1
00:00:06,281 --> 00:00:07,382
>> WELCOME BACK EVERYONE.

2
00:00:07,382 --> 00:00:08,450
HERE SUPER EXCITED TO HAVE YOU

3
00:00:08,450 --> 00:00:12,487
BACK HERE ON DAY 2 OF THE

4
00:00:12,487 --> 00:00:13,588
INSILIC O DRUG DISCOVERY

5
00:00:13,588 --> 00:00:13,855
WORKSHOP.

6
00:00:13,855 --> 00:00:15,123
THIS DAY WILL BE REALLY EXCITING

7
00:00:15,123 --> 00:00:16,458
AND WE HAVE A SUPER ECSIGHTING

8
00:00:16,458 --> 00:00:17,892
PROGRAM FOR YOU TODAY.

9
00:00:17,892 --> 00:00:21,963
WE WILL CONTINUE SESSION 3 ON

10
00:00:21,963 --> 00:00:23,898
APPLICATIONS OF INSILIC O

11
00:00:23,898 --> 00:00:24,966
METHODS ON DRUG DISCOVERY.

12
00:00:24,966 --> 00:00:28,536
WE WILL HAVE A SESSION 4 ON

13
00:00:28,536 --> 00:00:31,906
EMERGING TRENDS IN INSILIC O

14
00:00:31,906 --> 00:00:33,541
EMERGING TRENDS, AND IN THE

15
00:00:33,541 --> 00:00:38,913
2:00, WE WILL HAVE THE NEW NOBEL

16
00:00:38,913 --> 00:00:40,548
LAUREATE SPEAKER SPEAK ON DESIGN

17
00:00:40,548 --> 00:00:41,382
FUNCTIONS USING DEEP LEARNING

18
00:00:41,382 --> 00:00:43,651
AND THEN LATER IN THE AFTERNOON,

19
00:00:43,651 --> 00:00:47,055
WE WILL HAVE A SUPER EXCITING

20
00:00:47,055 --> 00:00:49,290
PANEL DISCUSSION ON BRIDGING THE

21
00:00:49,290 --> 00:00:50,925
GAPS IN TRANSLATION AND AS I

22
00:00:50,925 --> 00:00:51,693
MENTIONED YESTERDAY, IF WE

23
00:00:51,693 --> 00:00:52,794
DIDN'T COVER ANY OF THE

24
00:00:52,794 --> 00:00:55,230
QUESTIONS YOU HAD, AND YOU WANT

25
00:00:55,230 --> 00:00:56,898
TO BRING THEM BACK THIS IS A

26
00:00:56,898 --> 00:01:00,301
PERFECT PLACE DURING THE Q&A

27
00:01:00,301 --> 00:01:01,236
SESSION AT 420 TO BRING THOSE

28
00:01:01,236 --> 00:01:02,637
QUESTIONS BACK AND WE WILL

29
00:01:02,637 --> 00:01:06,841
TACKLE THEM .

30
00:01:06,841 --> 00:01:08,476
ALL RIGHT, LET'S CONTINUE TO

31
00:01:08,476 --> 00:01:09,077
SESSION 3 HERE.

32
00:01:09,077 --> 00:01:13,348
Y I WILL BRING BACK MY

33
00:01:13,348 --> 00:01:15,416
COLLEAGUE, DR. SHYAM RELE, WHO

34
00:01:15,416 --> 00:01:19,988
IS A PROGRAM OFFICER AT BARDA TO

35
00:01:19,988 --> 00:01:22,991
THE STAGE AND SHYAM, THE FLOOR

36
00:01:22,991 --> 00:01:24,526
IS YOURS.

37
00:01:24,526 --> 00:01:25,126
>> GOOD MORNING EVERYONE.

38
00:01:25,126 --> 00:01:28,830
WELCOME TO THE SECOND DAY OF

39
00:01:28,830 --> 00:01:32,433
THIS EXCITING WORKSHOP,

40
00:01:32,433 --> 00:01:34,736
CONTINUING FROM SEAN AND HIS

41
00:01:34,736 --> 00:01:38,206
EXCELLENT TALKS YESTERDAY, WE

42
00:01:38,206 --> 00:01:40,275
WILL CONTINUE WITH APPLICATIONS

43
00:01:40,275 --> 00:01:42,577
OF INSILIC O METHODS IN DRUG

44
00:01:42,577 --> 00:01:42,911
DISCOVERY.

45
00:01:42,911 --> 00:01:52,253
OUR FIRST SPEAKER TODAY IS

46
00:01:52,253 --> 00:01:56,157
MARISSA SAUNDERS, SHE IS AT

47
00:01:56,157 --> 00:01:57,191
REKURGZ PHARMACEUTICALS, SHE HAS

48
00:01:57,191 --> 00:01:59,561
ALREADY WORKED ON USING AI AND

49
00:01:59,561 --> 00:02:02,497
LEARNING TO ANSWER IMPACTFUL

50
00:02:02,497 --> 00:02:04,632
QUESTIONS AND SHE'S LEADING

51
00:02:04,632 --> 00:02:05,934
EFFORTS AT THE INTERFACE BETWEEN

52
00:02:05,934 --> 00:02:09,671
DEEP LEARNING AND MODELS TO

53
00:02:09,671 --> 00:02:10,572
ACCELERATE DISCOVERY AND

54
00:02:10,572 --> 00:02:12,740
DECISION MAKING TO ENABLE DRUG

55
00:02:12,740 --> 00:02:16,811
DISCOVERY AT SCALE USING

56
00:02:16,811 --> 00:02:17,845
AUTOMATED OMICS EXPERIMENTS, AI

57
00:02:17,845 --> 00:02:19,814
IS MACHINE LEARNING PREDICTIONS

58
00:02:19,814 --> 00:02:21,783
AND ANALYSIS AND STANDARDIZED

59
00:02:21,783 --> 00:02:25,687
WORK FLOWS AND DECISION ENGINES.

60
00:02:25,687 --> 00:02:27,188
[INDISCERNIBLE] IN THE LEARNING

61
00:02:27,188 --> 00:02:30,124
PLACE USING LANGUAGE MODELS TO

62
00:02:30,124 --> 00:02:34,195
DEVELOP TAXONOMY OF SKILLS AND

63
00:02:34,195 --> 00:02:37,932
AS A BIG DATA SCIENTIST

64
00:02:37,932 --> 00:02:40,969
CONSULTANT WORKING IN AREAS FROM

65
00:02:40,969 --> 00:02:42,070
BIOTECH TO VIDEO GAMES.

66
00:02:42,070 --> 00:02:43,605
SHE HAS A Ph.D. IN CHEMISTRY

67
00:02:43,605 --> 00:02:44,906
FROM THE UNIVERSITY OF CHICAGO

68
00:02:44,906 --> 00:02:46,841
AND COMPLETED POST DOC AT THE

69
00:02:46,841 --> 00:02:52,780
UNIVERSITY OF UTAH IN STRUCTURAL

70
00:02:52,780 --> 00:02:54,048
BIOLOGICAL [INDISCERNIBLE]

71
00:02:54,048 --> 00:02:57,919
ELECTRON MICROSCOPY IMAGES INTO

72
00:02:57,919 --> 00:02:58,353
PROTEIN STRUCTURES.

73
00:02:58,353 --> 00:03:00,688
MARISSA, BACK TO YOU, I THINK.

74
00:03:00,688 --> 00:03:02,256
>> THANK YOU SO MUCH.

75
00:03:02,256 --> 00:03:03,024
>> YOU HAVE 25 MINUTES FOR
THE

76
00:03:03,024 --> 00:03:05,226
TALK AND I WILL GIVE YOU A

77
00:03:05,226 --> 00:03:07,228
NOTICE AT AROUND 22 MINUTES IF

78
00:03:07,228 --> 00:03:07,996
THATIA OKAY.

79
00:03:07,996 --> 00:03:08,496
>> PERFECT.

80
00:03:08,496 --> 00:03:09,230
THAT SOUNDS GREAT.

81
00:03:09,230 --> 00:03:13,434
IS EVERYONE SEEING MY SLIDE

82
00:03:13,434 --> 00:03:15,069
OKAY.

83
00:03:15,069 --> 00:03:15,336
>> YES.

84
00:03:15,336 --> 00:03:15,803
>> WONDERFUL.

85
00:03:15,803 --> 00:03:16,971
>> WELL I'M SO HAPPY TO BE
HERE,

86
00:03:16,971 --> 00:03:18,906
THANK YOU SO MUCH FOR THE

87
00:03:18,906 --> 00:03:21,442
INVITATION TO TALK, I AM

88
00:03:21,442 --> 00:03:25,446
MARISSA, SANDERS AND I AM

89
00:03:25,446 --> 00:03:30,652
EXCITED TO SHARE TODAY'S

90
00:03:30,652 --> 00:03:36,824
TREALING EARLY DRUG DORPHY AT

91
00:03:36,824 --> 00:03:39,027
SCALE, TO ORIENT YOU IF YOU

92
00:03:39,027 --> 00:03:41,162
AREN'T FAMILIAR WITH REKURGZ, WE

93
00:03:41,162 --> 00:03:42,864
ARE A TECH BIOCOMPANY WITH OUR

94
00:03:42,864 --> 00:03:45,299
HEADQUARTERS IN SALT LAKE CITY,

95
00:03:45,299 --> 00:03:45,500
UTAH.

96
00:03:45,500 --> 00:03:49,270
WE ALSO HAVE OFFICES IN

97
00:03:49,270 --> 00:03:49,904
[INDISCERNIBLE], TORONTO,

98
00:03:49,904 --> 00:03:53,307
MONTREAL AND LONDON. 

99
00:03:53,307 --> 00:03:56,778
YOU MAY BE ASKING WHAT IS THIS

100
00:03:56,778 --> 00:03:58,680
BIOTECH THING, I THOUGHT THERE

101
00:03:58,680 --> 00:04:01,749
WAS A TERM OF BIOTECH, SO WHAT

102
00:04:01,749 --> 00:04:09,857
WE THINK AT RECURSION, THAT

103
00:04:09,857 --> 00:04:11,225
THERE ARE IMPORTANT TOOLS IN

104
00:04:11,225 --> 00:04:12,794
DRUG DISCOVERY.

105
00:04:12,794 --> 00:04:15,329
THIS IS WHERE WE STARTED TOO, WE

106
00:04:15,329 --> 00:04:17,532
STARTED APPROACHING DISCOVERY IN

107
00:04:17,532 --> 00:04:19,434
SCREENING USING MORPHOLOGICAL

108
00:04:19,434 --> 00:04:21,035
PROFILES AND SPECIFICALLY USING

109
00:04:21,035 --> 00:04:22,036
CELL PAINTING BUT AS WE

110
00:04:22,036 --> 00:04:24,539
CONTINUED ON THIS JOURNEY, WE

111
00:04:24,539 --> 00:04:26,240
REALIZED DISCOVERING AND

112
00:04:26,240 --> 00:04:27,141
DEVELOPING MEDICINE REQUIRES

113
00:04:27,141 --> 00:04:28,376
HUNDREDS OF STEPS AND IF YOU

114
00:04:28,376 --> 00:04:30,411
REALLY WANT TO ACCELERATE

115
00:04:30,411 --> 00:04:31,679
GETTING MEDICINES INTO THE HANDS

116
00:04:31,679 --> 00:04:33,815
OF PATIENTS, YOU HAVE TO

117
00:04:33,815 --> 00:04:34,615
INDUSTRIALIZE, DRUG DISCOVERY

118
00:04:34,615 --> 00:04:37,051
AND THAT MEANS YOU HAVE TO TAKE

119
00:04:37,051 --> 00:04:38,152
THESE POINT SOLUTIONS, AND

120
00:04:38,152 --> 00:04:40,021
INTEGRATE THEM AS MODULES ACROSS

121
00:04:40,021 --> 00:04:41,589
MANY DIVERSE STEPS, AND SO, THIS

122
00:04:41,589 --> 00:04:43,791
IS THE PIPELINE THAT WE USE TO

123
00:04:43,791 --> 00:04:46,194
THINK ABOUT HOW WE MOVE A DRUG

124
00:04:46,194 --> 00:04:47,695
FROM HYPOTHESIS, ALL THE WAY

125
00:04:47,695 --> 00:04:48,730
THROUGH THE CLINIC, AND YOU CAN

126
00:04:48,730 --> 00:04:51,032
SEE THAT THERE ARE MANY

127
00:04:51,032 --> 00:04:53,434
DIFFERENT MODULES, ALONG THE

128
00:04:53,434 --> 00:04:53,601
WAY.

129
00:04:53,601 --> 00:04:55,436
SO THIS IDEA OF USING POINT

130
00:04:55,436 --> 00:04:57,238
SOLUTIONS IN MANY PLACES ACROSS

131
00:04:57,238 --> 00:04:59,507
STEPS IS THE FIRST OF THE KEY

132
00:04:59,507 --> 00:05:02,143
IDEAS THAT I HOPE YOU WILL TAKE

133
00:05:02,143 --> 00:05:03,044
AWAY TODAY.

134
00:05:03,044 --> 00:05:04,345
THE OTHER 2 WHICH I WILL

135
00:05:04,345 --> 00:05:05,246
INTRODUCE HERE AND WILL TALK

136
00:05:05,246 --> 00:05:06,547
ABOUT THROUGH THE COURSE OF THIS

137
00:05:06,547 --> 00:05:08,783
TALK ARE THAT THE MOST EFFICIENT

138
00:05:08,783 --> 00:05:10,518
EXPERIMENT, 1 CAN RUN IS NO

139
00:05:10,518 --> 00:05:11,586
EXPERIMENT AT ALL.

140
00:05:11,586 --> 00:05:13,588
AND SO THAT IS THE REAL PROMISE

141
00:05:13,588 --> 00:05:15,089
OF INSILIC O METHODS.

142
00:05:15,089 --> 00:05:17,091
THEY HELP US TO PREDICT WHICH

143
00:05:17,091 --> 00:05:17,859
EXPERIMENTS ARE THE RIGHT

144
00:05:17,859 --> 00:05:19,060
EXPERIMENTS TO RUN AND THAT

145
00:05:19,060 --> 00:05:20,394
MEANS WE CAN BE MORE EFFICIENT

146
00:05:20,394 --> 00:05:23,297
IN BOTH TIME AND MONEY IN

147
00:05:23,297 --> 00:05:24,532
GETTING THROUGH DRUG DISCOVERY.

148
00:05:24,532 --> 00:05:26,467
TO BE ABLE TO GET BETTER AND

149
00:05:26,467 --> 00:05:28,236
BETTER AT KNOWING WHICH

150
00:05:28,236 --> 00:05:30,505
EXPERIMENTS WE SHOULD RUN, WE

151
00:05:30,505 --> 00:05:31,873
NEED BETTER MODELS AND FOR THAT

152
00:05:31,873 --> 00:05:38,212
WE NEED BETTER DATA SO A KEY

153
00:05:38,212 --> 00:05:39,614
IDEA AT RUECURSION, SO THAT IN

154
00:05:39,614 --> 00:05:41,749
ORDER TO HAVE THESE CYCLES OF

155
00:05:41,749 --> 00:05:42,784
LEARNING AND IMPROVING MODELS WE

156
00:05:42,784 --> 00:05:44,118
HAVE TO COUPLE THE PHYSICAL

157
00:05:44,118 --> 00:05:45,853
EXPERIMENTS WE USE TO PROGRESS

158
00:05:45,853 --> 00:05:47,021
PROGRAMS THROUGH WITH INSILIC O

159
00:05:47,021 --> 00:05:49,390
METHODS SO THAT WE CAN

160
00:05:49,390 --> 00:05:52,093
ITERATIVELY IMPROVE OUR MODELS.

161
00:05:52,093 --> 00:05:53,694
WE START BY THINKING ABOUT WHAT

162
00:05:53,694 --> 00:05:56,564
ARE THE DATA MODULES, WHERE ARE

163
00:05:56,564 --> 00:05:58,833
THE PLACES THAT WE NEED TO FOCUS

164
00:05:58,833 --> 00:06:00,902
AND YOU CAN SEE IN THIS

165
00:06:00,902 --> 00:06:02,703
PIPELINE, WE HAVE A RANGE OF

166
00:06:02,703 --> 00:06:04,505
DIFFERENT DATA POINTS WHERE

167
00:06:04,505 --> 00:06:06,174
WE'RE COLLECTING DATA, RANGING

168
00:06:06,174 --> 00:06:09,010
FROM PATIENT DATA THAT HELP US

169
00:06:09,010 --> 00:06:13,047
BUILD CAUSAL ML MODELS TO TARGET

170
00:06:13,047 --> 00:06:15,316
PREDICTIONS TO OUR SCALED OMICS

171
00:06:15,316 --> 00:06:19,086
CAPABILITIES AND ALL OF THESE

172
00:06:19,086 --> 00:06:20,087
DIFFERENT DATA MODALITIES THEN

173
00:06:20,087 --> 00:06:25,459
HELP US TO NAVIGATE THROUGH DRUG

174
00:06:25,459 --> 00:06:26,093
DISCOVERY.

175
00:06:26,093 --> 00:06:28,629
WE USE ML MODELS AND AI MODELS

176
00:06:28,629 --> 00:06:30,264
THROUGHOUT THIS PROCESS TO HELP

177
00:06:30,264 --> 00:06:31,032
INFORM WHICH EXPERIMENTS WE

178
00:06:31,032 --> 00:06:31,332
SHOULD RUN.

179
00:06:31,332 --> 00:06:33,835
SO IN THE PATIENT CONNECTIVITY

180
00:06:33,835 --> 00:06:35,703
AND NOVELTY STAGE AS AN EXAMPLE,

181
00:06:35,703 --> 00:06:37,538
WE USE LLMs TO FILTER THROUGH

182
00:06:37,538 --> 00:06:38,840
THE CORPUS OF DATA IN THE WORLD

183
00:06:38,840 --> 00:06:40,474
AND HELP US KNOW WHERE WE SHOULD

184
00:06:40,474 --> 00:06:44,145
BE INTERESTED IN LOOKING FOR NEW

185
00:06:44,145 --> 00:06:44,645
HYPOTHESIS.

186
00:06:44,645 --> 00:06:46,447
WHEN WE HAVE PATIENT DATA, WE

187
00:06:46,447 --> 00:06:48,950
USE CAUSE AT GENETIC MODELS TO

188
00:06:48,950 --> 00:06:50,985
HELP INFORM WHAT OUR UNKNOWN

189
00:06:50,985 --> 00:06:54,055
TARGETS OR CAUSAL GENES FOR DRUG

190
00:06:54,055 --> 00:06:54,322
DISCOVERY.

191
00:06:54,322 --> 00:06:56,357
WHEN WE MOVE TO HIDDEN TARGET

192
00:06:56,357 --> 00:06:58,226
VALIDATION, WE USE GENERATIVE

193
00:06:58,226 --> 00:07:00,962
CHEMISTRY MODELS WHICH YOU HEARD

194
00:07:00,962 --> 00:07:02,830
ABOUT EARLIER YESTERDAY IN ORDER

195
00:07:02,830 --> 00:07:04,832
TO GENERATE THE STRUCTURES THAT

196
00:07:04,832 --> 00:07:07,268
WE NEED TO BE FEEDING INTO THIS

197
00:07:07,268 --> 00:07:08,903
SYSTEM AND WE USE FOUNDATION

198
00:07:08,903 --> 00:07:10,404
MODELS FOR IMAGING WHICH I WILL

199
00:07:10,404 --> 00:07:12,273
TALK ABOUT IN JUST A MINUTE IN

200
00:07:12,273 --> 00:07:15,610
ORDER TO INTERPRET OUR SCALED

201
00:07:15,610 --> 00:07:15,943
GENOMICS DATA.

202
00:07:15,943 --> 00:07:17,912
WHEN WE GO INTO COMPOUND

203
00:07:17,912 --> 00:07:19,647
OPTIMIZATION, WE ARE USING

204
00:07:19,647 --> 00:07:21,115
PROTEIN LIGAND BINDING MODELS TO

205
00:07:21,115 --> 00:07:23,484
FIND OUT WHAT IS THE ACTUAL

206
00:07:23,484 --> 00:07:24,785
PROTEIN TARGET THAT COMPOWBDS

207
00:07:24,785 --> 00:07:26,420
OUR BINDING TO AND WE USE

208
00:07:26,420 --> 00:07:28,456
CHEMICAL PROPERTY PREDICTION IN

209
00:07:28,456 --> 00:07:29,824
ORDER TO KNOW WHICH THINGS ARE

210
00:07:29,824 --> 00:07:31,025
ACTUALLY GOING TO MOVE US TO A

211
00:07:31,025 --> 00:07:36,264
GOOD SPACE IN TERMS OF ADMIT

212
00:07:36,264 --> 00:07:36,564
PROPERTIES.

213
00:07:36,564 --> 00:07:38,799
AND THEN FINALLY IN TRANSLATION,

214
00:07:38,799 --> 00:07:40,001
WE USE THINGS LIKE VIDEO

215
00:07:40,001 --> 00:07:41,202
ANALYSIS AND ACTIVE LEARNING

216
00:07:41,202 --> 00:07:43,371
MODELS IN ORDER TO TELL US WHICH

217
00:07:43,371 --> 00:07:45,406
COMPOUNDS ARE STARTING TO SHOW

218
00:07:45,406 --> 00:07:48,743
EFFECTS OF TOXICITY IN OUR MICE,

219
00:07:48,743 --> 00:07:50,211
AND TO BETTER TRUNCATE THOSE

220
00:07:50,211 --> 00:07:54,215
EXPERIMENTS IF WE NEED TO.

221
00:07:54,215 --> 00:07:56,150
AND SO, IN ORDER TO BUILD ALL OF

222
00:07:56,150 --> 00:07:58,185
THESE MODELS THAT HELP US TO

223
00:07:58,185 --> 00:08:00,254
FILTER, WE REALLY DO NEED LARGE

224
00:08:00,254 --> 00:08:02,590
AMOUNTS OF DATA AND TO GET

225
00:08:02,590 --> 00:08:03,991
THERE, WE LEAN INTO THIS FLY

226
00:08:03,991 --> 00:08:08,896
WHEEL IDEA, WHERE WE HAVE REAL

227
00:08:08,896 --> 00:08:09,864
PHYSICAL EXPERIMENTS, HAPPENING

228
00:08:09,864 --> 00:08:12,600
IN HIGH DIMENSIONAL ASSAYS, AT

229
00:08:12,600 --> 00:08:15,403
SCALE AND THIS DATA THEN FEEDS

230
00:08:15,403 --> 00:08:16,771
INTO DIGITAL REPRESENTATIONS OF

231
00:08:16,771 --> 00:08:19,173
REALITY, SO WE GENERATE THIS

232
00:08:19,173 --> 00:08:21,776
CORPUS OF DATA AND THEN THAT

233
00:08:21,776 --> 00:08:25,513
DATA IN TURN AND FED INTO OUR

234
00:08:25,513 --> 00:08:26,847
ALGORITHMS AND METHODS INSILIC O

235
00:08:26,847 --> 00:08:30,718
IN ORDERED TO HELP US CREATE

236
00:08:30,718 --> 00:08:33,187
THESE IN SILIC O MODELS WHICH IN

237
00:08:33,187 --> 00:08:34,488
TURN INFORM WHICH EXPERIMENTS WE

238
00:08:34,488 --> 00:08:35,990
SHOULD RUN NEXT.

239
00:08:35,990 --> 00:08:38,559
TO DO THIS, WE ARE REALLY USING

240
00:08:38,559 --> 00:08:39,660
THIS DRUG DISCOVERY PROCESS TO

241
00:08:39,660 --> 00:08:41,595
GENERATE A DATA SET THAT IS FIT

242
00:08:41,595 --> 00:08:52,106
FOR PURPOSE FOR NEW ML MODELS.

243
00:09:07,188 --> 00:09:08,856
-- IN THESE HIGH DIMENSIONAL

244
00:09:08,856 --> 00:09:10,358
INTERACTIONS AND SO AS A RESULT

245
00:09:10,358 --> 00:09:12,393
THE KINDS OF DATA WE COLLECT AT

246
00:09:12,393 --> 00:09:17,131
SCALE ARE OMICS DATA, GENOMICS,

247
00:09:17,131 --> 00:09:19,800
TRANSCRIPT OMICS, PROTEOMICS AND

248
00:09:19,800 --> 00:09:21,702
IN ADDITION TO STANDARD ADME

249
00:09:21,702 --> 00:09:23,371
ASSAYS AND THEN SOME MOUSE

250
00:09:23,371 --> 00:09:24,605
MODELS AND REAL WORLD PATIENT

251
00:09:24,605 --> 00:09:24,939
DATA.

252
00:09:24,939 --> 00:09:26,707
AND OVER THE YEARS, WE'VE

253
00:09:26,707 --> 00:09:29,043
MANAGED TO COLLECT A SIGNIFICANT

254
00:09:29,043 --> 00:09:33,247
CORPUS OF DATA ACROSS ALL OF

255
00:09:33,247 --> 00:09:34,648
THESE LAYERS, EITHER THROUGH

256
00:09:34,648 --> 00:09:35,583
PARTNERSHIPS OR THROUGH

257
00:09:35,583 --> 00:09:37,885
GENERATING THAT DATA IN-HOUSE ON

258
00:09:37,885 --> 00:09:48,429
OUR PLATFORM AND YOU CAN SEE OUR

259
00:09:50,097 --> 00:09:51,499
PH ENOMICS AND WHERE YOU CAN

260
00:09:51,499 --> 00:09:53,434
WHERE THERE'S REAL WORLD DATA,

261
00:09:53,434 --> 00:09:54,835
GENOMICKINGS DATA WHERE WE

262
00:09:54,835 --> 00:09:55,536
PARTNER WITH PEOPLE TO HAVE

263
00:09:55,536 --> 00:09:56,804
ACCESS TO THIS DAT AND THEN

264
00:09:56,804 --> 00:09:58,105
THERE ARE ALSO LAYERS WHERE WE

265
00:09:58,105 --> 00:10:00,875
ARE JUST NOW STARTING TO ENTER

266
00:10:00,875 --> 00:10:03,077
THIS FLY WHEEL, PROTEOMICS WE'RE

267
00:10:03,077 --> 00:10:05,613
JUST STARTING THIS, WE STARTED

268
00:10:05,613 --> 00:10:08,315
EARLIER THIS YEAR AND OUR ADME

269
00:10:08,315 --> 00:10:10,484
PLATFORM WE STARTED RUNNING AT

270
00:10:10,484 --> 00:10:13,354
SCALE LATE LAST YEAR EMPLOY WHAT

271
00:10:13,354 --> 00:10:15,790
ALL THIS DATA GIVES US IS THE

272
00:10:15,790 --> 00:10:17,925
ABILITY TO BUILD A RANGE OF

273
00:10:17,925 --> 00:10:19,760
DIFFERENT MODELS AND YOU CAN SEE

274
00:10:19,760 --> 00:10:21,996
HERE, WE BUILD EVERYTHING FROM

275
00:10:21,996 --> 00:10:23,831
CAUSAL GENETIC MODELS THROUGH TO

276
00:10:23,831 --> 00:10:26,834
CHEMICAL PROPERTY PREDICTION

277
00:10:26,834 --> 00:10:30,404
MODELS, AND EVEN VIDEO TIME

278
00:10:30,404 --> 00:10:37,812
LAPSE BEHAVIORIAL MODELS FOR OUR

279
00:10:37,812 --> 00:10:38,479
IN VIVO OMICS.

280
00:10:38,479 --> 00:10:40,047
THIS IS OUR FOUNDATIONAL LAYER

281
00:10:40,047 --> 00:10:41,849
AND IT IS THE PLACE WE OFTEN

282
00:10:41,849 --> 00:10:43,717
START IN THINKING ABOUT

283
00:10:43,717 --> 00:10:44,151
PROGRAMS.

284
00:10:44,151 --> 00:10:52,359
THE IDEA BEHIND PH ENOMICS IS A

285
00:10:52,359 --> 00:10:53,627
MORE WHOLESTIC WAY TO CAPTURE

286
00:10:53,627 --> 00:10:53,994
THE RESPONSE.

287
00:10:53,994 --> 00:10:57,498
SO IF YOU THINK ABOUT IT BY

288
00:10:57,498 --> 00:10:58,799
DISEASE PRETURB ANT THAT

289
00:10:58,799 --> 00:10:59,934
PROTURBATION IS GOING TO CHANGE

290
00:10:59,934 --> 00:11:01,969
THE GENE EXPRESSION AND SO

291
00:11:01,969 --> 00:11:04,738
CHANGE THE TRANSCRIPTION

292
00:11:04,738 --> 00:11:05,906
ACTIVITIES IN THE CELL.

293
00:11:05,906 --> 00:11:07,475
IT WILL ALSO SHIFT PROTEIN

294
00:11:07,475 --> 00:11:10,311
LEVELSAs A RESULT OF THAT

295
00:11:10,311 --> 00:11:11,445
CHANGE IN TRANSCRIPTION, THAT IN

296
00:11:11,445 --> 00:11:12,947
TURN WILL SHIFT PROTEIN ACTIVITY

297
00:11:12,947 --> 00:11:14,448
AND ALL OF THOSE THINGS ARE

298
00:11:14,448 --> 00:11:17,284
THINGS WE CAN MEASURE THROUGH

299
00:11:17,284 --> 00:11:18,719
DIFFERENT MEASURE OMICS AND

300
00:11:18,719 --> 00:11:21,589
PROTEOMICS AND METRICS TAB O

301
00:11:21,589 --> 00:11:23,224
LOAMICS, BUT THE FANS, ALL OF

302
00:11:23,224 --> 00:11:24,658
THESE, THE MORPHOLOGY OF THIS

303
00:11:24,658 --> 00:11:25,960
END POINT OF THIS CELL IS

304
00:11:25,960 --> 00:11:27,695
INFORMED BY ALL OF THESE CHANGES

305
00:11:27,695 --> 00:11:30,264
RESULTING IN A CELLULAR END

306
00:11:30,264 --> 00:11:33,534
STATE AND WE INFACT SEE IF WE

307
00:11:33,534 --> 00:11:34,435
APPLY DIFFERENT PROTURBATIONS TO

308
00:11:34,435 --> 00:11:37,505
CELLS THAT WE CAN VISUALLY

309
00:11:37,505 --> 00:11:41,575
DISTINGUISH BETWEEN SOME VERY

310
00:11:41,575 --> 00:11:43,377
MORPHOLOGIES AND SO THIS IS THE

311
00:11:43,377 --> 00:11:53,888
FOUNDATION FOR ATION FOR WHICH

312
00:12:13,541 --> 00:12:14,375
WE'RE BUILDING.

313
00:12:14,375 --> 00:12:17,945
WHAT WAS EXCITING FOR THIS IS WE

314
00:12:17,945 --> 00:12:19,847
COULD TEST FOR PROTURBATION, WE

315
00:12:19,847 --> 00:12:20,781
COULD PHYSICALLY STRESS THESE

316
00:12:20,781 --> 00:12:22,550
CELLS OUT AND ALL OF THOSE

317
00:12:22,550 --> 00:12:26,654
THINGS COULD BE CAPTURED ON OUR

318
00:12:26,654 --> 00:12:27,521
PLATFORM.

319
00:12:27,521 --> 00:12:29,123
BUT IN THIS MODALITY WE WERE

320
00:12:29,123 --> 00:12:30,925
STILL IN THIS PLACE WHERE FOR

321
00:12:30,925 --> 00:12:31,725
EVERY DIFFERENT DISEASE WE

322
00:12:31,725 --> 00:12:33,160
WANTED TO TEST, WE HAD TO SCREEN

323
00:12:33,160 --> 00:12:34,795
ALL OF OUR COMPOUNDS AND SO WHAT

324
00:12:34,795 --> 00:12:36,697
THAT MEANT IS IF WE WANT TO

325
00:12:36,697 --> 00:12:39,033
THINK BROADLY, IF WE WANT TO

326
00:12:39,033 --> 00:12:40,968
SCREEN ACROSS A WIDE BREDTH OF

327
00:12:40,968 --> 00:12:43,103
DISEASES WE HAVE THIS QUADRATIC

328
00:12:43,103 --> 00:12:44,738
SCALING WHERE WE HAVE TO SCREEN

329
00:12:44,738 --> 00:12:46,040
THE ENTIRE LIBRARY FOR EACH 1 OF

330
00:12:46,040 --> 00:12:49,310
THOSE AND THAT WAS NOT SOMETHING

331
00:12:49,310 --> 00:12:52,680
THAT WOULD SCALE THE WAY WE HAD

332
00:12:52,680 --> 00:12:53,113
AMBITIONS TO SCALE.

333
00:12:53,113 --> 00:12:55,049
AND SO THE WAY THAT WE SOLVED

334
00:12:55,049 --> 00:12:57,851
THIS PROBLEM WAS TO THINK ABOUT

335
00:12:57,851 --> 00:12:59,220
USING AN INFERENTIAL MAP.

336
00:12:59,220 --> 00:13:00,788
SO THE IDEA WAS INSTEAD OF THE

337
00:13:00,788 --> 00:13:02,456
PARADIGM WHERE YOU HAVE THE

338
00:13:02,456 --> 00:13:03,490
DISEASE, AND ON TOP OF THAT

339
00:13:03,490 --> 00:13:05,593
DISEASE, YOU PUT A COMPOUND, THE

340
00:13:05,593 --> 00:13:08,028
QUESTION WAS, CAN WE TAKE A

341
00:13:08,028 --> 00:13:09,863
DISEASE AND THEN SEPARATELY TAKE

342
00:13:09,863 --> 00:13:11,932
A COMPOUND AND TREAT CELLS WITH

343
00:13:11,932 --> 00:13:14,735
THOSE AND IN SOME WAY FROM THE

344
00:13:14,735 --> 00:13:16,036
IMAGES CAPTURE BIOLOGICAL

345
00:13:16,036 --> 00:13:19,974
RELATIONSHIPS IN SUCH A WAY THAT

346
00:13:19,974 --> 00:13:20,774
THOSE PRESENTATIONS LET US INFER

347
00:13:20,774 --> 00:13:21,942
SOMETHING ABOUT WHAT WILL HAPPEN

348
00:13:21,942 --> 00:13:23,711
IF THOSE THINGS ARE PUT

349
00:13:23,711 --> 00:13:24,144
TOGETHER.

350
00:13:24,144 --> 00:13:27,081
AND SO WHAT WE'VE COME UP WITH

351
00:13:27,081 --> 00:13:28,182
IS THESE PHENOMAPS, YOU WILL SEE

352
00:13:28,182 --> 00:13:31,619
THIS PICTURE IN A FEW SLIDES

353
00:13:31,619 --> 00:13:33,254
HERE, THE RED SQUARES ARE WHERE

354
00:13:33,254 --> 00:13:35,422
THOSE 2 THINGS ARE VERY SIMILAR

355
00:13:35,422 --> 00:13:38,425
TO 1 ANOTHER, SO THE IMAGES LOOK

356
00:13:38,425 --> 00:13:40,461
SIMILAR BIOLOGICALLY AND THE

357
00:13:40,461 --> 00:13:41,862
BLUE SQUARES IS WHERE THEY LOOK

358
00:13:41,862 --> 00:13:44,531
OPPOSITE PRACTICES 1 ANOTHER IN

359
00:13:44,531 --> 00:13:45,266
THIS BIOLOGICAL REPRESENTATIONAL

360
00:13:45,266 --> 00:13:47,067
SPACE AND YOU MIGHT SAY, THE KEY

361
00:13:47,067 --> 00:13:49,169
THING HERE, IS THEN TO FIGURE

362
00:13:49,169 --> 00:13:51,238
OUT HOW TO TRANSFORM AN IMAGE

363
00:13:51,238 --> 00:13:53,007
INTO A REPRESENTATION THAT

364
00:13:53,007 --> 00:13:53,807
CARRIES THIS BIOLOGICAL MEANING

365
00:13:53,807 --> 00:13:56,677
AND HOW DO WE DO THIS.

366
00:13:56,677 --> 00:13:58,679
THE WAY WE'RE DOING THIS, IN

367
00:13:58,679 --> 00:14:00,714
FACT, IS THROUGH DEEP LEARNING

368
00:14:00,714 --> 00:14:03,083
MODELS AND MOST RECENTLY IT'S

369
00:14:03,083 --> 00:14:04,118
THROUGH FOUNDATION MODELS AND

370
00:14:04,118 --> 00:14:05,619
THESE MODELS ENABLE US TO TAKE

371
00:14:05,619 --> 00:14:07,655
ADVANTAGE OF THIS FLY WHEEL OF

372
00:14:07,655 --> 00:14:11,225
DATA GENERATION AS I'LL EXPLAIN

373
00:14:11,225 --> 00:14:11,659
NOW.

374
00:14:11,659 --> 00:14:13,927
SO WE HAVE GENERATED TO DATE

375
00:14:13,927 --> 00:14:15,996
OVER 250 MILLION DIFFERENT

376
00:14:15,996 --> 00:14:17,531
EXPERIMENTS RANGING ACROSS 50

377
00:14:17,531 --> 00:14:19,833
HUMAN CELL TYPES, INCLUDING A

378
00:14:19,833 --> 00:14:21,368
TRILLION DIFFERENT HUMAN NEURONS

379
00:14:21,368 --> 00:14:22,303
THAT HAVE BEEN GENERATED AND

380
00:14:22,303 --> 00:14:24,405
WHAT WE CAN DO WITH ALL OF THIS

381
00:14:24,405 --> 00:14:27,274
DATA IS WE CAN FEED IT INTO A

382
00:14:27,274 --> 00:14:29,343
FOUNDATION MODEL WHICH IS A

383
00:14:29,343 --> 00:14:30,411
MODEL WHERE WE'RE NOT TELLING

384
00:14:30,411 --> 00:14:31,779
THE MACHINE WHAT THE ANSWER IS

385
00:14:31,779 --> 00:14:34,782
THAT WE WANT IN TERMS OF A CLASS

386
00:14:34,782 --> 00:14:36,116
BUT WE'RE RATHER ASKING IT DO A

387
00:14:36,116 --> 00:14:37,885
TASK AND THE TASK WE'RE ASKING

388
00:14:37,885 --> 00:14:39,987
IT TO DO IS TO RECONSTRUCT A

389
00:14:39,987 --> 00:14:41,522
MASK IMAGE SO WE CAN TAKE THIS

390
00:14:41,522 --> 00:14:44,825
IMAGE WHICH IS OUR INPUT IMAGE

391
00:14:44,825 --> 00:14:47,861
ON THE RIGHT, AND WE CAN MASK

392
00:14:47,861 --> 00:14:50,464
OUT A LARGE CHUNK OF IT, SAY 75%

393
00:14:50,464 --> 00:14:52,232
OF THE IMAGE GETS MASKED OUT AND

394
00:14:52,232 --> 00:14:55,002
THEN WE ASK OUR ML ALGORITHM TO

395
00:14:55,002 --> 00:14:56,503
RECONSTRUCT THAT IMAGE FROM THIS

396
00:14:56,503 --> 00:14:57,538
MASKED INPUT AND AS YOU CAN SEE

397
00:14:57,538 --> 00:15:00,708
IT DOES A VERY, VERY GOOD JOB OF

398
00:15:00,708 --> 00:15:02,843
DOING THAT, AND WHAT'S MORE

399
00:15:02,843 --> 00:15:05,346
IMPORTANT IS THIS TASK CAPTURES

400
00:15:05,346 --> 00:15:06,947
INFORMATION ABOUT THE BIOLOGY

401
00:15:06,947 --> 00:15:08,215
THAT'S GOING ON ABOUT THE

402
00:15:08,215 --> 00:15:10,050
MORPHOLOGY OF THESE CELLS, THAT

403
00:15:10,050 --> 00:15:13,087
THEN WE CAN USE THOSE EMBEDDINGS

404
00:15:13,087 --> 00:15:15,956
TO REPRESENT OUR PROTURBATIONS

405
00:15:15,956 --> 00:15:16,790
IN THIS INFERENTIAL MAP.

406
00:15:16,790 --> 00:15:18,959
ONE OF THE THINGS THAT'S REALLY

407
00:15:18,959 --> 00:15:19,560
EXCITING ABOUT THIS MODEL IS

408
00:15:19,560 --> 00:15:22,196
THAT WE ARE SEEING THAT IT

409
00:15:22,196 --> 00:15:23,797
SCALES CONTINUES TO SCALE WITH

410
00:15:23,797 --> 00:15:25,265
THE MEMBER OF PARAMETERS SO WE

411
00:15:25,265 --> 00:15:27,034
CAN BUILD LARGER AND LARGER

412
00:15:27,034 --> 00:15:28,202
NETWORK MODELS AS WE GET MORE

413
00:15:28,202 --> 00:15:30,003
DATA AND AS WE HAVE ACCESS TO

414
00:15:30,003 --> 00:15:31,405
MORE COMPUTE AND THAT MEANS

415
00:15:31,405 --> 00:15:32,940
THESE MODELS WILL GET BETTER AND

416
00:15:32,940 --> 00:15:35,342
BETTER AND THAT'S REALLY

417
00:15:35,342 --> 00:15:35,642
DESIRABLE.

418
00:15:35,642 --> 00:15:38,746
WE'VE ALSO SEEN THAT THESE

419
00:15:38,746 --> 00:15:39,913
MODELS ARE GENERALIZABLE TO NEW

420
00:15:39,913 --> 00:15:42,750
PROBLEMS, AND SO IF WE COMPARE

421
00:15:42,750 --> 00:15:45,185
THIS TO OTHER WAYS OF EMBEDDING

422
00:15:45,185 --> 00:15:47,855
IMAGE DATA WE SEE THAT IT DOES

423
00:15:47,855 --> 00:15:48,789
SIGNIFICANTLY BETTER AT RECALL

424
00:15:48,789 --> 00:15:51,058
AND ACTUALLY IS ABLE TO CAPTURE

425
00:15:51,058 --> 00:15:52,559
THINGS BEYOND JUST THE CONTEXT

426
00:15:52,559 --> 00:15:54,361
WHERE IT WAS TRAINED, WHICH IS

427
00:15:54,361 --> 00:15:57,798
THE GREAT PROMISE OF FOUNDATION

428
00:15:57,798 --> 00:15:58,499
MODELS.

429
00:15:58,499 --> 00:16:02,202
PHENOM DATA WHICH IS AVAILABLE

430
00:16:02,202 --> 00:16:03,237
THROUGH VIDEOS BY

431
00:16:03,237 --> 00:16:03,871
[INDISCERNIBLE] FOR ANYONE TO

432
00:16:03,871 --> 00:16:06,173
USE IS 1 OF THOSE MODELS AND WE

433
00:16:06,173 --> 00:16:08,008
CONTINUE TO ITERATE ON THIS

434
00:16:08,008 --> 00:16:10,110
INTERNALLY AS WE GENERATE MORE

435
00:16:10,110 --> 00:16:12,446
AND MORE DATA IN THIS FLY WHEEL

436
00:16:12,446 --> 00:16:14,148
OF IMPROVEMENT AND WHEN WE DO

437
00:16:14,148 --> 00:16:15,749
THIS, WHEN WE PUT TOGETHER THIS

438
00:16:15,749 --> 00:16:16,817
MAP, WE CAN LOOK THEA THE MAP

439
00:16:16,817 --> 00:16:18,352
AND WE CAN SAY, FOR ANY

440
00:16:18,352 --> 00:16:19,820
COMBINATION THAT WE WANT TO

441
00:16:19,820 --> 00:16:22,156
THINK OF, WHAT IS THE SIGNAL

442
00:16:22,156 --> 00:16:24,224
THAT WE'RE SEEING IN THE MAP.

443
00:16:24,224 --> 00:16:26,593
ARE THESE SIMILAR, DO THEY HAVE

444
00:16:26,593 --> 00:16:27,594
NO RELATIONSHIPS OR ARE THEY

445
00:16:27,594 --> 00:16:28,896
OPPOSITE TO EACH OTHER AND WE

446
00:16:28,896 --> 00:16:31,999
CAN DO THIS NOT ONLY FOR

447
00:16:31,999 --> 00:16:33,233
GENE-GENE RELATIONSHIPPED WHERE

448
00:16:33,233 --> 00:16:34,701
WE KNOW THESE SIGNALING PATHWAYS

449
00:16:34,701 --> 00:16:38,806
AND WE SEE THAT THE MAP

450
00:16:38,806 --> 00:16:39,506
RECAPITULATES THESE SIGNALING

451
00:16:39,506 --> 00:16:41,975
PATHWAYS AND WE CAN DO THIS IN

452
00:16:41,975 --> 00:16:43,210
COMPARING GENES TO COMPOUNDS AND

453
00:16:43,210 --> 00:16:45,345
WE CAN SEE THAT THE MAPS ARE

454
00:16:45,345 --> 00:16:46,079
ALSO RECAPITULATING THE BIOLOGY

455
00:16:46,079 --> 00:16:48,148
WE KNOW FROM LITERATURE AND FROM

456
00:16:48,148 --> 00:16:49,817
PRIOR EXPERIMENTS WE SHOULD SEE.

457
00:16:49,817 --> 00:16:52,119
AND THEN WE CAN AM CO BIEN THOSE

458
00:16:52,119 --> 00:16:54,755
INTO 1 SO THAT WE SEE BOTH THE

459
00:16:54,755 --> 00:16:56,256
COMPOUNDS AND THE RELATED GENES

460
00:16:56,256 --> 00:16:59,393
AND GET THE FULL PICTURE OF THE

461
00:16:59,393 --> 00:17:07,000
KNOWN AND UNKNOWN BIOLOGY AROUND

462
00:17:07,000 --> 00:17:08,836
A PARTICULAR AREA.

463
00:17:08,836 --> 00:17:11,405
ONCE WE'VE GOT THE NOMICs

464
00:17:11,405 --> 00:17:13,740
DATA, WE ALSO WANT A METHOD TO

465
00:17:13,740 --> 00:17:16,243
VALENTINED DIADIC TWHA OR HAVE

466
00:17:16,243 --> 00:17:17,211
AN ORTHOGONAL ASSAY TO SHOW US

467
00:17:17,211 --> 00:17:18,912
WHAT WE'RE SEEING IS NOT JUST AN

468
00:17:18,912 --> 00:17:20,948
ARTIFACT OF HOW WE'RE DOING FOR

469
00:17:20,948 --> 00:17:22,082
GENOMICS BUT IT IS REAL BIOLOGY

470
00:17:22,082 --> 00:17:24,518
AND THE WAY WE'RE DOING THAT IS

471
00:17:24,518 --> 00:17:25,486
THROUGH TRANSCRIPTOMICS, THIS IS

472
00:17:25,486 --> 00:17:29,556
OUR SECOND DATA LAYER, WE HAVE

473
00:17:29,556 --> 00:17:30,958
GENERATED AT SCALE, THE FIRST

474
00:17:30,958 --> 00:17:37,998
IME NATIONAL LIBRARY  --

475
00:17:37,998 --> 00:17:41,401
GENOME SCALE THROUGH MAP TO

476
00:17:41,401 --> 00:17:43,737
PERTURB SEQ, IT GIVES US A WAY

477
00:17:43,737 --> 00:17:45,572
TO NOT ONLY TIE THIS BACK TO

478
00:17:45,572 --> 00:17:48,075
PATIENTS BUT IT HAS GIVEN US

479
00:17:48,075 --> 00:17:51,144
THIS MULTIMODAL WORK FLOW IN

480
00:17:51,144 --> 00:17:52,079
TRANSCRIPTOMICS AND PREDICT

481
00:17:52,079 --> 00:17:53,213
WHICH COMPOUNDS WILL FAIL DOWN

482
00:17:53,213 --> 00:17:54,948
STREAM TESTS AND WHICH COMPOUNDS

483
00:17:54,948 --> 00:17:56,250
WILL PASS DOWN STREAM TESTS AND

484
00:17:56,250 --> 00:17:57,918
THAT LETS US RUN THE RIGHT

485
00:17:57,918 --> 00:18:02,089
EXPERIMENTS AT THE RIGHT TIMES.

486
00:18:02,089 --> 00:18:03,490
OTHER MODELS WE'RE USING RIGHT

487
00:18:03,490 --> 00:18:06,994
NOW, WE'RE USING PATIENT DATA TO

488
00:18:06,994 --> 00:18:08,829
CREATE CAUSAL MODELS, SO IN THIS

489
00:18:08,829 --> 00:18:10,531
MODEL SETTING, THE DATA THAT

490
00:18:10,531 --> 00:18:13,500
WE'RE USING AS PARTNERSHIP DATA

491
00:18:13,500 --> 00:18:14,768
FROM [INDISCERNIBLE] WHICH HAVE

492
00:18:14,768 --> 00:18:17,437
PATIENT RECORDS THAT ARE

493
00:18:17,437 --> 00:18:18,405
DEIDENTIFIED AND STANDARDIZED

494
00:18:18,405 --> 00:18:21,241
AND BY COMBINING THIS DATA,

495
00:18:21,241 --> 00:18:22,309
WHICH REALLY DOES FORWARD

496
00:18:22,309 --> 00:18:23,877
GENETICS WHERE WE KNOW THE

497
00:18:23,877 --> 00:18:26,280
PHENOTYPE OF THE PATIENT AND WE

498
00:18:26,280 --> 00:18:28,115
CAN MAP FORWARD WHAT ARE THE

499
00:18:28,115 --> 00:18:30,217
GENETIC DRIVERS THAT ARE

500
00:18:30,217 --> 00:18:31,485
ASSOCIATED WITH THAT PHENOTYPE,

501
00:18:31,485 --> 00:18:33,921
WE CAN COUPLE THAT WITH REKURGZS

502
00:18:33,921 --> 00:18:35,656
PLATFORM WHICH LETS US KNOCK OUT

503
00:18:35,656 --> 00:18:37,157
PARTICULAR GENES AND OBSERVE A

504
00:18:37,157 --> 00:18:39,359
PHENOTYPE IN CELLS AND SO WE GET

505
00:18:39,359 --> 00:18:40,761
THIS VIRTUOUS CYCLE OF LEARNING

506
00:18:40,761 --> 00:18:43,430
WHERE WE'RE USING BOTH FORWARD

507
00:18:43,430 --> 00:18:44,765
IMENETTICS AND REVERSE GENETICS

508
00:18:44,765 --> 00:18:48,302
TO GET TO NOVEL TARGETS.

509
00:18:48,302 --> 00:18:50,137
ONCE WE HAVE THOSE TARGETS, AND

510
00:18:50,137 --> 00:18:52,239
WE HAVE COMPOUNDS THAT LOOK LIKE

511
00:18:52,239 --> 00:18:55,876
THEY'RE AFFECTING THEM THE

512
00:18:55,876 --> 00:18:58,478
INFORMIC SPACE WE CAN LOOK AT

513
00:18:58,478 --> 00:19:00,080
OUR CHEMISTRY MODELS LIKE OUR

514
00:19:00,080 --> 00:19:01,982
MATCH MAKER MODEL THAT PREDICTS

515
00:19:01,982 --> 00:19:02,883
WHICH PROTEINS A PARTICULAR

516
00:19:02,883 --> 00:19:04,051
COMPOUND WILL BYPASSED WITH AND

517
00:19:04,051 --> 00:19:06,453
FROM THIS MODEL WE ACTUALLY HAVE

518
00:19:06,453 --> 00:19:09,089
OVER 2 QUADRILLION DIFFERENT

519
00:19:09,089 --> 00:19:09,957
PROTEIN LIGAND INTERACTIONS

520
00:19:09,957 --> 00:19:13,293
WHERE WE SCORED HOW LIKELY IT IS

521
00:19:13,293 --> 00:19:14,861
THAT THOSE 2 ARE INTERACTING AND

522
00:19:14,861 --> 00:19:20,267
THAT LETS US GO FROM THE

523
00:19:20,267 --> 00:19:22,803
PHENOMMICS TARGET TO BEING ABLE

524
00:19:22,803 --> 00:19:24,104
TO HYPOTHESIZE WHAT BINDING

525
00:19:24,104 --> 00:19:24,671
TARGETS MIGHT BE.

526
00:19:24,671 --> 00:19:26,440
AND SO DOES THIS WORK?

527
00:19:26,440 --> 00:19:27,307
DOES THIS ACTUALLY MAKE A

528
00:19:27,307 --> 00:19:28,508
DIFFERENCE IN HOW FAST WE'RE

529
00:19:28,508 --> 00:19:30,777
ABLE TO MOVE OR HOW MANY

530
00:19:30,777 --> 00:19:31,645
DIFFERENT HYPOTHESIS WE'RE ABLE

531
00:19:31,645 --> 00:19:36,049
TO TEST, SO WE HAVE DATA ON THE

532
00:19:36,049 --> 00:19:38,251
FIRST 2 STAGES HERE OF A VOLUME

533
00:19:38,251 --> 00:19:40,387
WE CAN START TO AGGREGATE AND

534
00:19:40,387 --> 00:19:41,521
WHAT WE SLEEP APNEA AND OBESITY

535
00:19:41,521 --> 00:19:43,824
IS BEFORE WE DID THIS

536
00:19:43,824 --> 00:19:44,658
TREALIZATION PROCESS IN 2022, IT

537
00:19:44,658 --> 00:19:46,860
TOOK A TEAM OF 40 PEOPLE AND WE

538
00:19:46,860 --> 00:19:48,962
WERE ABLE TO EXPLORE 30 PROGRAM

539
00:19:48,962 --> 00:19:49,896
HYPOTHESIS, THE FOLLOWING YEAR,

540
00:19:49,896 --> 00:19:52,032
WHEN WE HAD STARTED THIS

541
00:19:52,032 --> 00:19:52,833
INDUSTRIALIZED PROCESS, THAT

542
00:19:52,833 --> 00:19:54,868
TEAM WAS COUNSEL TO 7 PEOPLE,

543
00:19:54,868 --> 00:19:56,703
AND THE NUMBER OF PROGRAM

544
00:19:56,703 --> 00:19:58,605
HYPOTHESIS HAD EXPLODED TO ABOUT

545
00:19:58,605 --> 00:19:59,906
115 OVER THE COURSE OF THE YEAR.

546
00:19:59,906 --> 00:20:02,342
AND THEN IF WE LOOK AT THE FIRST

547
00:20:02,342 --> 00:20:07,614
HALF OF 2024, THIS YEAR, WE NEED

548
00:20:07,614 --> 00:20:08,982
LESS THAN 1 FULL-TIME PERSON TO

549
00:20:08,982 --> 00:20:10,050
GET PROGRAMS FROM INITIATION ALL

550
00:20:10,050 --> 00:20:15,956
THE WAY TO WHERE WE'RE

551
00:20:15,956 --> 00:20:16,723
NOMINATING THEM FOR

552
00:20:16,723 --> 00:20:17,991
[INDISCERNIBLE] AND WE'RE ABLE

553
00:20:17,991 --> 00:20:21,695
TO DO THAT FOR 201 PROGRAM

554
00:20:21,695 --> 00:20:22,863
HYPOTHESIS GOING BO THE SYSTEM.

555
00:20:22,863 --> 00:20:24,965
SO WHAT THAT'S DO SUGGEST

556
00:20:24,965 --> 00:20:26,967
ENABLING US TO SHIFT THE

557
00:20:26,967 --> 00:20:28,402
TIMELINES UPON AND THE SHAPE OF

558
00:20:28,402 --> 00:20:31,471
THE DRUG DORPHY FUNNEL FROM A

559
00:20:31,471 --> 00:20:33,774
V-SHAPED FUNNEL THAT TAKES 1-2

560
00:20:33,774 --> 00:20:36,476
YEARS FOR THIS EARLY HIT

561
00:20:36,476 --> 00:20:37,310
DISCOVERY AND MILLIONS OF

562
00:20:37,310 --> 00:20:38,912
DOLLARS TO SOMETHING THAT TAKES

563
00:20:38,912 --> 00:20:40,947
WEEKS OR MONTHS AND CAN BE

564
00:20:40,947 --> 00:20:44,317
MEASURED IN THE THOUSANDS OF

565
00:20:44,317 --> 00:20:44,618
DOLLARS.

566
00:20:44,618 --> 00:20:46,086
SO IN THE TIME I HAVE LEFT, I

567
00:20:46,086 --> 00:20:47,320
WOULD LIKE TO TALK THROUGH A

568
00:20:47,320 --> 00:20:48,755
CASE STUDY TO SHOW YOU HOW THIS

569
00:20:48,755 --> 00:20:52,359
MIGHT BE APPLIED TO A PARTICULAR

570
00:20:52,359 --> 00:20:52,793
DISCOVERY PROGRAM.

571
00:20:52,793 --> 00:20:55,262
I'M GOING TO TALK ABOUT RBM39

572
00:20:55,262 --> 00:20:57,531
THIS, IS A PROGRAM THAT WAS 1 OF

573
00:20:57,531 --> 00:20:59,566
OUR EARLIER PROGRAMS IN IT SO IT

574
00:20:59,566 --> 00:21:01,168
DOESN'T HAVE ALL THE MODELS THAT

575
00:21:01,168 --> 00:21:02,636
I'VE TALKED THROUGH BUT IT IS 1

576
00:21:02,636 --> 00:21:04,504
THAT WE'VE BROUGHT ALL THE WAY

577
00:21:04,504 --> 00:21:05,706
THROUGH TO CLEARING AYE IERK ND

578
00:21:05,706 --> 00:21:09,376
AND SO I THINK IT'S A GREAT

579
00:21:09,376 --> 00:21:11,645
STORY TO TELL.

580
00:21:11,645 --> 00:21:12,446
SO CDK12 WAS INITIALLY

581
00:21:12,446 --> 00:21:14,314
IDENTIFIED AS A TARGET FOR HR

582
00:21:14,314 --> 00:21:15,215
PROFICIENT CANCERS BUT THE

583
00:21:15,215 --> 00:21:21,955
CHALLENGE IS BECAUSE THIS IS A

584
00:21:21,955 --> 00:21:23,657
KINASE, WE HAD NONSPECIFICKITY

585
00:21:23,657 --> 00:21:27,294
ISSUES FOR AN INDICATOR FOR

586
00:21:27,294 --> 00:21:31,531
CDK12 AND CDK13 ARE VERY SIMILAR

587
00:21:31,531 --> 00:21:35,502
AND SO MOST CDK INHIBITORS ALSO

588
00:21:35,502 --> 00:21:36,369
ENDED UP INHIBITING CDK13 AND

589
00:21:36,369 --> 00:21:38,305
THAT CAUSED A LOT OF TOXICITY.

590
00:21:38,305 --> 00:21:42,976
WHEN WE LOOKED AT OUR PHENOMAP,

591
00:21:42,976 --> 00:21:45,812
WE SAW A GENE RBM39, WHERE THE

592
00:21:45,812 --> 00:21:49,015
KNOCK OUT OF RBM39 MIMICS THE

593
00:21:49,015 --> 00:21:51,218
KNOCK OUT OF CDK12 BUT IT

594
00:21:51,218 --> 00:21:52,919
SHOULDN'T SHOW THAT SAME

595
00:21:52,919 --> 00:21:54,054
RELATIONSHIP WITH CDK13 SO THIS

596
00:21:54,054 --> 00:21:56,056
SEEMED LIKE A GREAT TARGET FOR

597
00:21:56,056 --> 00:22:00,927
GETTING THE EFFECTS OF CDK12

598
00:22:00,927 --> 00:22:03,463
WITHOUT HAVING THE INHIBITION OF

599
00:22:03,463 --> 00:22:04,064
CDK13.

600
00:22:04,064 --> 00:22:05,398
WHEN WE SCREENED SMALL

601
00:22:05,398 --> 00:22:07,801
MOLECULES, WHAT WE SAW, WAS THAT

602
00:22:07,801 --> 00:22:10,437
THERE ARE SMALL MOLECULES THAT

603
00:22:10,437 --> 00:22:13,373
TARGET RBM 39, AND SHOW

604
00:22:13,373 --> 00:22:14,441
SIMILARITY TO CDK12 WITHOUT

605
00:22:14,441 --> 00:22:16,610
HAVING ANY SIMILARITY TO CDK13

606
00:22:16,610 --> 00:22:21,882
AND SO THIS FELT REALLY, REALLY

607
00:22:21,882 --> 00:22:22,149
EXCITING.

608
00:22:22,149 --> 00:22:23,784
WITH OUR MATCH MAKER PRODUCT,

609
00:22:23,784 --> 00:22:27,287
WHICH PREDICTS THE BINDING OF

610
00:22:27,287 --> 00:22:29,956
COMPOUNDS TO PROTEIN POCKETS, WE

611
00:22:29,956 --> 00:22:31,725
WERE ABLE TO IDENTIFY -- WE WERE

612
00:22:31,725 --> 00:22:33,627
ABLE TO PREDICT THAT THESE NEW

613
00:22:33,627 --> 00:22:36,763
COMPOUNDS WERE NOT IN FACT CDK

614
00:22:36,763 --> 00:22:39,933
INHIBITORS AND SO HERE, TOO, IS

615
00:22:39,933 --> 00:22:40,867
A SIGNIFICANT SIMILARITY SCORE.

616
00:22:40,867 --> 00:22:43,503
WE CAN SEE THAT POSITIVE

617
00:22:43,503 --> 00:22:46,473
CONTROLS, CDK12 AND 13

618
00:22:46,473 --> 00:22:48,041
INHIBITORS ARE SHOWING, ARE

619
00:22:48,041 --> 00:22:52,445
SCORING AS BEING PREDICTED AS

620
00:22:52,445 --> 00:22:55,315
BEING THE KINASES HERE BUT THESE

621
00:22:55,315 --> 00:23:00,554
2 NEW MOLECULES, THESE RECURSION

622
00:23:00,554 --> 00:23:01,655
MOLECULES ARE NOT PREDICTED THAT

623
00:23:01,655 --> 00:23:03,056
WAY EMPLOY WE CAN PHYSICALLY

624
00:23:03,056 --> 00:23:05,091
KREEN TO CONFIRM THAT DIGITAL

625
00:23:05,091 --> 00:23:06,026
HYPOTHESIS AND RUN AN EXPERIMENT

626
00:23:06,026 --> 00:23:07,727
THAT MATTERS AND WHAT WE CAN SEE

627
00:23:07,727 --> 00:23:11,731
HERE IS WHEN WE SCREEN FOR CDK12

628
00:23:11,731 --> 00:23:14,701
ACTIVITY, WE CAN SEE THAT CDK 12

629
00:23:14,701 --> 00:23:16,670
INHIBITORS ARE ACTUALLY BINDING

630
00:23:16,670 --> 00:23:20,006
TO CDK12, OUR HIT COMPOUNDS ARE

631
00:23:20,006 --> 00:23:22,742
NOT BINDING TO CDK12 DESPITE

632
00:23:22,742 --> 00:23:23,643
HAVING THE SAME PHENOMMIC CENTER

633
00:23:23,643 --> 00:23:24,845
FOR EXCELLENCE ON AGING AND THAT

634
00:23:24,845 --> 00:23:26,479
LED US TO FEEL LIKE A REALLY EBS

635
00:23:26,479 --> 00:23:28,048
SIGHTING DREK TO TAKE THIS

636
00:23:28,048 --> 00:23:28,381
PROGRAM.

637
00:23:28,381 --> 00:23:30,550
IN ORDER TO DEVELOP THAT

638
00:23:30,550 --> 00:23:32,853
COMPOUND AND GET IT TO A

639
00:23:32,853 --> 00:23:34,955
COMPOUND THAT WE CAN DOSE INTO

640
00:23:34,955 --> 00:23:41,795
AN ANIMAL WE HAVE TO ITERATE ON

641
00:23:41,795 --> 00:23:43,430
THE COMPOUND DESEEN AND THAT

642
00:23:43,430 --> 00:23:45,131
GIVES US A CHANCE TO TEST IT

643
00:23:45,131 --> 00:23:47,100
AGAINST THAT RESPONSE AND NOT

644
00:23:47,100 --> 00:23:52,472
JUST ABOUT RBM12 AND CDK39, BUT

645
00:23:52,472 --> 00:23:53,673
MAKING SURE WE'RE ARE NOT

646
00:23:53,673 --> 00:23:55,542
PICKING UP ACTIVITY WE DON'T

647
00:23:55,542 --> 00:23:55,876
WANT.

648
00:23:55,876 --> 00:23:58,211
SO AS WE INCREASE CONCENTRATION,

649
00:23:58,211 --> 00:23:59,646
WE CAN EVALUATE SIMILARITY TO

650
00:23:59,646 --> 00:24:01,214
THESE DIFFERENT TARGETS AND GET

651
00:24:01,214 --> 00:24:02,549
A NICE CONCENTRATION RESPONSE

652
00:24:02,549 --> 00:24:04,985
CURVE THAT ALLOWS US TO ITERATE

653
00:24:04,985 --> 00:24:06,753
ON IMPROVING MOLECULES AND THAT

654
00:24:06,753 --> 00:24:08,455
LETS US DRIVE STRUCTURE ACTIVITY

655
00:24:08,455 --> 00:24:08,788
RELATIONSHIPS.

656
00:24:08,788 --> 00:24:11,057
WE CAN SEE HERE THAT FROM THIS

657
00:24:11,057 --> 00:24:13,460
COMPOUND WE'RE ABLE TO MAKE

658
00:24:13,460 --> 00:24:14,728
VARIATIONS, CHANGE SOME OF THE

659
00:24:14,728 --> 00:24:18,698
R-GROUPS AND THEN PROFILE WITHIN

660
00:24:18,698 --> 00:24:21,201
THE PHENOMMIC MAP TO MAKE SURE

661
00:24:21,201 --> 00:24:22,002
WE'RE MAINTAINING THE RESPONSE

662
00:24:22,002 --> 00:24:25,839
WE WANT TO SEE AND WHEN WE TEST

663
00:24:25,839 --> 00:24:26,873
FOR RBM39 DEGRADATION WE SEE AN

664
00:24:26,873 --> 00:24:28,208
ECHOING OF THAT EFFECT.

665
00:24:28,208 --> 00:24:30,911
SO WE KNOW THAT THE PHENOMMIC

666
00:24:30,911 --> 00:24:33,947
INFER ENSEL MAP IS TELLING US

667
00:24:33,947 --> 00:24:35,081
WHAT THE PHYSICAL EXPERIMENTS

668
00:24:35,081 --> 00:24:36,783
WOULD TELL US AS WELL.

669
00:24:36,783 --> 00:24:38,285
AND THEN FINALLY EXCITING 3 THIS

670
00:24:38,285 --> 00:24:40,954
COMPOUND GOT TO THE PLACE WHERE

671
00:24:40,954 --> 00:24:44,457
WE WERE ABLE TO DOSE IT IN A

672
00:24:44,457 --> 00:24:47,127
TUMOR MODEL, AND SO, WHAT WE SEE

673
00:24:47,127 --> 00:24:49,029
HERE IS WITH OUR HIT COMPOUND,

674
00:24:49,029 --> 00:24:52,399
WE SEE A SIGNIFICANT REDUCTION

675
00:24:52,399 --> 00:24:53,667
IN THE TUMOR VOLUME OVER TIME

676
00:24:53,667 --> 00:24:56,469
AND WE ALSO SEE AN INCREASE IN

677
00:24:56,469 --> 00:24:58,505
OUR SURVIVAL RATES OVER TIME

678
00:24:58,505 --> 00:25:04,244
COMPARED TO THE VEHICLE.

679
00:25:04,244 --> 00:25:05,779
AND SO WHAT THIS STORY SHOULD

680
00:25:05,779 --> 00:25:07,147
ILLUSTRATE IS THAT THESE

681
00:25:07,147 --> 00:25:08,982
PREDICTIONS AND MINIMAL STANDARD

682
00:25:08,982 --> 00:25:11,551
EXPERIMENTS WILL LET US RAPIDLY

683
00:25:11,551 --> 00:25:12,852
AND EFFICIENTLY IDENTIFY A

684
00:25:12,852 --> 00:25:13,486
DEVELOPMENT CANDIDATE.

685
00:25:13,486 --> 00:25:15,889
THESE ARE THE NUMBERS FOR THE

686
00:25:15,889 --> 00:25:18,191
RBM 39 PROGRAM WHERE YOU CAN SEE

687
00:25:18,191 --> 00:25:19,359
INSILICO, WE STARTED VERY, VERY

688
00:25:19,359 --> 00:25:20,794
WIDE AND WE WERE ABLE TO SCREEN

689
00:25:20,794 --> 00:25:23,330
A VERY LARGE NUMBER OF MOLECULES

690
00:25:23,330 --> 00:25:25,999
INSILICO, WE WERE ABLE TO IN

691
00:25:25,999 --> 00:25:26,666
PHENOMMICS EXPERIMENTS GO

692
00:25:26,666 --> 00:25:30,270
THROUGH A SMALLER SET OF THOSE

693
00:25:30,270 --> 00:25:31,871
COMPOUNDS AND DRIVE RELATIVELY

694
00:25:31,871 --> 00:25:33,440
QUICKLY TO CANDIDATE QUALITY

695
00:25:33,440 --> 00:25:33,740
COMPOUNDS.

696
00:25:33,740 --> 00:25:35,942
AND WE WERE ABLE TO DO THIS IN A

697
00:25:35,942 --> 00:25:37,610
RELATIVELY SHORT TIME, IT TOOK

698
00:25:37,610 --> 00:25:39,145
US 9 MONTHS FROM STARTING THIS

699
00:25:39,145 --> 00:25:40,981
PROGRAM TO HAVING SOMETHING THAT

700
00:25:40,981 --> 00:25:43,416
FELT LIKE IT HAD AN APPROPRIATE

701
00:25:43,416 --> 00:25:45,151
THERAPEUTIC INDEX TO MOVE

702
00:25:45,151 --> 00:25:45,485
FORWARD WITH.

703
00:25:45,485 --> 00:25:50,590
AND IN FACT, THIS COMPOUND JUST

704
00:25:50,590 --> 00:25:51,725
CLEARED IND EARLIER THIS MONTH

705
00:25:51,725 --> 00:25:53,159
AND SO WE ANTICIPATE STARTING

706
00:25:53,159 --> 00:25:54,961
PHASE 1 PHASE 2 TRIALS FOR THAT

707
00:25:54,961 --> 00:25:58,398
AT THE BEGINNING OF THE NEXT

708
00:25:58,398 --> 00:25:58,698
YEAR.

709
00:25:58,698 --> 00:26:02,168
AND SO IN SUMMARY, WE'RE REALLY

710
00:26:02,168 --> 00:26:04,204
EXCITED ABOUT THIS IDEA OF DATA

711
00:26:04,204 --> 00:26:06,272
LOOPS, THIS IDEA OF RUNNING

712
00:26:06,272 --> 00:26:08,641
EXPERIMENTS WHERE WE NEED TO IN

713
00:26:08,641 --> 00:26:09,709
ORDER TO IMPROVE OUR MODELS

714
00:26:09,709 --> 00:26:11,544
WHICH WILL TELL US BETTER WHICH

715
00:26:11,544 --> 00:26:20,053
SPRMS WE NEED TO RUN.

716
00:26:20,053 --> 00:26:21,521
ANDLET END GAME IS THAT PATIENTS

717
00:26:21,521 --> 00:26:22,756
ARE WAITING FOR THIS AND WE NEED

718
00:26:22,756 --> 00:26:26,326
TO FIND NEW WAYS TO DELIVER

719
00:26:26,326 --> 00:26:27,694
THIS, AND SO FAR WHAT WE'VE

720
00:26:27,694 --> 00:26:28,661
TALKED ABOUT AT THIS WORKSHOP

721
00:26:28,661 --> 00:26:29,929
ARE THE WAY TO MOVE THIS

722
00:26:29,929 --> 00:26:30,163
FORWARD.

723
00:26:30,163 --> 00:26:35,802
THAT'S ALL I HAVE.

724
00:26:35,802 --> 00:26:37,037
>> THANK YOU.

725
00:26:37,037 --> 00:26:40,440
THANK YOU SO MUCH, MARISSA FOR

726
00:26:40,440 --> 00:26:41,174
AN EXCELLENT PRESENTATION.

727
00:26:41,174 --> 00:26:43,043
I WILL SEE IF THERE ARE ANY

728
00:26:43,043 --> 00:26:43,376
QUESTIONS.

729
00:26:43,376 --> 00:26:46,913
THERE ARE SOME QUESTIONS IN THE

730
00:26:46,913 --> 00:26:50,517
CHAT, WHICH I WILL LEAVE FOR YOU

731
00:26:50,517 --> 00:26:51,684
MARISSA, IF YOU CAN ANSWER THOSE

732
00:26:51,684 --> 00:26:53,453
THAT WOULD BE GREAT.

733
00:26:53,453 --> 00:26:54,954
I AM EXCITED TO LEARN MORE ABOUT

734
00:26:54,954 --> 00:26:57,924
THE MATCH MAKER MODEL, I AM

735
00:26:57,924 --> 00:27:00,527
CURIOUS WHAT INPUTS WERE USED

736
00:27:00,527 --> 00:27:02,262
FOR THE MODEL, WAS IT FRAMED ON

737
00:27:02,262 --> 00:27:03,430
DIFFERENT PROTEIN LIVING

738
00:27:03,430 --> 00:27:05,265
CONTEXTS AND THE WORD OF

739
00:27:05,265 --> 00:27:07,100
PROTEINS AND LIGANDS, DELIVERED

740
00:27:07,100 --> 00:27:07,400
SEPARATELY.

741
00:27:07,400 --> 00:27:10,503
SO IT WAS ACTUALLY TRAINED ON A

742
00:27:10,503 --> 00:27:13,840
LARGE SUITE OF DATA INCLUDING

743
00:27:13,840 --> 00:27:15,942
KNOWN PROTEIN LIGAND

744
00:27:15,942 --> 00:27:16,676
INTERACTIONS INCLUDING

745
00:27:16,676 --> 00:27:17,477
BIOCHEMICAL ASSAY INFORMATION

746
00:27:17,477 --> 00:27:19,679
FOR BINDING AND SO, THOSE --

747
00:27:19,679 --> 00:27:21,915
THAT WAS THE INPUT AND THEN IT

748
00:27:21,915 --> 00:27:24,484
HAS BEEN SHOWN TO BE ABLE TO

749
00:27:24,484 --> 00:27:25,618
MAKE THESE PREDICTIONS FOR

750
00:27:25,618 --> 00:27:27,587
THINGS THAT IT HAD NOT SEEN

751
00:27:27,587 --> 00:27:31,624
BEFORE, SO IT DOES REST ON THAT

752
00:27:31,624 --> 00:27:32,459
KIND OF DATA.

753
00:27:32,459 --> 00:27:34,461
>> WE WILL TAKE 1 MORE
QUESTION 

754
00:27:34,461 --> 00:27:36,262
IT'S GREAT TAKEN--THEY THE MODEL

755
00:27:36,262 --> 00:27:39,833
HAS A LOT IN TIME AND COST,

756
00:27:39,833 --> 00:27:41,835
EARLY LEAD DISCOVERY, WHAT'S

757
00:27:41,835 --> 00:27:43,903
YOUR STRATEGY TO DO THE SAME IN

758
00:27:43,903 --> 00:27:44,904
THE CLINICAL OUTCOME?

759
00:27:44,904 --> 00:27:46,439
>> IT'S A REALLY, REALLY
GOOD

760
00:27:46,439 --> 00:27:46,739
QUESTION.

761
00:27:46,739 --> 00:27:48,908
IN THE CLINICAL SPACE, WE KNOW

762
00:27:48,908 --> 00:27:50,076
THAT EARLY CLINICAL SCREENS ARE

763
00:27:50,076 --> 00:27:51,478
RELATIVELY SMALL IN TERMS OF

764
00:27:51,478 --> 00:27:54,481
PATIENT NUMBERS, WE RUN THE

765
00:27:54,481 --> 00:27:55,915
FIRST FEW CYCLES THROUGH IN A

766
00:27:55,915 --> 00:27:57,817
SMALL HANDFUL OF PATIENTS AND SO

767
00:27:57,817 --> 00:27:59,319
IT'S REALLY IMPORTANT TO BE ABLE

768
00:27:59,319 --> 00:28:02,388
TO GET THE PATIENT PROFILE RIGHT

769
00:28:02,388 --> 00:28:04,257
BECAUSE WE -- WE KNOW THAT IN

770
00:28:04,257 --> 00:28:06,860
THIS AGE OF PERSONALIZED

771
00:28:06,860 --> 00:28:08,027
MEDICINE, PARTICULARLY IN

772
00:28:08,027 --> 00:28:09,896
ONCOLOGY, EACH PATIENT IS

773
00:28:09,896 --> 00:28:10,730
SLIDELY DIFFERENT AND WHAT WE

774
00:28:10,730 --> 00:28:12,465
WANT TO DO IS IDENTIFY THE

775
00:28:12,465 --> 00:28:15,435
PATIENT COHORT THAT BEST FITS

776
00:28:15,435 --> 00:28:16,736
OUR TREATMENT, AND SO 1 OF THE

777
00:28:16,736 --> 00:28:19,639
WAYS THAT MODELS CAN HELP IS IN

778
00:28:19,639 --> 00:28:20,840
FINDING THEERATE PATIENT

779
00:28:20,840 --> 00:28:22,575
POPULATION TO BRING INTO OUR

780
00:28:22,575 --> 00:28:23,710
CLINICAL TRIALS.

781
00:28:23,710 --> 00:28:24,744
THERE'S ALSO MORE LOGISTICAL

782
00:28:24,744 --> 00:28:27,080
THINGS WE CAN DO IN THE CLINICAL

783
00:28:27,080 --> 00:28:30,383
TRIAL SPACE, IN TERMS OF KNOWING

784
00:28:30,383 --> 00:28:33,453
HOW BEST TO RECRUIT PEOPLE INTO

785
00:28:33,453 --> 00:28:36,656
THESE TRIALS IN ORDER TO

786
00:28:36,656 --> 00:28:37,957
IDENTIFY WHAT IS THE POPULATION

787
00:28:37,957 --> 00:28:39,392
WE NEED TO OPEN THIS UP TO IN

788
00:28:39,392 --> 00:28:40,827
ORDER TO GET ENOUGH PEOPLE

789
00:28:40,827 --> 00:28:42,128
ENROLLED AND SO I THINK THERE

790
00:28:42,128 --> 00:28:44,664
ARE A WIDE RANGE OF WAYS THAT WE

791
00:28:44,664 --> 00:28:46,799
CAN USE SOME OF THESE SAME TYPES

792
00:28:46,799 --> 00:28:51,504
OF STRATEGIES IN ORDER TO

793
00:28:51,504 --> 00:28:54,741
ACCELERATE CLINICAL PHASES.

794
00:28:54,741 --> 00:28:56,009
>> ONE MORE QUESTION:  CAN
YOU

795
00:28:56,009 --> 00:28:59,312
PLEASE SHED MORE LIGHT ON --

796
00:28:59,312 --> 00:29:00,413
OOPS, SORRY ON VALIDATION OF

797
00:29:00,413 --> 00:29:05,518
MODEL AND IN CASE OF ISOFORMS

798
00:29:05,518 --> 00:29:07,086
OUR SELECTIVITY AND SPECIFICITY

799
00:29:07,086 --> 00:29:07,887
WILL HANDLE?

800
00:29:07,887 --> 00:29:08,888
>> YEAH, SO I'M ACTUALLY
HAPPY

801
00:29:08,888 --> 00:29:11,991
TO REFER YOU TO SOME OF THE

802
00:29:11,991 --> 00:29:12,792
PUBLICATIONS AROUND MATCH MAKER.

803
00:29:12,792 --> 00:29:15,161
I DON'T HAVE THOSE CITATIONS ON

804
00:29:15,161 --> 00:29:15,795
HAND.

805
00:29:15,795 --> 00:29:17,463
I WILL LOOK THEM UP OFFLINE HERE

806
00:29:17,463 --> 00:29:19,365
AND HAPPY TO SEND THAT ON TO

807
00:29:19,365 --> 00:29:19,532
YOU.

808
00:29:19,532 --> 00:29:21,968
IN GENERAL MOST OF THE WORK THAT

809
00:29:21,968 --> 00:29:25,471
WE'RE DOING HERE AT RECURSION,

810
00:29:25,471 --> 00:29:27,574
MOST OF THE PAPERS WE PUBLISHED

811
00:29:27,574 --> 00:29:33,112
ARE AVAILABLE ON THE WEBSITE, SO

812
00:29:33,112 --> 00:29:33,446
RECURSION.COM.

813
00:29:33,446 --> 00:29:34,847
THERE IS A LITERATURE AREA THERE

814
00:29:34,847 --> 00:29:36,115
THAT YOU CAN LOOK AND SEE THE

815
00:29:36,115 --> 00:29:38,218
BACKING SCIENCE FOR THESE

816
00:29:38,218 --> 00:29:38,451
MODELS.

817
00:29:38,451 --> 00:29:40,353
>> ONE LAST QUESTION, I
WILL GO

818
00:29:40,353 --> 00:29:42,822
WITH, HOW TO ELUC DATE DRUG

819
00:29:42,822 --> 00:29:45,858
MECHANISM OF ACTION BASED ON

820
00:29:45,858 --> 00:29:46,526
PHENOMMICS DATA?

821
00:29:46,526 --> 00:29:48,361
>> YOU KNOW THAT'S AN
AWESOME

822
00:29:48,361 --> 00:29:49,195
QUESTION AND IT'S SOMETHING WE

823
00:29:49,195 --> 00:29:50,463
THINK ABOUT A LOT.

824
00:29:50,463 --> 00:29:54,000
I THINK THE TERM MECHANISM OF

825
00:29:54,000 --> 00:29:54,801
ACTION IS ACTUALLY A LITTLE BIT

826
00:29:54,801 --> 00:29:56,469
FUZZY AND IT MEANS DIFFERENT

827
00:29:56,469 --> 00:29:57,737
THINGS TO DIFFERENT PEOPLE, SO 1

828
00:29:57,737 --> 00:29:59,739
OF THE THINGS THAT WE THINK

829
00:29:59,739 --> 00:30:01,007
ABOUT WHEN WE THINK ABOUT

830
00:30:01,007 --> 00:30:03,476
MECHANISM OF ACTION IS WHAT ARE

831
00:30:03,476 --> 00:30:04,544
THE GENERAL BIOLOGICAL PATHWAYS

832
00:30:04,544 --> 00:30:07,547
THAT ARE BEING ACTIVATED AND TO

833
00:30:07,547 --> 00:30:08,648
ANSWER THAT QUESTION.

834
00:30:08,648 --> 00:30:10,650
WE USE THE PHENOMMICS DATA AND

835
00:30:10,650 --> 00:30:12,151
THE TRANSKRPT ORDER OF

836
00:30:12,151 --> 00:30:14,787
MICRONSICS DATA AND WE'RE --

837
00:30:14,787 --> 00:30:17,323
TRANSCRIPTOMICS DATA AND WE'RE

838
00:30:17,323 --> 00:30:18,691
ABLE TO FIND SIMILAR YEENS AND

839
00:30:18,691 --> 00:30:19,959
KNOW HOW THEY'RE ACTING AND ALSO

840
00:30:19,959 --> 00:30:23,162
ABLE TO LOOK AT THE

841
00:30:23,162 --> 00:30:25,598
TRANSCRIPTOMIC REGULATION AND

842
00:30:25,598 --> 00:30:25,865
 SAY

843
00:30:25,865 --> 00:30:28,968
WHAT ARE THE PATHWAYS THAT ARE

844
00:30:28,968 --> 00:30:29,369
SHIFTING.

845
00:30:29,369 --> 00:30:30,903
THERE'S A QUESTION AROUND MOA

846
00:30:30,903 --> 00:30:32,538
HAD IS WHAT IS THE TARGET THE

847
00:30:32,538 --> 00:30:33,873
DRUG IS BINDING TOO IN AND THAT

848
00:30:33,873 --> 00:30:36,509
IS 1 THAT REALLY HELPS TO

849
00:30:36,509 --> 00:30:37,110
ACCELERATE DRUG STOEVERY DOWN

850
00:30:37,110 --> 00:30:38,144
STREAM BUT IT'S NOT NECESSARILY

851
00:30:38,144 --> 00:30:39,445
SOMETHING THAT WE HAVE TO HAVE

852
00:30:39,445 --> 00:30:41,981
EARLY ON AND IT'S ALSO NOT

853
00:30:41,981 --> 00:30:43,549
SOMETHING THAT'S ABSOLUTELY

854
00:30:43,549 --> 00:30:45,318
REQUIRED AND SO, THE EXACT

855
00:30:45,318 --> 00:30:46,519
BINDING TARGET IS SOMETHING WE

856
00:30:46,519 --> 00:30:47,820
WOULD TRY TO PREDICT THROUGH

857
00:30:47,820 --> 00:30:49,289
MATCH MAKER AND WE ALSO TEST

858
00:30:49,289 --> 00:30:50,857
THINGS THAT ARE SHOWING STRONG

859
00:30:50,857 --> 00:30:52,191
SIMILARITY IN THE MAP TO SEE IF

860
00:30:52,191 --> 00:30:55,428
WE DO SEE THAT BINDING AND IT IS

861
00:30:55,428 --> 00:30:57,330
SOMETHING THAT IS AN ITERATIVE

862
00:30:57,330 --> 00:30:58,431
EFFORT THAT WE PARTICULARLY PAY

863
00:30:58,431 --> 00:31:02,702
ATTENTION TO IN LATER STAGES OF

864
00:31:02,702 --> 00:31:04,504
DRUG DESIGN.

865
00:31:04,504 --> 00:31:06,306
>> MARISSA, IF YOU CAN
STAY

866
00:31:06,306 --> 00:31:08,041
ONLINE AND ANSWER SOME OF THE

867
00:31:08,041 --> 00:31:09,742
QUESTIONS IN THE CHAT, THAT

868
00:31:09,742 --> 00:31:10,109
WOULD BE GREAT.

869
00:31:10,109 --> 00:31:15,281
WE HAVE TO MOVE TO OUR NEXT

870
00:31:15,281 --> 00:31:16,015
SPEAKER.

871
00:31:16,015 --> 00:31:16,416
>> ABSOLUTELY.

872
00:31:16,416 --> 00:31:18,484
>> THANK YOU AGAIN
MARISSA.

873
00:31:18,484 --> 00:31:24,991
THE LAST SPEAKER IN THIS SESSION

874
00:31:24,991 --> 00:31:27,727
IS ALEX ZHAVORONKOV, ALEX IS THE

875
00:31:27,727 --> 00:31:32,699
FOUNDER AND LEADER OF THE

876
00:31:32,699 --> 00:31:35,601
INSILICO MEDICINE GENERATING

877
00:31:35,601 --> 00:31:37,003
ARTIFICIAL IN. TELEIGENCE AND

878
00:31:37,003 --> 00:31:41,874
PLATFORMS FOR DRUG DISCOVERY. 

879
00:31:41,874 --> 00:31:43,643
SINCE 2007 HE HAS DONE

880
00:31:43,643 --> 00:31:45,545
REINFORCEMENT LEARNING AND THE

881
00:31:45,545 --> 00:31:47,747
GENERATION OF MOLECULAR

882
00:31:47,747 --> 00:31:50,850
STRUCTURES THE, THE DESIRED

883
00:31:50,850 --> 00:31:52,752
PROPERTIES AND [INDISCERNIBLE]

884
00:31:52,752 --> 00:31:55,021
AND PATIENT DATA.

885
00:31:55,021 --> 00:31:57,957
UNDER HIS LEADERSHIP INSILICO

886
00:31:57,957 --> 00:31:59,992
RAISED OVER 400 MILLION IN

887
00:31:59,992 --> 00:32:01,060
MEDICAL GRAPTS FROM

888
00:32:01,060 --> 00:32:03,963
BIOTECHNOLOGY, HEALTHCARE AND

889
00:32:03,963 --> 00:32:08,334
FINANCIAL INVESTORS, R&D

890
00:32:08,334 --> 00:32:10,670
RESEARCH CENTERS IN SEVERAL

891
00:32:10,670 --> 00:32:11,237
REGIONS.

892
00:32:11,237 --> 00:32:13,072
THE COMPANY HAS NOMINATED 18

893
00:32:13,072 --> 00:32:15,007
PRECLINICAL CANDIDATES SO FAR,

894
00:32:15,007 --> 00:32:16,409
STARTED CLINICAL TRIALS AND

895
00:32:16,409 --> 00:32:18,644
ENTERED PHASE 2 WITH AN AI

896
00:32:18,644 --> 00:32:23,015
DISCOVERED NOVEL TARGET AND AI

897
00:32:23,015 --> 00:32:24,016
DESIGNED NOVEL MOLECULE.

898
00:32:24,016 --> 00:32:26,619
ALEX, I CAN HAND IT OVER TO YOU

899
00:32:26,619 --> 00:32:30,656
IF YOU ARE READY WITH YOUR

900
00:32:30,656 --> 00:32:30,990
PRESENTATION.

901
00:32:30,990 --> 00:32:32,859
>> READY WITH THE
PRESENTATION,

902
00:32:32,859 --> 00:32:34,327
THANK YOU SHYAM, AND GREAT TO BE

903
00:32:34,327 --> 00:32:35,461
PRESENTING TO YOU TODAY.

904
00:32:35,461 --> 00:32:37,630
THANK YOU FOR THIS GENEROUS

905
00:32:37,630 --> 00:32:39,565
INTRODUCTION AND THABILITY TO

906
00:32:39,565 --> 00:32:42,168
PRESENT WITH SUCH ESTEEMED

907
00:32:42,168 --> 00:32:42,435
SPEAKERS.

908
00:32:42,435 --> 00:32:49,509
LET ME SHARE MY SCREEN AND START

909
00:32:49,509 --> 00:32:49,809
PRESENTING.

910
00:32:49,809 --> 00:32:54,747
SO MY TALK TODAY WILL FOCUS ON

911
00:32:54,747 --> 00:32:57,083
THE CAPABILITIES OF END-TO-END

912
00:32:57,083 --> 00:32:59,285
GENERATIVE AI AND ROBOTICS FOR

913
00:32:59,285 --> 00:33:03,089
DRUG DORPHY.

914
00:33:03,089 --> 00:33:05,691
I HAVE CASE STUDIES SHOWING WHAT

915
00:33:05,691 --> 00:33:06,993
AI DRUG DISCOVERY CAN DO TODAY

916
00:33:06,993 --> 00:33:10,329
AND WHAT WE EXPECT IT TO BE ABLE

917
00:33:10,329 --> 00:33:14,133
TO DO IN THE NEAR FUTURE.

918
00:33:14,133 --> 00:33:17,637
THERE IS A LOT OF HYPE AROUND

919
00:33:17,637 --> 00:33:20,673
THIS AREA, AND MANY COMPANIES

920
00:33:20,673 --> 00:33:22,909
ARE FORMED EVERY DAY, OR EVERY

921
00:33:22,909 --> 00:33:28,014
WEEK AND OF COURSE, LET'S HOPE

922
00:33:28,014 --> 00:33:29,182
FOR PROGRESS IN ACADEMIA, MUCH

923
00:33:29,182 --> 00:33:30,983
OF THAT IS REAL AND I WOULD LIKE

924
00:33:30,983 --> 00:33:33,953
TO TALK A LITTLE BIT ABOUT THE

925
00:33:33,953 --> 00:33:36,589
RECENT EXAMPLES.

926
00:33:36,589 --> 00:33:40,993
SO, A LITTLE BIT ABOUT INSILICO,

927
00:33:40,993 --> 00:33:45,097
JUST LIKE MARISSA PRESENTED FOR

928
00:33:45,097 --> 00:33:46,999
RECURSION, WE ARE ALSO DIVERSE

929
00:33:46,999 --> 00:33:48,100
AND FLOABALLY PRESENT.

930
00:33:48,100 --> 00:33:50,369
OUR HEADQUARTERS, GLOBAL

931
00:33:50,369 --> 00:33:51,737
HEADQUARTERS IS IN BOSTON, WE

932
00:33:51,737 --> 00:33:55,575
JUST MOVED IT FROM NEW YORK.

933
00:33:55,575 --> 00:34:00,179
SO 1000 MASS AVENUE, WELCOME TO

934
00:34:00,179 --> 00:34:00,513
JOIN.

935
00:34:00,513 --> 00:34:05,151
OUR AI IS DONE IN MONTREAL

936
00:34:05,151 --> 00:34:08,621
CANADA, THAT'S WHERE WE DO

937
00:34:08,621 --> 00:34:13,392
THEORY AI AND IT'S ALSO DONE IN

938
00:34:13,392 --> 00:34:13,960
ABU DHABI, HIGHLY RECOMMEND

939
00:34:13,960 --> 00:34:15,428
VISITING AND WE ALSO HAVE GREAT

940
00:34:15,428 --> 00:34:20,366
PRESENCE IN CHINA, SMALL TEEM IN

941
00:34:20,366 --> 00:34:26,506
HONG KONG, TAI PEI, SHANGHAI,

942
00:34:26,506 --> 00:34:29,475
SUZHOW, AND WE UTILIZE THESE TO

943
00:34:29,475 --> 00:34:31,344
RAPIDLY DESIGN AND DISCOVER

944
00:34:31,344 --> 00:34:33,045
DRUGS, HOWEVER OUR MISSION AS A

945
00:34:33,045 --> 00:34:35,548
COMPANY IS TO EXTENT HEALTHY

946
00:34:35,548 --> 00:34:36,182
PRODUCTIVE LONGEVITY FOR

947
00:34:36,182 --> 00:34:40,953
EVERYONE AND OUR GROUND GOAL IS

948
00:34:40,953 --> 00:34:41,420
TO SOLVE AGING.

949
00:34:41,420 --> 00:34:43,356
HAD IS HOW THE COMPANY IS

950
00:34:43,356 --> 00:34:46,459
ORGANIZED SO PRETTY IMPORTANT

951
00:34:46,459 --> 00:34:46,993
SLIDE.

952
00:34:46,993 --> 00:34:51,030
WE HAVE OUR OWN LLMs THAT HELP

953
00:34:51,030 --> 00:34:53,432
ORCHESTRATE THE MAIN SPECIFIC

954
00:34:53,432 --> 00:34:53,666
MODELS.

955
00:34:53,666 --> 00:34:56,402
OF COURSE WE ALSO UTILIZE LARGE

956
00:34:56,402 --> 00:34:58,004
FOUNDATIONAL MODELS TO CREATE

957
00:34:58,004 --> 00:35:01,240
SOME OF THE AGENTS THAT WE USE

958
00:35:01,240 --> 00:35:02,575
EVERYWHERE.

959
00:35:02,575 --> 00:35:04,176
WE HAVE OVER 800, THE MAIN

960
00:35:04,176 --> 00:35:06,579
SPECIFIC MODELS, SOME OF THEM

961
00:35:06,579 --> 00:35:10,182
ARE TRANSFORMER BASED, SOME ARE

962
00:35:10,182 --> 00:35:13,119
[INDISCERNIBLE], SOME ARE

963
00:35:13,119 --> 00:35:14,820
DIFFUSION, DIFFUSION

964
00:35:14,820 --> 00:35:16,188
TRANSFORMERS, AND WHAT ARE

965
00:35:16,188 --> 00:35:18,424
LLMs ARE REALLY GOOD AT IS

966
00:35:18,424 --> 00:35:20,593
ORCHESTRATING SO IF YOU HAVE API

967
00:35:20,593 --> 00:35:23,563
TO THOSE MODEL ANDS GREAT

968
00:35:23,563 --> 00:35:26,966
DESCRIPTION, YOU CAN HAVE AGENTS

969
00:35:26,966 --> 00:35:29,535
EVEN CREATING RESEARCH PLANS FOR

970
00:35:29,535 --> 00:35:32,972
YOU USING THE MAIN SPECIFIC

971
00:35:32,972 --> 00:35:33,205
MODELS.

972
00:35:33,205 --> 00:35:35,675
ANDUREENTLY WE HAVE OVER 2000

973
00:35:35,675 --> 00:35:37,577
AGENTS IN THE COMPANY THAT WE'VE

974
00:35:37,577 --> 00:35:39,545
CREATED, SO WE PRETTY MUCH

975
00:35:39,545 --> 00:35:43,449
REPLACED A LOT OF KIND OF LOW

976
00:35:43,449 --> 00:35:44,317
LEVEL DATA LIMITATION AND EVEN

977
00:35:44,317 --> 00:35:46,819
SOME OF THE CODING TASKS WITH

978
00:35:46,819 --> 00:35:48,688
AGENTS, WE HAVE SOFTWARE IN THE

979
00:35:48,688 --> 00:35:51,223
MARKET, CURRENTLY 10 OUT OF THE

980
00:35:51,223 --> 00:35:53,392
24 COMPANIES ARE USING OUR

981
00:35:53,392 --> 00:35:54,594
TOOLS, AND WE PROTECT OURSELVES

982
00:35:54,594 --> 00:35:56,629
PRACTICES OUR DATA HOWEVER, WE

983
00:35:56,629 --> 00:35:58,965
REALLY LIKE TO RECEIVE FEEDBACK

984
00:35:58,965 --> 00:36:03,569
ON THE MODELS AND FEEDBACK IT

985
00:36:03,569 --> 00:36:08,908
WAS INVALUABLE AND IT GOES BACK

986
00:36:08,908 --> 00:36:12,111
TO OUR QC PROGRAMS AND WE DO

987
00:36:12,111 --> 00:36:14,280
EXPERT LEARNING WITH EXPERT

988
00:36:14,280 --> 00:36:16,549
FEEDBACK, SO WE REGULARLY

989
00:36:16,549 --> 00:36:17,984
RETRAIN THE MODELS AND MADE THEM

990
00:36:17,984 --> 00:36:20,519
AVAILABLE FOR THE COMMUNITY.

991
00:36:20,519 --> 00:36:22,655
AND WE USE THOSE SAME MODELS TO

992
00:36:22,655 --> 00:36:24,624
PUT THE DRUGS IN CLINICAL TRIALS

993
00:36:24,624 --> 00:36:27,927
ITSELF, WE WILL WORK WITH SMALL

994
00:36:27,927 --> 00:36:32,798
MOLL KIEWMS BIOLOGICS AND DO

995
00:36:32,798 --> 00:36:33,199
TARGET DISCOVERY.

996
00:36:33,199 --> 00:36:35,334
WE ALSO DO A LOT OF WORK TBH

997
00:36:35,334 --> 00:36:38,304
AGING WORK SO PLEASE FOLLOW US

998
00:36:38,304 --> 00:36:39,639
ON GOOGLE SCHOLAR, WE SIEWCIALLY

999
00:36:39,639 --> 00:36:44,577
PUT OUT MAYBE 1 OR 2 INTERESTING

1000
00:36:44,577 --> 00:36:46,512
PAPER IS IN AGING RESEARCH EVERY

1001
00:36:46,512 --> 00:36:49,181
QUARTER AND WE START ALSO

1002
00:36:49,181 --> 00:36:51,917
UTILIZING GENERATIVE ABILITY

1003
00:36:51,917 --> 00:36:55,388
ANDY FOR AGRICULTURAL TASKS, SO

1004
00:36:55,388 --> 00:36:58,224
PESTICIDES, HERBICIDES AND ALSO

1005
00:36:58,224 --> 00:37:01,894
FOR US TO CAPTURE OUR MATERIAL

1006
00:37:01,894 --> 00:37:02,495
DESIGN TASKS.

1007
00:37:02,495 --> 00:37:09,602
SO LOOKING FOR MOLECULES AND

1008
00:37:09,602 --> 00:37:11,237
PROVIDING CO2 EFFICIENTLY, WE'VE

1009
00:37:11,237 --> 00:37:12,872
RECENTLY DRIVEN REVENUE FROM

1010
00:37:12,872 --> 00:37:14,373
THAT AS WELL, IT'S NONAPOPTOTIC

1011
00:37:14,373 --> 00:37:15,541
THE JUST HEALTHY

1012
00:37:15,541 --> 00:37:17,410
[INDISCERNIBLE], IT'S A REALLY

1013
00:37:17,410 --> 00:37:18,177
INTERESTING BUSINESS.

1014
00:37:18,177 --> 00:37:20,413
AND NOW I'M GOING TO SHOW YOU

1015
00:37:20,413 --> 00:37:22,348
THE MOST IMPORTANT SLIDE DURING

1016
00:37:22,348 --> 00:37:26,852
THIS HOPEFULLY THIS TALK BY VERY

1017
00:37:26,852 --> 00:37:30,022
OFTEN PEOPLE TRY TO PROJECT WHAT

1018
00:37:30,022 --> 00:37:32,491
DRUG DISCOVERY IS GOING TO BE

1019
00:37:32,491 --> 00:37:39,331
WITH AI, IN THE FUTURE, IN TERMS

1020
00:37:39,331 --> 00:37:41,300
OF SPEED, COST, AND PROBABILITY

1021
00:37:41,300 --> 00:37:44,704
OF SUCCESS, HOWEVER, THERE ARE

1022
00:37:44,704 --> 00:37:45,304
VERY FEW BENCHMARKS OUT THERE

1023
00:37:45,304 --> 00:37:47,740
THAT CAN BE USED TO SEE, OKAY,

1024
00:37:47,740 --> 00:37:50,409
ARE YOU OUT  PERFORMING OR NOT.

1025
00:37:50,409 --> 00:37:53,913
SO HERE ARE SOME INTERNAL

1026
00:37:53,913 --> 00:37:55,014
BENCHMARKS FROM INSILICO, MANY

1027
00:37:55,014 --> 00:37:56,615
OF THOSE ARE DATE STAMPED WITH

1028
00:37:56,615 --> 00:38:00,052
PRESS RELEASES OR ACADEMIC

1029
00:38:00,052 --> 00:38:02,421
PAPERS, SO IN 2019, WE RAISED

1030
00:38:02,421 --> 00:38:04,523
OUR FIRST BIG ROUND TO DO OUR

1031
00:38:04,523 --> 00:38:06,325
OWN DRUG DISCOVERY AND ALSO PUT

1032
00:38:06,325 --> 00:38:08,794
THE SOFTWAREOT MARKET, SO THEN

1033
00:38:08,794 --> 00:38:11,297
WE NOMINATED 20 PRECLINICAL

1034
00:38:11,297 --> 00:38:14,467
CANDIDATES, 9 REACHED IND

1035
00:38:14,467 --> 00:38:16,869
CLEARANCE, 2 ARE IN PHASE 2, 1

1036
00:38:16,869 --> 00:38:18,404
JUST COMPLOATED PHASE 2 A, WITH

1037
00:38:18,404 --> 00:38:19,939
SAFETY AND DOSE DEPENDENT

1038
00:38:19,939 --> 00:38:22,174
EFFICACY AND ACROSS ALL THESE,

1039
00:38:22,174 --> 00:38:24,677
NOW OUR AVERAGE TIME TO

1040
00:38:24,677 --> 00:38:27,079
PRECLINICAL CANDIDATE IS 13

1041
00:38:27,079 --> 00:38:28,814
MONTHS.

1042
00:38:28,814 --> 00:38:32,585
SHORTEST TIME TO PRECLINICAL

1043
00:38:32,585 --> 00:38:34,720
CANDIDATE WAS 9 NONTHS.

1044
00:38:34,720 --> 00:38:37,656
THAT WAS ALSO PARTNERED SO IT

1045
00:38:37,656 --> 00:38:39,225
ALREADY PROGRESSED INTO PHASE 1

1046
00:38:39,225 --> 00:38:40,459
FROM 0 TO PRECLINICAL CANDIDATE

1047
00:38:40,459 --> 00:38:44,263
WITH A NOVEL TARGET, IT WAS A

1048
00:38:44,263 --> 00:38:45,998
NOVEL TARGET FOR CANCER, WE

1049
00:38:45,998 --> 00:38:48,100
MANAGED TO ACHIEVE A PRECLINICAL

1050
00:38:48,100 --> 00:38:49,168
CANDIDATE IN 9 MONTHS.

1051
00:38:49,168 --> 00:38:52,838
FOR US THE PC C PACKAGE INCLUDES

1052
00:38:52,838 --> 00:38:55,608
2AN MALEFFICACY MODELS AND 28

1053
00:38:55,608 --> 00:38:57,943
DAY NONGLP TALKS.

1054
00:38:57,943 --> 00:38:59,945
SO, IT'S PRETTY COMPREHENSIVE

1055
00:38:59,945 --> 00:39:06,519
AND AT A MINIMUM 2 MODELS.

1056
00:39:06,519 --> 00:39:08,788
IN OUR ACTINIC PAPER WE

1057
00:39:08,788 --> 00:39:11,991
DEMONSTRATED WE PUT OUT ATM

1058
00:39:11,991 --> 00:39:15,094
EXPERIMENTS IN 1 PAPER

1059
00:39:15,094 --> 00:39:17,396
PRECLINICAL AND THAT TOOK US 18

1060
00:39:17,396 --> 00:39:17,963
MONTHS.

1061
00:39:17,963 --> 00:39:23,769
THAT'S OUR LONGEST PROGRAM FOR

1062
00:39:23,769 --> 00:39:25,004
T-NICK, FOR 2022 NOMINATED 9 PC

1063
00:39:25,004 --> 00:39:26,772
Cs, IF WE GO AT CURRENT

1064
00:39:26,772 --> 00:39:28,607
CAPACITY AND HAD UNLIMITED

1065
00:39:28,607 --> 00:39:31,977
FUNDING WE COULD PROBABLY DO 12

1066
00:39:31,977 --> 00:39:33,112
PRECLINICAL CANDIDATES AND WE

1067
00:39:33,112 --> 00:39:37,583
CURRENTLY DO NOT HAVE ANY

1068
00:39:37,583 --> 00:39:38,117
[INDISCERNIBLE] PROMISING

1069
00:39:38,117 --> 00:39:41,287
TARGETS, I HAVE MORE THAN 400

1070
00:39:41,287 --> 00:39:43,122
TARGETS AFTER THIS TALK TO GO

1071
00:39:43,122 --> 00:39:45,357
AFTER IF ANYBODY UPONS AND USUAL

1072
00:39:45,357 --> 00:39:48,360
LEE WE TRY TO BALANCE TARGET AND

1073
00:39:48,360 --> 00:39:50,129
COMPOUND NOVELTY, CONFIDENCE AND

1074
00:39:50,129 --> 00:39:50,963
COMMERCIAL ATTRACTABILITY.

1075
00:39:50,963 --> 00:39:56,202
SO THOSE ARE THE 3 CRITERIA, 1

1076
00:39:56,202 --> 00:39:56,902
SELECTING THE PROGRAM.

1077
00:39:56,902 --> 00:39:59,505
BECAUSE AT THE END OF THE DAY,

1078
00:39:59,505 --> 00:40:03,209
YOU ALSO WANT TO INSURE THE YOB

1079
00:40:03,209 --> 00:40:03,776
CORPS IS MARKETABLE.

1080
00:40:03,776 --> 00:40:05,244
OUR YOB CORPS IS NONAPOPTOTIC

1081
00:40:05,244 --> 00:40:07,246
THE TO TAKE EVERY PROGRAM INTO

1082
00:40:07,246 --> 00:40:08,948
THE CLINIC OURSELVES, WE TRY TO

1083
00:40:08,948 --> 00:40:12,051
DEVELOP THE DRUG TO THE POINT

1084
00:40:12,051 --> 00:40:14,653
WHERE IT'S DERISKED AND THEN BIG

1085
00:40:14,653 --> 00:40:17,690
PHARMACEUTICAL COMPANIES TAKE IT

1086
00:40:17,690 --> 00:40:21,760
OVER AND DO LATER STAGE CLINICAL

1087
00:40:21,760 --> 00:40:21,994
TRIALS.

1088
00:40:21,994 --> 00:40:24,063
SO HOW DO WE LOOK AT NOVELTY, WE

1089
00:40:24,063 --> 00:40:26,565
USUALLY LOOK AT WHETHER THE

1090
00:40:26,565 --> 00:40:28,467
TARGET HAS ALREADY BEEN IN THE

1091
00:40:28,467 --> 00:40:30,002
CLINICAL TRIALS, IS THERE

1092
00:40:30,002 --> 00:40:33,038
CLINICAL POC, OTHER SELECTIVE

1093
00:40:33,038 --> 00:40:33,472
COMPOUNDS, HOW MANY

1094
00:40:33,472 --> 00:40:35,074
PUBLICATIONS, SO FOR EXAMPLE,

1095
00:40:35,074 --> 00:40:37,209
FOR TNICK WHEN WE STARTED

1096
00:40:37,209 --> 00:40:38,477
WORKING ON THE PROGRAM, I WILL

1097
00:40:38,477 --> 00:40:42,514
TALK ABOUT IT TODAY, THERE WERE

1098
00:40:42,514 --> 00:40:44,250
ONLY I THINK 70 PAPERS JUST

1099
00:40:44,250 --> 00:40:49,088
MENTIONING THE GENE IN DIFFERENT

1100
00:40:49,088 --> 00:40:49,788
CONTEXT.

1101
00:40:49,788 --> 00:40:51,991
NO SELECTIVE COMPOUNDS, NO

1102
00:40:51,991 --> 00:40:52,992
CLINICAL PSC, NO CLINICAL

1103
00:40:52,992 --> 00:40:54,593
TRIALS, SO THAT WAS PRETTY NOVEL

1104
00:40:54,593 --> 00:40:57,329
AND WE USUALLY CONSIDER THAT TO

1105
00:40:57,329 --> 00:41:02,968
BE MAYBE 75% NOVELTY, AND YOU DO

1106
00:41:02,968 --> 00:41:04,270
SCORING WITH EVERYTHING THAT HAS

1107
00:41:04,270 --> 00:41:06,639
BEEN DEVELOPED AGAINST THE

1108
00:41:06,639 --> 00:41:07,673
TARGET, DORPHED AGAINST THIS

1109
00:41:07,673 --> 00:41:09,308
TARGET AND THEN WE HAVE TO LOOK

1110
00:41:09,308 --> 00:41:11,076
AT, YOU KNOW IS THAT GOING TO BE

1111
00:41:11,076 --> 00:41:14,013
FIRST IN CLASS, BEST IN CLASS,

1112
00:41:14,013 --> 00:41:15,347
PATTERN DIFFERENTIATED, WHAT IS

1113
00:41:15,347 --> 00:41:18,450
THE PATENT LIFE ON, SO FOR OUR

1114
00:41:18,450 --> 00:41:22,187
T-NICK PROGRAM IT WAS A

1115
00:41:22,187 --> 00:41:24,189
SCAFFOLD, LESS THAN 60%

1116
00:41:24,189 --> 00:41:27,726
[INDISCERNIBLE] SIMILARITY WITH

1117
00:41:27,726 --> 00:41:28,127
EVERYTHING KNOWN.

1118
00:41:28,127 --> 00:41:34,099
THIS IS THIS SLIDE THAT SHOWS

1119
00:41:34,099 --> 00:41:36,535
OUR INSILICO PRICE SO WE HAVE

1120
00:41:36,535 --> 00:41:37,870
SOFTWARE FOR SCIENCE, YOU CAN

1121
00:41:37,870 --> 00:41:40,239
BUY MOST OF THOSE COMMERCIALLY,

1122
00:41:40,239 --> 00:41:46,078
WE HAVE MORE VERTICALS, 42

1123
00:41:46,078 --> 00:41:47,212
BIOLOGY, 42 MEDICINE, 42 SCIENCE

1124
00:41:47,212 --> 00:41:49,315
AND MEDICINE, SO IT ALLOWS TO

1125
00:41:49,315 --> 00:41:52,084
YOU DISCOVER TARGETS UTILIZING

1126
00:41:52,084 --> 00:41:55,254
MORE THAN 60 DIFFERENT TARGET

1127
00:41:55,254 --> 00:41:57,890
DISCOVERY PHILOSOPHIES SO 23

1128
00:41:57,890 --> 00:41:58,590
DIFFERENT MODELS.

1129
00:41:58,590 --> 00:42:02,895
IN 1 PLACE, SO DEPENDING ON YOUR

1130
00:42:02,895 --> 00:42:04,463
KIND OF THERAPEUTIC AREA,

1131
00:42:04,463 --> 00:42:07,533
WHETHER HE OR SHE LIKES,

1132
00:42:07,533 --> 00:42:09,001
[INDISCERNIBLE] WILL BE ABLE TO

1133
00:42:09,001 --> 00:42:10,569
SATISFY THE REQUIREMENTS MOST

1134
00:42:10,569 --> 00:42:12,271
LIKELY, IF NOT, THEY HAVE THEIR

1135
00:42:12,271 --> 00:42:15,507
OWN PHILOSOPHY, I CAN TELL US

1136
00:42:15,507 --> 00:42:18,243
THIS PHILOSOPHY WILL INTEGRATE

1137
00:42:18,243 --> 00:42:23,215
INTO THE METABOLOMICS IF IT HAS

1138
00:42:23,215 --> 00:42:24,750
RECENTLY SOLID THEORETICAL BASE.

1139
00:42:24,750 --> 00:42:29,054
WE HAVE A SYSTEM CALLED GENERAL

1140
00:42:29,054 --> 00:42:33,993
ANTIBIOTICS WHICH ALLOWS TO YOU

1141
00:42:33,993 --> 00:42:38,530
GENERATE ANTIBIOTICS ANTIBODIES

1142
00:42:38,530 --> 00:42:41,500
AND WE STARTED THIS IN PILOT

1143
00:42:41,500 --> 00:42:42,601
MODE, CUSTOMERS STARTED BUYING.

1144
00:42:42,601 --> 00:42:44,403
WE DON'T HAVE OUR OWN BIOLOGICS

1145
00:42:44,403 --> 00:42:45,704
PROGRAMS YET AND SOME OF THE

1146
00:42:45,704 --> 00:42:49,808
SOFTWARE IS OPEN SOURCE, SOME OF

1147
00:42:49,808 --> 00:42:50,509
THE VERY EXCITING PROGRAMS WE

1148
00:42:50,509 --> 00:42:52,444
HAVE A LOOK IN OUR LIFE MODELS

1149
00:42:52,444 --> 00:42:55,581
SO THINK OF LIKE OPEN AI'S

1150
00:42:55,581 --> 00:42:56,382
[INDISCERNIBLE] THAT CAN

1151
00:42:56,382 --> 00:42:59,752
GENERATE 1 MINUTE TO HIGH

1152
00:42:59,752 --> 00:43:00,486
QUALITY VIDEO.

1153
00:43:00,486 --> 00:43:02,855
WE DO SIMILAR THINGS WITH

1154
00:43:02,855 --> 00:43:05,657
BIOLOGICAL DATA AND

1155
00:43:05,657 --> 00:43:08,560
[INDISCERNIBLE] 3 GPT,

1156
00:43:08,560 --> 00:43:10,863
MULTIOMIMS TRANSFORMER

1157
00:43:10,863 --> 00:43:11,630
MULTIMODAL, MULTIOMICS,

1158
00:43:11,630 --> 00:43:13,098
TRANSFORM IS CURRENTLY OPEN

1159
00:43:13,098 --> 00:43:15,200
SOURCE AND YOU CAN PLAY WITH IT

1160
00:43:15,200 --> 00:43:19,371
ON DISCORD, SO WE'RE BUILDING A

1161
00:43:19,371 --> 00:43:22,841
COMMUNITY TO UNDERSTAND BIOLOGY

1162
00:43:22,841 --> 00:43:23,142
IN TIME.

1163
00:43:23,142 --> 00:43:24,043
[INDISCERNIBLE] ALLOWS TO YOU

1164
00:43:24,043 --> 00:43:25,711
GENERATE TOOLS WITH PROPERTIES,

1165
00:43:25,711 --> 00:43:28,280
IT HAS OVER 40 GENERATIVE MODELS

1166
00:43:28,280 --> 00:43:31,350
AND OVER 700 PREDICTIVE MODELS

1167
00:43:31,350 --> 00:43:32,918
FOR REINFORCEMENT LEARNING, VERY

1168
00:43:32,918 --> 00:43:34,153
STABLE SYSTEM, CHEMISTRY ALLOWS

1169
00:43:34,153 --> 00:43:37,256
YOU TO DO DFT, CHEAPLY,

1170
00:43:37,256 --> 00:43:39,525
ACCURATELY, IT COMES FREE WITH

1171
00:43:39,525 --> 00:43:41,093
CHEMISTRY 42, ON MEDICINE, 42,A

1172
00:43:41,093 --> 00:43:43,695
LOWS TO YOU PREDICT THE OUTCOMES

1173
00:43:43,695 --> 00:43:44,363
OF CLINICAL TRIALS, HAVE A LOOK

1174
00:43:44,363 --> 00:43:47,699
AT SOME OF OUR PUBLISHED PAPERS

1175
00:43:47,699 --> 00:43:52,171
WHERE WE DID RETROSPECTIVE,

1176
00:43:52,171 --> 00:43:54,940
PERSPECTIVE AND QUASI

1177
00:43:54,940 --> 00:43:56,108
PERSPECTIVE, AND IT ALLOWS YOU

1178
00:43:56,108 --> 00:43:57,709
TO WORK OUTSIDE PHARMA AND WE

1179
00:43:57,709 --> 00:44:01,180
HAVE A FUN TOOL CALLED DORA, IT

1180
00:44:01,180 --> 00:44:06,852
ALLOWS YOU TO SPAWN TEAMS OF AI

1181
00:44:06,852 --> 00:44:07,820
RESEARCHERS.

1182
00:44:07,820 --> 00:44:08,921
THEY GO AND   PERFORM DIFFERENT

1183
00:44:08,921 --> 00:44:10,255
RESEARCH TASKS AND THEN BRING

1184
00:44:10,255 --> 00:44:12,558
THEIR RESEARCH DATA BACK AND IT

1185
00:44:12,558 --> 00:44:15,461
GETS COMPILED INTO A RESEARCH

1186
00:44:15,461 --> 00:44:16,261
PAPER, IT'S $17 A MONTH.

1187
00:44:16,261 --> 00:44:19,665
YOU CAN GET IT FROM OUR WEBSITE

1188
00:44:19,665 --> 00:44:22,401
AND WE USE THESE TOOLS

1189
00:44:22,401 --> 00:44:24,403
OURSELVES, WHATEVER WE PROVIDE

1190
00:44:24,403 --> 00:44:26,839
TO THE COMMUNITY TO VALIDATE

1191
00:44:26,839 --> 00:44:28,006
THEM USING OUR OWN PROGRAMS.

1192
00:44:28,006 --> 00:44:36,014
SO THE FIRST PROGRAM THAT WE

1193
00:44:36,014 --> 00:44:38,383
STARTED IN 2019 IS T-NICK, AND

1194
00:44:38,383 --> 00:44:40,652
WITH IMD SAFETY AND UNEXPECTED

1195
00:44:40,652 --> 00:44:41,887
DOSE EFFICACY WITH ACTUAL

1196
00:44:41,887 --> 00:44:45,724
IMPROVE AMS WE SEE, SO WE ARE

1197
00:44:45,724 --> 00:44:48,827
SITTING ON GOOD DATA, IT'S ALSO

1198
00:44:48,827 --> 00:44:51,797
ABOUT EFFICACY ON KIDNEY AND

1199
00:44:51,797 --> 00:44:53,465
FIBROSIS, AND ALSO OTHER

1200
00:44:53,465 --> 00:44:55,367
FIBROSIS, AND WE NOMINATE 2

1201
00:44:55,367 --> 00:44:56,101
CANDIDATES AND WE BALANCE THE

1202
00:44:56,101 --> 00:44:59,571
REST OF THE PIPELINE WITH

1203
00:44:59,571 --> 00:45:04,443
DIFFERENT ABILITIES OF OUR

1204
00:45:04,443 --> 00:45:05,611
PLATFORM.

1205
00:45:05,611 --> 00:45:07,946
BALANCING NOVELTY CONFIDENCE AND

1206
00:45:07,946 --> 00:45:08,847
COMMERCIAL TRACTABILITY, SOME OF

1207
00:45:08,847 --> 00:45:10,415
THOSE ARE PROGRAMS WE HAVE

1208
00:45:10,415 --> 00:45:13,185
ALREADY LICENSED, SO USP 1 WAS A

1209
00:45:13,185 --> 00:45:15,254
PRETTY SUCCESSFUL LICENSE SAYING

1210
00:45:15,254 --> 00:45:20,392
ON EXERCISE TO [INDISCERNIBLE],

1211
00:45:20,392 --> 00:45:21,994
$18 MILLION CASH UPFRONT TO

1212
00:45:21,994 --> 00:45:24,129
REACH I&  D.

1213
00:45:24,129 --> 00:45:27,799
SO NOT A SUPER NOVEL TARGET BUT

1214
00:45:27,799 --> 00:45:29,601
A HIGH CONFIDENCE 1 AND WE HAD

1215
00:45:29,601 --> 00:45:40,012
BEST IN CLASS CHEMISTRY.

1216
00:45:41,980 --> 00:45:43,749
-- IT HAS PROGRESSED NICELY AND

1217
00:45:43,749 --> 00:45:46,985
MANY OF THESE ARE UNDER TERM

1218
00:45:46,985 --> 00:45:49,021
SHEETS AS WE'RE CONTINUING TO

1219
00:45:49,021 --> 00:45:49,988
SUPPLY FROM PHARMACEUTICAL

1220
00:45:49,988 --> 00:45:51,957
COMPANIES WHERE HIGH QUALITY

1221
00:45:51,957 --> 00:45:52,157
DRUGS.

1222
00:45:52,157 --> 00:45:57,062
USUALLY TO BE ABLE TO LICENSE

1223
00:45:57,062 --> 00:45:59,264
OUT AN AI DISCOVERED THERAPEUTIC

1224
00:45:59,264 --> 00:46:01,166
YOU NEED TO HAVE HIGH LEVEL OF

1225
00:46:01,166 --> 00:46:03,535
QUALITY OF THE DATA RULE.

1226
00:46:03,535 --> 00:46:07,039
SO THAT'S WHY WE WANT TO

1227
00:46:07,039 --> 00:46:08,173
MULTIPLE PHARMACEUTICAL

1228
00:46:08,173 --> 00:46:10,242
COMPANIES AND FROM BIOFARM O

1229
00:46:10,242 --> 00:46:12,778
SUITICAL COMPANIES TO CREATE

1230
00:46:12,778 --> 00:46:15,380
THOSE THAT EXCEED THE QUALITY

1231
00:46:15,380 --> 00:46:16,548
STANDARDS OF LARGE

1232
00:46:16,548 --> 00:46:17,549
PHARMACEUTICAL COMPANIES AND

1233
00:46:17,549 --> 00:46:19,351
INSURE THAT THE AVERAGE CHECK

1234
00:46:19,351 --> 00:46:20,619
MARK IS CHECKED.

1235
00:46:20,619 --> 00:46:22,821
USUALLY THE LEVEL OF QUALITY

1236
00:46:22,821 --> 00:46:25,624
THAT YOU NEED TO ACHIEVE FOR AN

1237
00:46:25,624 --> 00:46:27,326
EXTERNAL LICENSING DEAL IS

1238
00:46:27,326 --> 00:46:31,830
ACTUALLY HIGHER THAN BIG

1239
00:46:31,830 --> 00:46:32,431
PHARMACEUTICAL COMPANIES HAVE

1240
00:46:32,431 --> 00:46:33,832
INTERNALLY SO IT'S VERY, VERY

1241
00:46:33,832 --> 00:46:35,000
IMPORTANT TO COMMIT YOURSELF TO

1242
00:46:35,000 --> 00:46:38,437
ULTRA HIGH LEVEL OF QUALITY AND

1243
00:46:38,437 --> 00:46:43,475
ALSO MANY OF OUR TOOLS HAVE

1244
00:46:43,475 --> 00:46:45,210
RESEARCH PAPERS EXPLAINING OUR

1245
00:46:45,210 --> 00:46:46,178
APPLICATION KNOWLEDGE,

1246
00:46:46,178 --> 00:46:48,747
EXPLAINING HOW THEY WORK, SO

1247
00:46:48,747 --> 00:46:50,749
METABOLOMICS YOU CAN READ THE

1248
00:46:50,749 --> 00:46:52,084
APPLICATION AND NOTE HOW IT

1249
00:46:52,084 --> 00:46:54,353
WORKS AND THEN EXPLAIN CASE

1250
00:46:54,353 --> 00:46:57,589
STUDIES AND HOW MODELS WORK FOR

1251
00:46:57,589 --> 00:46:58,957
CHEMISTRY FOR THE TOOL, ALSO AND

1252
00:46:58,957 --> 00:47:01,260
1 OF MY FAVORITE JOURNALS, WE

1253
00:47:01,260 --> 00:47:03,629
PUT OUT THE APPLICATION OUT

1254
00:47:03,629 --> 00:47:06,398
EXPLAINING HOW IT WORKS.

1255
00:47:06,398 --> 00:47:08,066
NOW SEVERAL SMALLER CUTCHES

1256
00:47:08,066 --> 00:47:09,434
ACTUALLY ALSO PICKED UP THIS

1257
00:47:09,434 --> 00:47:12,137
MODEL AND STARTED DEVELOPING AND

1258
00:47:12,137 --> 00:47:15,641
PROVIDING TOOLS IN A VERY

1259
00:47:15,641 --> 00:47:19,177
SIMILAR MODEL SO DEEP

1260
00:47:19,177 --> 00:47:19,645
FLUORESCENT LEARNING.

1261
00:47:19,645 --> 00:47:20,512
I HIGHLY RECOMMEND LOOKING AT

1262
00:47:20,512 --> 00:47:23,448
THIS PAPER AND USING THIS

1263
00:47:23,448 --> 00:47:25,584
APPROACH BECAUSE IT'S PRETTY

1264
00:47:25,584 --> 00:47:25,851
VALIDATED.

1265
00:47:25,851 --> 00:47:29,187
DORA IS A SUPER COOL TOOL.

1266
00:47:29,187 --> 00:47:31,990
YOU CAN GET IT FOR $17 A MONTH

1267
00:47:31,990 --> 00:47:36,628
OR A 50% DISCOUNT. 

1268
00:47:36,628 --> 00:47:41,800
PREVIOUS SPEAKER MARIS SA FROM

1269
00:47:41,800 --> 00:47:42,701
RECURSION, WAS SPEAKING FROM

1270
00:47:42,701 --> 00:47:44,903
BDECKER MI, AND WHY DON'T WE

1271
00:47:44,903 --> 00:47:46,038
WRITE A RESEARCH PAPER ABOUT

1272
00:47:46,038 --> 00:47:49,675
THAT WHILE I'M RESENTING SO

1273
00:47:49,675 --> 00:47:51,843
LET'S DO RBM 39, AND I THINK

1274
00:47:51,843 --> 00:47:55,314
THEY WENT FOR LYMPHOMA.

1275
00:47:55,314 --> 00:48:00,218
YOU CAN ADD ADDITIONAL PAPERS

1276
00:48:00,218 --> 00:48:02,854
HERE YOURSELF, SO THE OBJECTIVE

1277
00:48:02,854 --> 00:48:03,722
FOR SCIENTISTS IS USUALLY TO

1278
00:48:03,722 --> 00:48:05,524
WRITE A PAPER, SO WE ALLOW YOU

1279
00:48:05,524 --> 00:48:08,960
TO WRITE A PAPER HERE SO IT'S

1280
00:48:08,960 --> 00:48:11,229
ALWAYS IN THE FORM OF THE PAPER,

1281
00:48:11,229 --> 00:48:13,231
YOU CAN ADD YOUR OWN DATA HERE

1282
00:48:13,231 --> 00:48:15,500
BUT LET'S SAY WE GO AUTOPILOT SO

1283
00:48:15,500 --> 00:48:17,336
WE GIVE IT A SECOND AND IT WILL

1284
00:48:17,336 --> 00:48:19,004
PROVIDE YOU WITH A RESEARCH

1285
00:48:19,004 --> 00:48:22,741
PLAN, SO NOW, THE SYSTEM IS

1286
00:48:22,741 --> 00:48:23,809
PROVIDING RESEARCH PLAN BASED ON

1287
00:48:23,809 --> 00:48:26,611
MY REQUEST, SO IT WILL   PERFORM

1288
00:48:26,611 --> 00:48:29,481
OMICS DATA ANALYSIS, GO THROUGH

1289
00:48:29,481 --> 00:48:32,784
THOUSANDS OF DATA SETS THAT HAVE

1290
00:48:32,784 --> 00:48:34,853
BEEN INTEGRATED INTO

1291
00:48:34,853 --> 00:48:37,022
METABOLOMICS,   PERFORM NETWORK

1292
00:48:37,022 --> 00:48:38,023
ANALYSIS, HYPOTHESIS DEVELOPMENT

1293
00:48:38,023 --> 00:48:43,628
BUT LET'S SAY WE ALSO WANT TO

1294
00:48:43,628 --> 00:48:52,471
PROVIDE, ALSO   PERFORM

1295
00:48:52,471 --> 00:48:54,940
COMPREHENSIVE LITERATURE REVIEW,

1296
00:48:54,940 --> 00:48:56,775
AND SUBMIT, GENERATE CONTENT,

1297
00:48:56,775 --> 00:48:59,811
LET IT RUN, SO NOW WE CAN SEE IT

1298
00:48:59,811 --> 00:49:02,481
SPAWNED AN ARMY OF AGENTS, THE

1299
00:49:02,481 --> 00:49:08,720
MASTER AGENTS OF THE PRINCIPAL

1300
00:49:08,720 --> 00:49:10,088
INVESTIGATOR, TRANSCRIPTOMICS,

1301
00:49:10,088 --> 00:49:11,390
KNOWLEDGE GRAPHS, SCIENCE

1302
00:49:11,390 --> 00:49:12,157
RESEARCHER, WEB REFERENCE FINDER

1303
00:49:12,157 --> 00:49:13,825
AND BY THE TIME I FINISH TALKING

1304
00:49:13,825 --> 00:49:18,263
IT WILL HAVE A PAPER WITH

1305
00:49:18,263 --> 00:49:18,730
ORIGINAL BIBLIOGRAPHY.

1306
00:49:18,730 --> 00:49:20,799
SO LET'S GO BACK TO THE

1307
00:49:20,799 --> 00:49:23,135
PRESENTATION WHILE WE'RE

1308
00:49:23,135 --> 00:49:23,368
WRITING.

1309
00:49:23,368 --> 00:49:25,070
SO I JUST WANTED TO COVER A

1310
00:49:25,070 --> 00:49:26,772
LITTLE BIT OF HISTORY OF

1311
00:49:26,772 --> 00:49:28,740
GENERATIVE AI AND DRUG

1312
00:49:28,740 --> 00:49:29,007
DISCOVERY.

1313
00:49:29,007 --> 00:49:30,175
SO IN 2016 THERE WERE 3

1314
00:49:30,175 --> 00:49:31,410
COMPETING GROUPS AS I REMEMBER

1315
00:49:31,410 --> 00:49:34,846
IT IN THIS FIELD, SO 1 WAS OURS,

1316
00:49:34,846 --> 00:49:37,015
BUT OF COURSE, VERY FAMOUS GROUP

1317
00:49:37,015 --> 00:49:40,385
IS [INDISCERNIBLE]'S LAB AND

1318
00:49:40,385 --> 00:49:44,623
[INDISCERNIBLE] LAB THE

1319
00:49:44,623 --> 00:49:45,657
ASTRAZENECA, AND PUBLISHED HIS

1320
00:49:45,657 --> 00:49:48,727
FIRST VERY, VERY FAMOUS PAPER

1321
00:49:48,727 --> 00:49:50,796
WITH CANNED WHAT CHEMICAL DESIGN

1322
00:49:50,796 --> 00:49:52,564
USING DATA DRIVEN CONTINUOUS

1323
00:49:52,564 --> 00:49:53,598
PRESENTATION OF MOLECULES, IT

1324
00:49:53,598 --> 00:49:55,967
WENT ON ARCHIVE IN 2016.

1325
00:49:55,967 --> 00:49:59,070
IN OCTOBER I HAD TO

1326
00:49:59,070 --> 00:50:01,273
[INDISCERNIBLE] BECAUSE MINE WAS

1327
00:50:01,273 --> 00:50:02,974
IN PEER REVIEW IN THE ACADEMIC

1328
00:50:02,974 --> 00:50:07,913
JOURNAL AT THE TIME AND SIMILAR

1329
00:50:07,913 --> 00:50:12,050
IDEA, AND NOW PUBLISHED THE

1330
00:50:12,050 --> 00:50:13,585
[INDISCERNIBLE] CENTRAL SCIENCE

1331
00:50:13,585 --> 00:50:16,087
FROM CHEN ALSO A VERY EARLY

1332
00:50:16,087 --> 00:50:19,591
WORKS IN 2017 MOLECULAR NOVEL

1333
00:50:19,591 --> 00:50:20,559
DESIGN THROUGH REINFORCE AM

1334
00:50:20,559 --> 00:50:22,861
LEARNING AND MANY OTHERS, WE

1335
00:50:22,861 --> 00:50:24,563
TOGETHER PUBLISHED MOSES, SOME

1336
00:50:24,563 --> 00:50:29,668
OF MY PAPERS, SO OUR FIRST 1 WAS

1337
00:50:29,668 --> 00:50:32,471
FROM 2016 BUT FOR SOME REASON IT

1338
00:50:32,471 --> 00:50:36,074
CHOSE 2017, WE USED

1339
00:50:36,074 --> 00:50:36,908
[INDISCERNIBLE] FOR SMALL

1340
00:50:36,908 --> 00:50:38,777
MOLECULE DRUG DESIGN BUT THEN

1341
00:50:38,777 --> 00:50:40,512
DOING EXPERIMENTS THERE AND THEN

1342
00:50:40,512 --> 00:50:41,680
REALLY BIG EXPERIMENTAL

1343
00:50:41,680 --> 00:50:44,316
VALIDATION STUDY FOR US FOR

1344
00:50:44,316 --> 00:50:46,751
GENERATIVE [INDISCERNIBLE],

1345
00:50:46,751 --> 00:50:48,086
ENFORCEMENT LEARNING, PLUS 2019,

1346
00:50:48,086 --> 00:50:50,088
HERE ARE THE TIMELINE OF SOME OF

1347
00:50:50,088 --> 00:50:52,724
THOSE EXPERIMENTS SO GREEN IS

1348
00:50:52,724 --> 00:50:54,860
INSILICO PAPERS, OUR FIRST

1349
00:50:54,860 --> 00:50:56,161
EXPERIMENTAL STUDY WAS

1350
00:50:56,161 --> 00:50:57,696
CONDITIONAL [INDISCERNIBLE],

1351
00:50:57,696 --> 00:51:00,131
WHERE WE DESIGNED OUR FIRST

1352
00:51:00,131 --> 00:51:03,802
SELECTIVE [INDISCERNIBLE] AND

1353
00:51:03,802 --> 00:51:04,603
PUBLISHED AND TESTED IT.

1354
00:51:04,603 --> 00:51:08,907
WE DID IT IN 2017, BUT MANAGED

1355
00:51:08,907 --> 00:51:13,011
TO PUBLISH IN 2018, AND IN 2019

1356
00:51:13,011 --> 00:51:14,813
IT WAS GENERAL.

1357
00:51:14,813 --> 00:51:17,682
THE GREATEST BOTTLENECK WE

1358
00:51:17,682 --> 00:51:20,085
QUICKLY REALIZED IS IT'S

1359
00:51:20,085 --> 00:51:21,720
EXPERIMENTAL VALIDATION, IT

1360
00:51:21,720 --> 00:51:24,956
TAKES YOU SOMETIMES 5 WEEKS TO

1361
00:51:24,956 --> 00:51:27,659
CREATE AND TRAIN NEW MODEL, NEW

1362
00:51:27,659 --> 00:51:28,527
GENERATIVE MODEL REGARDLESS OF

1363
00:51:28,527 --> 00:51:33,098
HOW COMPLEX IT IS, USUALLY IT

1364
00:51:33,098 --> 00:51:33,765
DOESN'T TAKE LONGER.

1365
00:51:33,765 --> 00:51:35,567
BUT IT CAN TAKE YOU SEVERAL

1366
00:51:35,567 --> 00:51:38,803
YEARS TO INSURE THE THAT ALL

1367
00:51:38,803 --> 00:51:39,504
VALIDATION EXPERIMENTS ARE

1368
00:51:39,504 --> 00:51:41,039
CONCERNED EMPLOY BY THE WAY, A

1369
00:51:41,039 --> 00:51:43,975
GRAY PAPER THAT THE PR ININATE

1370
00:51:43,975 --> 00:51:45,310
AND YOU ARE INTELIENCE THIS

1371
00:51:45,310 --> 00:51:47,512
APRIL HOW THE OF TOM

1372
00:51:47,512 --> 00:51:48,980
[INDISCERNIBLE] LAB IN

1373
00:51:48,980 --> 00:51:50,515
CAMBRIDGE, I HAVEN'T SEEN IT

1374
00:51:50,515 --> 00:51:52,918
PUBLISHED BUT WHEN I READ IT, I

1375
00:51:52,918 --> 00:51:54,586
WAS BLOWN AWAY HOW MUCH WORK

1376
00:51:54,586 --> 00:51:57,255
THEY   PERFORMED, SO THEY

1377
00:51:57,255 --> 00:51:58,723
ACTUALLY LOOKED AT ONLY

1378
00:51:58,723 --> 00:51:59,624
VALIDATION EXPERIMENTS THAT WERE

1379
00:51:59,624 --> 00:52:02,127
-- THAT APPEARED IN LITERATURE,

1380
00:52:02,127 --> 00:52:04,162
OR EVERY MODEL PUBLISHED FOR

1381
00:52:04,162 --> 00:52:06,464
GENERATIVE CHEMIST RADIOY WITH

1382
00:52:06,464 --> 00:52:10,101
INPUT, OUTPUT, DESIGN TASK,

1383
00:52:10,101 --> 00:52:13,271
TARGET, KIT RATE, RESULT AND

1384
00:52:13,271 --> 00:52:15,974
YEAR PUBLISHED FOR DISTRIBUTION

1385
00:52:15,974 --> 00:52:17,275
AND ALSO FOR [INDISCERNIBLE]

1386
00:52:17,275 --> 00:52:18,243
LEARNING AND THESE ARE SOME OF

1387
00:52:18,243 --> 00:52:19,444
OUR EXPERIMENTS SO I WAS HAPPY

1388
00:52:19,444 --> 00:52:22,614
TO SEE THAT OUR WORK WAS VERY

1389
00:52:22,614 --> 00:52:24,182
WELL REPRESENTED SO WE DID

1390
00:52:24,182 --> 00:52:25,717
PUBLISH QUITE A BIT WITH

1391
00:52:25,717 --> 00:52:26,017
EXPERIMENTS.

1392
00:52:26,017 --> 00:52:28,219
AND HERE ARE SOME CASE STUDIES I

1393
00:52:28,219 --> 00:52:34,759
WOULD LIKE TO PRESENT, SO NEW

1394
00:52:34,759 --> 00:52:35,493
COMPOUNDS INFUSING AI, WORK

1395
00:52:35,493 --> 00:52:38,363
EXPERIMENTALLY, SO THAT WAS OUR

1396
00:52:38,363 --> 00:52:39,130
2019 PAPER.

1397
00:52:39,130 --> 00:52:42,133
WE DID THE EXPERIMENT 2018 FOR A

1398
00:52:42,133 --> 00:52:44,536
YEAR TO PUBLISH, ALMOST GOT

1399
00:52:44,536 --> 00:52:46,004
REJECTED, AND HOW PRETTY HIGHLY

1400
00:52:46,004 --> 00:52:49,240
SIGHTED SO HERE, WE WERE

1401
00:52:49,240 --> 00:52:50,775
CHALLENGED BY OUR PARTNER FOR

1402
00:52:50,775 --> 00:52:52,477
ANY TARGET OUT OF THE

1403
00:52:52,477 --> 00:52:53,845
[INDISCERNIBLE].

1404
00:52:53,845 --> 00:52:56,681
THEY GIVE US THE KINASE, 21 DAYS

1405
00:52:56,681 --> 00:52:59,217
TO GENERATE 6 COMPOUNDS AND THEN

1406
00:52:59,217 --> 00:53:01,820
WE SYNTHESIZE AND TESTED.

1407
00:53:01,820 --> 00:53:03,989
[INDISCERNIBLE] WAS THE LAST

1408
00:53:03,989 --> 00:53:06,324
AUTHOR, I WAS THE FIRST AUTHOR,

1409
00:53:06,324 --> 00:53:08,059
SO AT THAT TIME WE BECAME

1410
00:53:08,059 --> 00:53:09,527
FRIENDS DURING THIS EXPERIMENT.

1411
00:53:09,527 --> 00:53:11,463
PRETTY FAMOUS PAPER AND IN 46

1412
00:53:11,463 --> 00:53:15,734
DAYS WE MANAGED TO DOSE THE

1413
00:53:15,734 --> 00:53:16,034
MOUSE.

1414
00:53:16,034 --> 00:53:19,004
AND OUT OF THOSE 6 COMPOUNDS FOR

1415
00:53:19,004 --> 00:53:19,804
WORK, 2 WENT THROUGH

1416
00:53:19,804 --> 00:53:21,740
MICROSTUDIES OF MULTIPLE

1417
00:53:21,740 --> 00:53:23,808
ENDOCRINAL ASSAYS AND 1 WAS

1418
00:53:23,808 --> 00:53:24,542
TESTED IN MOUSE.

1419
00:53:24,542 --> 00:53:26,044
WE WERE ACTUALLY LIMITED BY THE

1420
00:53:26,044 --> 00:53:28,146
TIME AND ALSO THE ABILITY TO

1421
00:53:28,146 --> 00:53:30,849
SYNTHESIZE, SO OF COURSE SOME OF

1422
00:53:30,849 --> 00:53:33,051
THOSE MOLECULES SHARED

1423
00:53:33,051 --> 00:53:36,488
STRUCTURAL SIMILARITY WITH

1424
00:53:36,488 --> 00:53:37,656
NONCOMPOUNDS AND WE DO NOT DENY

1425
00:53:37,656 --> 00:53:40,358
THAT BUT THE FACT THAT GENERAL

1426
00:53:40,358 --> 00:53:41,393
ENFORCEMENT LEARNING MANAGED TO

1427
00:53:41,393 --> 00:53:42,927
DO THAT VERY QUICKLY WAS PRETERM

1428
00:53:42,927 --> 00:53:44,229
BIRTH IMPEDIMENTS PRESSIVE, THEN

1429
00:53:44,229 --> 00:53:48,233
IT WAS CRITICIZED BY SOME OF THE

1430
00:53:48,233 --> 00:53:52,737
COMPETING GROUPS AND WE

1431
00:53:52,737 --> 00:53:54,706
RESPONDED AND EXPLAINED THAT

1432
00:53:54,706 --> 00:53:55,206
IT'S SCALABLE.

1433
00:53:55,206 --> 00:53:57,308
BUT AT THAT TIME AGAIN, PEOPLE

1434
00:53:57,308 --> 00:54:00,779
WERE DEBATING THE POTENTIAL OF

1435
00:54:00,779 --> 00:54:02,380
GENERATIVE CHEMISTRY IN THE

1436
00:54:02,380 --> 00:54:05,550
INDUSTRY, AND WE HAD TO RESPOND.

1437
00:54:05,550 --> 00:54:07,585
BUT THEN ONCE WE MANAGED TO

1438
00:54:07,585 --> 00:54:09,320
RAISE SOME MONEY WE STARTED

1439
00:54:09,320 --> 00:54:10,121
DOING REAL PROGRAMS AND YOU'VE

1440
00:54:10,121 --> 00:54:13,958
SEEN SOME OF THE RESULTS, SINCE

1441
00:54:13,958 --> 00:54:17,462
2019, SO THE PAPER WAS PUBLISHED

1442
00:54:17,462 --> 00:54:20,331
AND SOME OF THE POCs, WE'VE

1443
00:54:20,331 --> 00:54:24,569
MANAGED TO PUBLISH WITHOUT GOING

1444
00:54:24,569 --> 00:54:26,071
INTO COMMERCIAL PROGRAMS, SO 1

1445
00:54:26,071 --> 00:54:27,706
REALLY COOL EXPERIMENT WAS THIS

1446
00:54:27,706 --> 00:54:28,139
1.

1447
00:54:28,139 --> 00:54:33,678
WE DID IT IN 2021 OF THE END OF

1448
00:54:33,678 --> 00:54:34,379
THE 2021.

1449
00:54:34,379 --> 00:54:40,752
SO WE USED -- SORRY, THE

1450
00:54:40,752 --> 00:54:47,258
BEGINNING OF 2022, AND WE USED

1451
00:54:47,258 --> 00:54:51,096
PANNA OMICS TO IDENTIFY CDK20 AS

1452
00:54:51,096 --> 00:54:53,631
A MARKER FOR CARCINOMA, USED

1453
00:54:53,631 --> 00:54:55,200
ALPHA FOLD PROTEIN STRUCTURE AND

1454
00:54:55,200 --> 00:54:58,903
CHEMISTRY 42 TO VERY RAPIDLY DO

1455
00:54:58,903 --> 00:55:00,939
2 ROUNDS OF SYNTHESIS, SO IN

1456
00:55:00,939 --> 00:55:04,576
FIRST ROUND WE GOT 7.3

1457
00:55:04,576 --> 00:55:05,877
MICROMOLAR HITS SO NOT GREAT.

1458
00:55:05,877 --> 00:55:08,046
GAVE IT BACK TO CHEMISTRY 42 FOR

1459
00:55:08,046 --> 00:55:11,049
LIAISON GABBED DRUG BASED DESIGN

1460
00:55:11,049 --> 00:55:12,817
AND GOT 180 NANO MOLAR HITS AND

1461
00:55:12,817 --> 00:55:15,153
NOT TOO BAD AND TESTED INVITRO

1462
00:55:15,153 --> 00:55:18,022
AND IT WORKED SO CHEMICAL

1463
00:55:18,022 --> 00:55:19,557
SCIENCE, SO CURRENTLY 1 ACADEMIC

1464
00:55:19,557 --> 00:55:23,294
GROUP IS TAKING THIS VERSION OF

1465
00:55:23,294 --> 00:55:25,029
THIS MOLECULE FORWARD WHICH WAS

1466
00:55:25,029 --> 00:55:27,632
VERY COOL BECAUSE THEY -- THEY

1467
00:55:27,632 --> 00:55:29,868
KNOW, THEY SEE IT REALLY, REALLY

1468
00:55:29,868 --> 00:55:30,068
WELL.

1469
00:55:30,068 --> 00:55:32,103
AND WE ALSO CAN USE AI TO

1470
00:55:32,103 --> 00:55:34,038
REPURPOSE EXISTING DRUGS AS SOME

1471
00:55:34,038 --> 00:55:35,840
COMPANIES IS ACTUALLY DEVELOP

1472
00:55:35,840 --> 00:55:38,376
BUSINESS MODELS AROUND THIS

1473
00:55:38,376 --> 00:55:39,010
PHILOSOPHY.

1474
00:55:39,010 --> 00:55:40,945
SO IN THIS PARTICULAR

1475
00:55:40,945 --> 00:55:44,849
EXPERIMENT, I PRESENTED IN MARCH

1476
00:55:44,849 --> 00:55:50,088
AT [INDISCERNIBLE] CONSORTIUM IN

1477
00:55:50,088 --> 00:55:51,990
2022 AND MET BRILLIANT

1478
00:55:51,990 --> 00:55:56,060
[INDISCERNIBLE] AND JEFFREY

1479
00:55:56,060 --> 00:55:57,695
[INDISCERNIBLE] WHO RUN

1480
00:55:57,695 --> 00:55:58,630
[INDISCERNIBLE], WE COMBINED IT

1481
00:55:58,630 --> 00:56:00,064
WITH THE DATA OMICS SET

1482
00:56:00,064 --> 00:56:02,667
TAKEN--THEY WE HAVE IN THE

1483
00:56:02,667 --> 00:56:04,435
DATABASE AND GOT SEVERAL NEW

1484
00:56:04,435 --> 00:56:06,938
TARGETS AND ALSO DRUGS TO

1485
00:56:06,938 --> 00:56:07,906
REPURPOSE, TESTED THEM IN THE

1486
00:56:07,906 --> 00:56:12,343
FLY MODEL OUT OF MAYO CLINIC

1487
00:56:12,343 --> 00:56:14,312
WITH [INDISCERNIBLE] AND OUT OF

1488
00:56:14,312 --> 00:56:20,118
THE 26 PREDICTED TARGETS, 18

1489
00:56:20,118 --> 00:56:23,721
FELT STRONG OR MODERATE RESCUE

1490
00:56:23,721 --> 00:56:28,159
AND IN FLY MODELS AND ONLY 2

1491
00:56:28,159 --> 00:56:30,261
SHOWED MODERATE ENHANCEMENT

1492
00:56:30,261 --> 00:56:33,865
MOTILITY AND LESS PHENOTYPE AND

1493
00:56:33,865 --> 00:56:34,699
[INDISCERNIBLE] AT THAT TIME,

1494
00:56:34,699 --> 00:56:37,068
DEAN OF THE MEDICAL SCHOOL AT

1495
00:56:37,068 --> 00:56:38,069
[INDISCERNIBLE] UNIVERSITY NOW

1496
00:56:38,069 --> 00:56:43,074
HE IS IN SHANGHAI, VERY FAMOUS

1497
00:56:43,074 --> 00:56:44,142
AMERICAN-CHINESE SCIENTIST IN

1498
00:56:44,142 --> 00:56:46,411
NEUROSCIENCE, HE ALSO WAS THE

1499
00:56:46,411 --> 00:56:47,579
CONFOUNDER OF [INDISCERNIBLE]

1500
00:56:47,579 --> 00:56:49,681
TECHNOLOGIES SO THEY PICKED 1 OF

1501
00:56:49,681 --> 00:56:52,717
THE TARGETS THAT WE'VE

1502
00:56:52,717 --> 00:56:53,685
IDENTIFIED DURING THE

1503
00:56:53,685 --> 00:56:56,487
EXPERIMENT, TOOK ALSO 1 OF THE

1504
00:56:56,487 --> 00:56:57,589
REPURPOSING CANDIDATES AND IN

1505
00:56:57,589 --> 00:57:00,391
UNDER 1 YEAR, THEY ENROLLED 64

1506
00:57:00,391 --> 00:57:01,459
PATIENTS IN THE CLINICAL TRIAL.

1507
00:57:01,459 --> 00:57:04,395
SO THE READ OUT FOR THIS 1, IS

1508
00:57:04,395 --> 00:57:05,730
EXPERIMENTED BEFORE THE END OF

1509
00:57:05,730 --> 00:57:11,135
THIS YEAR, SO FINGERS CROSSED.

1510
00:57:11,135 --> 00:57:12,770
BUT THAT'S ALSO THE POWER OF

1511
00:57:12,770 --> 00:57:14,239
CHINA, SO SOME OF THE DISEASES

1512
00:57:14,239 --> 00:57:15,206
ARE NOT THAT RARE AND IF YOU

1513
00:57:15,206 --> 00:57:18,610
WANT TO DO A QUICK INVESTIGATOR

1514
00:57:18,610 --> 00:57:19,577
INITIATED STUDY YOU CAN ACTUALLY

1515
00:57:19,577 --> 00:57:23,047
DO THAT XI  -- AND I THINK THIS

1516
00:57:23,047 --> 00:57:25,216
IS WHERE THE POWER OF AI HAS TO

1517
00:57:25,216 --> 00:57:27,352
BE EXPLERRED AND EXPLOITED.

1518
00:57:27,352 --> 00:57:29,087
AND THEN CAN QUEE GO AFTER A

1519
00:57:29,087 --> 00:57:34,292
NOVEL TARGET WITH A NOVEL

1520
00:57:34,292 --> 00:57:34,993
MOLECULE OR REAL COMMERCIAL

1521
00:57:34,993 --> 00:57:36,227
PROGRAM AND IT'S VERY DIFFICULT

1522
00:57:36,227 --> 00:57:36,494
AND RARE.

1523
00:57:36,494 --> 00:57:39,330
AS A MATTER OF FACT IF YOU WERE

1524
00:57:39,330 --> 00:57:41,499
TO SURVEY MY TWITTER FOLLOWERS,

1525
00:57:41,499 --> 00:57:43,735
YOU KNOW IF THE CHINA EVER

1526
00:57:43,735 --> 00:57:45,103
DISCOVERED THE NOVEL MOLECULE,

1527
00:57:45,103 --> 00:57:46,571
THE ANSWER WOULD BE USUALLY NO

1528
00:57:46,571 --> 00:57:49,207
AND SOME PEOPLE DEBATE BUT THOSE

1529
00:57:49,207 --> 00:57:52,143
ARE USUALLY HOLDER TARGETS.

1530
00:57:52,143 --> 00:57:56,614
IF YOU ASK CHATGPT PREVIEW, THE

1531
00:57:56,614 --> 00:57:57,916
SAME QUESTION, THE ANSWER WOULD

1532
00:57:57,916 --> 00:58:00,218
USUALLY BE NO, AND THIS IS THE

1533
00:58:00,218 --> 00:58:02,086
PREROGATIVE OF VERY WELL

1534
00:58:02,086 --> 00:58:02,587
DEVELOPED COUNTRIES.

1535
00:58:02,587 --> 00:58:06,758
SO IF YOU LOOK AT ALSO ASK

1536
00:58:06,758 --> 00:58:10,161
CHATGPT HOW MANY COUNTRIES EVER

1537
00:58:10,161 --> 00:58:11,996
DISCOVERED A NOVEL DRUG, NOVEL

1538
00:58:11,996 --> 00:58:13,197
TARGET, NOVEL MOLECULE, THERE

1539
00:58:13,197 --> 00:58:16,567
ARE ACTUALLY NOT THAT MANY, AND

1540
00:58:16,567 --> 00:58:17,669
BEFORE TARGETED THERAPEUTICS

1541
00:58:17,669 --> 00:58:19,637
WHICH IS ACTUALLY PRETTY SCARY

1542
00:58:19,637 --> 00:58:23,474
AS WELL SO WE NEED TO GET MORE

1543
00:58:23,474 --> 00:58:25,476
LAYERS TO THE MOLECULAR CASINO

1544
00:58:25,476 --> 00:58:27,045
AND THAT'S WHY WE ALSO RELEASE

1545
00:58:27,045 --> 00:58:30,048
THE SOFTWARE AND WE WANTED TO

1546
00:58:30,048 --> 00:58:31,516
TRY AND GO AFTER THIS GOAL

1547
00:58:31,516 --> 00:58:34,385
OURSELVES FROM THE VERY

1548
00:58:34,385 --> 00:58:35,653
BEGINNING WITHOUT REPURPOSING

1549
00:58:35,653 --> 00:58:37,188
WITH NOVEL TARGET, NOVEL

1550
00:58:37,188 --> 00:58:39,424
MOLECULE, WE ACTUALLY THOUGHT WE

1551
00:58:39,424 --> 00:58:43,761
WOULD FAIL THAT'S WHY WE STARTED

1552
00:58:43,761 --> 00:58:44,595
HAVING A VIDEO DOCUMENTARY

1553
00:58:44,595 --> 00:58:46,164
AROUND THIS TARGET AND THIS

1554
00:58:46,164 --> 00:58:50,668
PROGRAM, BUT WE SUCCEEDED WHICH

1555
00:58:50,668 --> 00:58:51,703
WAS VERY UNEXPECTED.

1556
00:58:51,703 --> 00:58:53,104
THIS PROGRAM, IS DRIEBED IN THIS

1557
00:58:53,104 --> 00:58:55,940
PAPER, THAT WAS PUBLISHED IN THE

1558
00:58:55,940 --> 00:58:59,477
NATURE BIOTECH EIGHTH OF

1559
00:58:59,477 --> 00:59:00,011
MARCH 2024, PRETTY POPULAR

1560
00:59:00,011 --> 00:59:02,380
PAPER, IN THIS PAPER, WE

1561
00:59:02,380 --> 00:59:04,449
DESCRIBED HOW WE IDENTIFIED THE

1562
00:59:04,449 --> 00:59:06,150
TARGETS THAT HALLMARKS OF AGING

1563
00:59:06,150 --> 00:59:08,052
ASSESSMENT, KIND OF A LONG

1564
00:59:08,052 --> 00:59:09,520
STORY, I WON'T GO TOO DEEP INTO

1565
00:59:09,520 --> 00:59:10,355
THAT BECAUSE I'M PROBABLY A

1566
00:59:10,355 --> 00:59:14,726
LITTLE BIT OUT OF TIME,

1567
00:59:14,726 --> 00:59:15,560
SYNTHESIZED ONLY 79 MOLECULES

1568
00:59:15,560 --> 00:59:16,260
FOR THIS PROGRAM.

1569
00:59:16,260 --> 00:59:19,764
BY THE WAY OUR AVERAGE NOW PER

1570
00:59:19,764 --> 00:59:23,067
PROGRAM OVER 20 PRECLINICAL

1571
00:59:23,067 --> 00:59:24,135
CANDIDATES, WE SYNTHESIZE AROUND

1572
00:59:24,135 --> 00:59:24,769
70 MOLECULES PER PROGRAM.

1573
00:59:24,769 --> 00:59:27,939
SO IF YOU ARE TO BENCHMARK,

1574
00:59:27,939 --> 00:59:28,973
THAT'S AROUND THAT NUMBER EMPLOY

1575
00:59:28,973 --> 00:59:35,046
SO HERE WITH THE 79, ARE 055 IS

1576
00:59:35,046 --> 00:59:37,448
FIFTH MOLECULE SYNTHESIZED WITH

1577
00:59:37,448 --> 00:59:39,784
INDICATION PRIORITIZE EGG,

1578
00:59:39,784 --> 00:59:42,120
A-TINNIC FILLED THE WORK IN MANY

1579
00:59:42,120 --> 00:59:46,024
INDICATIONS BUT IPF WAS THE

1580
00:59:46,024 --> 00:59:48,526
CLASS PROBABILITY AND ALSO MOST

1581
00:59:48,526 --> 00:59:51,162
COMMERCIALLY TRACTABLE ROUTE  

1582
00:59:51,162 --> 00:59:52,163
PERFORMED MULTIPLE INVITRO

1583
00:59:52,163 --> 00:59:54,098
SPRMS, MULTIPLE IN VIVO

1584
00:59:54,098 --> 00:59:55,967
EXPERIMENTS, NOMINATED 3

1585
00:59:55,967 --> 00:59:57,402
PRECLINICAL CANDIDATES IN 1

1586
00:59:57,402 --> 00:59:59,971
PAPER, THE INDICATION EXPANSION,

1587
00:59:59,971 --> 01:00:02,874
PHASE 0 IN AUSTRALIA, AND

1588
01:00:02,874 --> 01:00:05,476
HEALTHY VOLUNTEERS, 2 PHASE 1S

1589
01:00:05,476 --> 01:00:08,546
IN NEW ZEALAND AND IN CHINA AND

1590
01:00:08,546 --> 01:00:11,849
NOW, IF YOU WANT TO TALK TO THIS

1591
01:00:11,849 --> 01:00:14,085
PAPER, IT HAS LOTS OF

1592
01:00:14,085 --> 01:00:16,087
SUPPLEMENTARY MATERIALS, WE ALSO

1593
01:00:16,087 --> 01:00:17,488
CREATED THE CONVERSATIONAL

1594
01:00:17,488 --> 01:00:19,257
INTERFACE IN JUNE 2023, IT WAS

1595
01:00:19,257 --> 01:00:20,258
ACTUALLY VERY COOL, NOWA KAYS

1596
01:00:20,258 --> 01:00:24,729
YOU CAN DO IT WITH CHAT GPT, AND

1597
01:00:24,729 --> 01:00:29,067
I'LL SKIP THIS BUT  FOR NOW WE

1598
01:00:29,067 --> 01:00:31,702
HAVE COMPLETED THE CHINA PHASE

1599
01:00:31,702 --> 01:00:35,907
2A AND OBSERVED VERY GOOD SAFETY

1600
01:00:35,907 --> 01:00:37,875
AND PROBABILITY, SO PRIMARY

1601
01:00:37,875 --> 01:00:39,911
OBJECTIVE WAS MET BUT WE ALSO

1602
01:00:39,911 --> 01:00:42,780
MET THE SECOND OBJECTIVE SO WE

1603
01:00:42,780 --> 01:00:44,549
SHOWED DOSE DEPENDENT INCREASE

1604
01:00:44,549 --> 01:00:46,584
IN VITAL FORCE CAPACITY WHICH

1605
01:00:46,584 --> 01:00:47,752
WAS VERY UNEXPECTED AND

1606
01:00:47,752 --> 01:00:49,053
CURRENTLY I'M NOT GOING TO SHOW

1607
01:00:49,053 --> 01:00:51,489
YOU DATA WHERE WORKING ON A

1608
01:00:51,489 --> 01:00:52,957
PROFILE PUBLICATION SO IF YOU

1609
01:00:52,957 --> 01:00:57,462
KNOW ANYBODY IN HIGH PROFILE

1610
01:00:57,462 --> 01:00:59,130
CLINICAL JOURNAL PING ME PLEASE,

1611
01:00:59,130 --> 01:01:00,164
WE'RE SUBMITTING BUT NOW THIS IS

1612
01:01:00,164 --> 01:01:03,601
THE PROGRAM THAT IS LIKELY TO

1613
01:01:03,601 --> 01:01:05,136
INCREASE THE CONFIDENCE IN THE

1614
01:01:05,136 --> 01:01:06,938
ENTIRE AI DRUG DISCOVERY.

1615
01:01:06,938 --> 01:01:08,039
I WILL SKIP A FEW THINGS.

1616
01:01:08,039 --> 01:01:10,975
BY THE WAY, IF YOU LIKE TO SEE

1617
01:01:10,975 --> 01:01:12,777
HOW HAN ULTRA MODERN ROBOTICS

1618
01:01:12,777 --> 01:01:16,114
LAB WORKS, PLEASE WELCOME TO OUR

1619
01:01:16,114 --> 01:01:18,316
LAB IN [INDISCERNIBLE].

1620
01:01:18,316 --> 01:01:21,619
WE HAVE CENTER CONNECTED ROOMS

1621
01:01:21,619 --> 01:01:25,189
FULLY AUTOMATED THAT WORK WITH

1622
01:01:25,189 --> 01:01:26,290
CELLS, TISSUES, ORGANOIDS THEY

1623
01:01:26,290 --> 01:01:29,560
CAN DO SAMPLE PROCESSING AND FOR

1624
01:01:29,560 --> 01:01:33,598
EVERY SAMPLE WE USUALLY GET

1625
01:01:33,598 --> 01:01:34,198
METHYLATION, TRANSCRIPTOMICS,

1626
01:01:34,198 --> 01:01:37,635
GENOME AND TYPES OF IMAGING AND

1627
01:01:37,635 --> 01:01:38,803
AI PROCESSES THIS DATA, FIX WILL

1628
01:01:38,803 --> 01:01:40,705
TARGETS AND FOR THE TARGETS THAT

1629
01:01:40,705 --> 01:01:45,776
WE MIGHT HAVE 2 COMPOUNDS, FIX 2

1630
01:01:45,776 --> 01:01:46,677
COMPOUNDS, INCUBATES THE

1631
01:01:46,677 --> 01:01:48,946
ORIGINAL SAMPLE WITH THOSE 2

1632
01:01:48,946 --> 01:01:52,750
COMPOUNDS AND   PERFORMS THE

1633
01:01:52,750 --> 01:01:54,051
ENTIRE SEQUENCE 1 MORE TIME TO

1634
01:01:54,051 --> 01:01:58,389
SEE IF THE PREDICTIONS WERE

1635
01:01:58,389 --> 01:01:58,723
CONFIRMED.

1636
01:01:58,723 --> 01:02:00,024
IF THEY WERE AND THERE ARE SOME

1637
01:02:00,024 --> 01:02:01,459
TARGETS THAT DON'T HAVE 2

1638
01:02:01,459 --> 01:02:04,662
COMPOUNDS AND SCORE HIGHER WE  

1639
01:02:04,662 --> 01:02:07,999
PERFORM CRSPR, AND THE SiRNA

1640
01:02:07,999 --> 01:02:09,767
STUDIES AND FOR THOSE THAT

1641
01:02:09,767 --> 01:02:13,538
SCORED HIGH AND WERE UNXCTED,

1642
01:02:13,538 --> 01:02:15,973
AND HAVE 2 COMPOUNDS, WE

1643
01:02:15,973 --> 01:02:16,741
ACTUALLY PROSECUTE THEM FARTHER

1644
01:02:16,741 --> 01:02:18,643
AS WELL AND NOW WE HAVE A LARGE

1645
01:02:18,643 --> 01:02:20,244
DATABASE OF TARGETS SO CURRENTLY

1646
01:02:20,244 --> 01:02:22,980
IF YOU NEED TARGETS FOR MULTIPLE

1647
01:02:22,980 --> 01:02:27,818
TYPES OF CANCER, FOR FIBROSIS,

1648
01:02:27,818 --> 01:02:28,986
MUSCLE WASTING, IBD, PLEASE

1649
01:02:28,986 --> 01:02:36,260
WELCOME TO TALK TO US.

1650
01:02:36,260 --> 01:02:37,728
>> ARE YOU FINALIZING
SLIDES.

1651
01:02:37,728 --> 01:02:38,029
>> SURE.

1652
01:02:38,029 --> 01:02:39,830
SO I JUST WANT TO MENTION THAT

1653
01:02:39,830 --> 01:02:41,866
AGING RESEARCH IS ALSO VERY,

1654
01:02:41,866 --> 01:02:44,869
VERY IMPORTANT FOR DRUG

1655
01:02:44,869 --> 01:02:46,103
DISCOVERY, MANY BIOLOGICAL

1656
01:02:46,103 --> 01:02:50,308
PROCESSES HAPPEN IN TIME AND WE

1657
01:02:50,308 --> 01:02:51,709
GO THROUGH OUR LIFE FROM CRADLE

1658
01:02:51,709 --> 01:02:52,743
TO GRAVE.

1659
01:02:52,743 --> 01:02:54,078
DEEP NEURAL NETWORKS ARE A VERY

1660
01:02:54,078 --> 01:02:56,981
POWERFUL TOOLS TO ACTUALLY

1661
01:02:56,981 --> 01:02:59,684
CAPTURE THE PATTERNS FROM THE

1662
01:02:59,684 --> 01:03:01,118
CRADLE TO THE GRAVE EVEN WITHOUT

1663
01:03:01,118 --> 01:03:04,488
DISEASE OR WITH DISEASE AND WE

1664
01:03:04,488 --> 01:03:06,824
LIKE TO DISCUSS THE CAPABILITIES

1665
01:03:06,824 --> 01:03:10,461
OF AGING RESEARCH AS A PLATFORM

1666
01:03:10,461 --> 01:03:12,663
FOR DRUG DISCOVERY AT THE

1667
01:03:12,663 --> 01:03:13,631
CONFERENCE WE ORGANIZE, NOW IT'S

1668
01:03:13,631 --> 01:03:17,702
GOING TO BE THE 12th YEAR IN

1669
01:03:17,702 --> 01:03:19,770
COPENHAGEN, IT'S A 5 DAY EVENT.

1670
01:03:19,770 --> 01:03:21,138
PLEASE WELCOME.

1671
01:03:21,138 --> 01:03:23,641
USUALLY IT GETS SOLD OUT AT

1672
01:03:23,641 --> 01:03:27,745
LEAST 2 MONTHS IN ADVANCE, SO

1673
01:03:27,745 --> 01:03:28,913
BIG PHARMA, ACADEMIA ALL COME

1674
01:03:28,913 --> 01:03:29,914
TOGETHER, PLEASE JOIN AND HAPPY

1675
01:03:29,914 --> 01:03:32,383
TO HOST YOU THERE.

1676
01:03:32,383 --> 01:03:34,719
>> YEAH, THANK YOU SO MUCH
ALEX,

1677
01:03:34,719 --> 01:03:36,821
THAT WAS QUITE IMPRESSIVE BASED

1678
01:03:36,821 --> 01:03:37,855
ON THE COMMENTS, IT SEEMS LIKE

1679
01:03:37,855 --> 01:03:43,427
THE 1  1 -- WORK YOU ARE DOING T

1680
01:03:43,427 --> 01:03:44,695
INSILICO IS AN INSPIRATION FOR

1681
01:03:44,695 --> 01:03:46,130
MANY FOLKS SO THAT LOOKS LIKE

1682
01:03:46,130 --> 01:03:51,402
THAT IS THE NEW STANDARD PROBE

1683
01:03:51,402 --> 01:03:52,103
DEVELOPMENTABLY, BUT WE WILL

1684
01:03:52,103 --> 01:03:53,571
JUST TAKE A COUPLE OF QUESTIONS.

1685
01:03:53,571 --> 01:03:58,142
ONE OF THE QUESTIONS WAS HOW

1686
01:03:58,142 --> 01:04:00,344
MUCH COMPUTING RESOURCES CAN I

1687
01:04:00,344 --> 01:04:02,113
EXPECT TO SPEND, AND SECOND

1688
01:04:02,113 --> 01:04:03,047
RELATIVE TO THAT IS YOU

1689
01:04:03,047 --> 01:04:06,751
MENTIONED THAT THERE ARE NO GOOD

1690
01:04:06,751 --> 01:04:12,123
BENCHMARKS FOR [INDISCERNIBLE]

1691
01:04:12,123 --> 01:04:14,592
MODELS, CFNS, AND CR, ARE WIDELY

1692
01:04:14,592 --> 01:04:16,827
USED FOR THEY APPLICABLE FOR

1693
01:04:16,827 --> 01:04:17,194
YOU?

1694
01:04:17,194 --> 01:04:17,461
>> SURE.

1695
01:04:17,461 --> 01:04:20,131
FIRST OF ALL MOST OF WHAT WE DO

1696
01:04:20,131 --> 01:04:22,433
ARE GENERATIVE MODELING, SO NOT

1697
01:04:22,433 --> 01:04:23,968
ONLY VIRTUAL SCREENING, VIRTUAL

1698
01:04:23,968 --> 01:04:25,336
KREENING IN OUR CASE

1699
01:04:25,336 --> 01:04:27,705
REINFORCEMENT LEARNING IS VERY

1700
01:04:27,705 --> 01:04:28,506
CHEAP.

1701
01:04:28,506 --> 01:04:30,474
SO USUALLY FOR 1 ACCESS SIZE,

1702
01:04:30,474 --> 01:04:37,882
YET TO A HIT, OR KIND OF A

1703
01:04:37,882 --> 01:04:38,783
REASONABLY DIFFICULT TARGET, YOU

1704
01:04:38,783 --> 01:04:41,919
WOULD SPEND ABOUT 72 HOURS IN

1705
01:04:41,919 --> 01:04:44,789
COMPUTE, MAYBE ABOUT -- MAYBE

1706
01:04:44,789 --> 01:04:46,557
SORRY ABOUT $5000 IN ADDITION TO

1707
01:04:46,557 --> 01:04:50,928
ANNUAL LICENSE FOR THE TOOL.

1708
01:04:50,928 --> 01:04:52,496
THAT WOULD INVOLVE GENERATION

1709
01:04:52,496 --> 01:04:53,964
AND SCREENING AT THE SAME TIME

1710
01:04:53,964 --> 01:04:56,133
BECAUSE YOU GENERATE USING 42

1711
01:04:56,133 --> 01:04:59,170
MODEL ANDS THEN YOU SCREEN USING

1712
01:04:59,170 --> 01:05:01,238
MORE THAN 700 MODELS.

1713
01:05:01,238 --> 01:05:02,907
WE ALSO HAVE ALL CHEMISTRY WHICH

1714
01:05:02,907 --> 01:05:05,443
ALLOWS TO YOU DO PHYSICS BASED

1715
01:05:05,443 --> 01:05:07,345
MODELING BUT THAT 1 IS SUPER

1716
01:05:07,345 --> 01:05:08,779
CHEAP AS LUSUALLY YOU'RE LEFT

1717
01:05:08,779 --> 01:05:13,351
WITH LIKE 20 MOLECULES AS OUT OF

1718
01:05:13,351 --> 01:05:16,053
CHEMISTRY 42 AND YOU WANT TO DO

1719
01:05:16,053 --> 01:05:19,256
HIGHER CONFIDENCE, YOU CAN DO

1720
01:05:19,256 --> 01:05:19,890
CERTAINLY--CERTAINLY FT DOCKING

1721
01:05:19,890 --> 01:05:24,128
ASK THAT WOULD COST YOU MAYBE AN

1722
01:05:24,128 --> 01:05:26,497
ADDITIONAL THOUSAND DOLLARS.

1723
01:05:26,497 --> 01:05:27,264
AND TRADITIONAL BENCHMARKS FOR

1724
01:05:27,264 --> 01:05:30,167
THE MODELS OF COURSE THEY ARE

1725
01:05:30,167 --> 01:05:33,571
APPLICABLE BUT WE HAVE OUR OWN

1726
01:05:33,571 --> 01:05:35,773
PIPELINE FOR [INDISCERNIBLE]

1727
01:05:35,773 --> 01:05:38,142
IDs, SO WE ALREADY ESTABLISHED

1728
01:05:38,142 --> 01:05:39,477
OUR OWN BENCHMARKS.

1729
01:05:39,477 --> 01:05:41,345
>> WE WILL TAKE 1 MORE,
PROBABLY

1730
01:05:41,345 --> 01:05:42,913
LAST QUESTION AND IT YOU CAN

1731
01:05:42,913 --> 01:05:44,048
STAY ONLINE TO ANSWER SOME OF

1732
01:05:44,048 --> 01:05:44,915
THE OTHER QUESTIONS, THAT WOULD

1733
01:05:44,915 --> 01:05:45,182
BE GREAT.

1734
01:05:45,182 --> 01:05:46,784
ONE OF THE QUESTIONS WE HEARD IS

1735
01:05:46,784 --> 01:05:49,587
IN CASE OF AN ABSENT CRYSTAL

1736
01:05:49,587 --> 01:05:50,454
STRUCTURE, YOU HAVE SHOWN 1

1737
01:05:50,454 --> 01:05:51,055
SUCCESS.

1738
01:05:51,055 --> 01:05:52,356
HAVE YOU TRIED OTHER TARGETS

1739
01:05:52,356 --> 01:05:56,827
WHILE GOING WITH THE SAME

1740
01:05:56,827 --> 01:05:58,763
PIPELINE, IF YES, HOW MANY OF

1741
01:05:58,763 --> 01:05:59,530
THOSE RESULTED IN ACCESS?

1742
01:05:59,530 --> 01:06:01,465
>> IF YOU ARE TALKING
ABOUT

1743
01:06:01,465 --> 01:06:05,603
ALPHA FOLD, WE PUBLISHED 1 ON

1744
01:06:05,603 --> 01:06:11,142
SIGNATURES 2 TARGET,  POC STUDY,

1745
01:06:11,142 --> 01:06:13,144
VERY RARELY WOULD WE GO FOR

1746
01:06:13,144 --> 01:06:14,545
CRYSTAL STRUCTURE FOR A

1747
01:06:14,545 --> 01:06:14,979
COMMERCIAL PROGRAM.

1748
01:06:14,979 --> 01:06:17,014
USUALLY FOR A COMMERCIAL PROGRAM

1749
01:06:17,014 --> 01:06:18,215
WHEN YOU'RE GOING OR IF YOU'RE

1750
01:06:18,215 --> 01:06:19,950
PLANNING TO INVEST, YOU KNOW

1751
01:06:19,950 --> 01:06:22,420
$20 MILLION IN THE PROGRAM, YOU

1752
01:06:22,420 --> 01:06:26,323
PROBABLY GET 4 OR 5 CRYSTAL

1753
01:06:26,323 --> 01:06:29,260
STRUCTURES, BECAUSE IT'S, YOU

1754
01:06:29,260 --> 01:06:31,529
KNOW $30,000 IN 2 OR 3 WEEKS,

1755
01:06:31,529 --> 01:06:34,398
DEPENDING ON WHICH CRO YOU USE,

1756
01:06:34,398 --> 01:06:37,301
SO MY RECOMMENDATION IS ALWAYS

1757
01:06:37,301 --> 01:06:38,068
GET THE CRYSTAL.

1758
01:06:38,068 --> 01:06:41,672
>> I WOULD LIKE TO THANK
ALL THE

1759
01:06:41,672 --> 01:06:41,939
SPEAKERS.

1760
01:06:41,939 --> 01:06:42,973
THANKS ALEX AGAIN, I WOULD LIKE

1761
01:06:42,973 --> 01:06:46,377
TO THANK THE SPEAKERS IN IN

1762
01:06:46,377 --> 01:06:49,146
PARTICULAR SESSION, SEAN, WOODY,

1763
01:06:49,146 --> 01:06:51,549
MARISSA AS WELL AS ALEX,

1764
01:06:51,549 --> 01:06:52,550
EXCELLENT PRESENTATIONS AND A

1765
01:06:52,550 --> 01:06:54,151
LOT TO THINK ABOUT FOR THE FIELD

1766
01:06:54,151 --> 01:06:55,986
IN GENERAL EMPLOY I WILL THANKED

1767
01:06:55,986 --> 01:06:57,054
OVER TO YOU BEFORE THE LUNCH

1768
01:06:57,054 --> 01:06:58,489
BREAK IF THERE ARE ANY

1769
01:06:58,489 --> 01:07:01,892
ANNOUNCEMENTS TO BREAK.

1770
01:07:01,892 --> 01:07:05,329
THANK YOU, EVERYONE. 

1771
01:07:05,329 --> 01:07:05,996
>> YES, THANK YOU.

1772
01:07:05,996 --> 01:07:06,831
ALL RIGHT.

1773
01:07:06,831 --> 01:07:10,201
OH MY GOODNESS, HOW IMPRESSIVE

1774
01:07:10,201 --> 01:07:12,736
LIKE THESE 2 TALKS THIS MORNING

1775
01:07:12,736 --> 01:07:15,973
WERE AMAZING SO, THANK YOU FOR

1776
01:07:15,973 --> 01:07:16,340
THE SPEAKERS.

1777
01:07:16,340 --> 01:07:19,477
SO I HAVE PUT THE AGENDA UP

1778
01:07:19,477 --> 01:07:21,779
AGAIN, SO, WE'RE GOING TO TAKE A

1779
01:07:21,779 --> 01:07:23,247
SHORT LUNCH BREAK RIGHT NOW, AND

1780
01:07:23,247 --> 01:07:27,451
WE WILL BE BACK EXACTLY AT 12:30

1781
01:07:27,451 --> 01:07:28,285
TO START SESSION 4. OKAY, THIS
IS GOING TO BE ON

1782
01:07:28,285 --> 01:07:33,691
EMERGING TRENDS IN THIS SILICO
DRUG DISCOVERY AND WE WILL HAVE

1783
01:07:33,691 --> 01:07:35,059
THE NEWLY IMPLEMENTED NOBEL
LAUREATE AFTER THIS LUNCH

1784
01:07:35,059 --> 01:07:41,866
SESSION EMPLOY SO 12:30, THANK
YOU VERY MUCH, SO WE STAY ON
TIME.

1785
01:07:41,866 --> 01:07:52,111
SEE YOU BACK HERE AT 12:30.

1786
01:07:52,111 --> 01:07:53,212
>> WELCOME BACK EVERYONE,
LET'S

1787
01:07:53,212 --> 01:07:56,482
TART SESSION 4 WHICH WILL BE ON

1788
01:07:56,482 --> 01:07:58,317
EMERGING TRENDS IN INSILICO DRUG

1789
01:07:58,317 --> 01:07:58,784
DORPHY.

1790
01:07:58,784 --> 01:07:59,518
I WILL INTRODUCE MY COLLEAGUE

1791
01:07:59,518 --> 01:08:04,090
AND CO CHAIR OF THIS ORGANIZING

1792
01:08:04,090 --> 01:08:09,795
COMMITTEE, ALEX EY ZAKHAROV,

1793
01:08:09,795 --> 01:08:11,097
HERE WITH NCATS.

1794
01:08:11,097 --> 01:08:12,465
THE FLOOR IS YOURS.

1795
01:08:12,465 --> 01:08:13,366
>> EXCITED TO SEE
EVERYBODY ON

1796
01:08:13,366 --> 01:08:17,636
ZOOM AS WELL AS VIDEOCAST.

1797
01:08:17,636 --> 01:08:20,473
>> LET ME OPEN UP SESSION
4 IN

1798
01:08:20,473 --> 01:08:23,776
WHICH WE WILL DISCUSS EMERGING

1799
01:08:23,776 --> 01:08:25,211
TRENDS IN INSILICO DRUG

1800
01:08:25,211 --> 01:08:25,678
DISCOVERY.

1801
01:08:25,678 --> 01:08:28,781
SO JUST A REMINDER, WE WILL HAVE

1802
01:08:28,781 --> 01:08:30,683
25 MINUTES FOR THE SPEAKER AND 5

1803
01:08:30,683 --> 01:08:33,819
MINUTES IF ARE Q&A, WE HAVE
A

1804
01:08:33,819 --> 01:08:34,487
PRETTY BUSY SCHEDULE TODAY, AND

1805
01:08:34,487 --> 01:08:37,289
I WILL TURN ON MY CAMERA 2

1806
01:08:37,289 --> 01:08:39,425
MINUTES BEFORE Q&A, WE WILL
DO

1807
01:08:39,425 --> 01:08:43,763
THE HARD STOP AFTER 30  MINUTES

1808
01:08:43,763 --> 01:08:44,497
IN TOTAL.

1809
01:08:44,497 --> 01:08:46,532
WITH THAT SAID, OUR FIRST

1810
01:08:46,532 --> 01:08:50,036
SPEAKER OF THE SESSION IS

1811
01:08:50,036 --> 01:08:52,538
BRYN TAYLOR, IS WITH JANUARY
& 

1812
01:08:52,538 --> 01:08:53,606
JANUARY INNOVATIVE MEDICINE, SHE

1813
01:08:53,606 --> 01:08:55,775
GOT OUR PHASH D AT THE

1814
01:08:55,775 --> 01:08:56,776
UNIVERSITY OF CALIFORNIA SAN

1815
01:08:56,776 --> 01:08:57,410
DIEGO IN 2020.

1816
01:08:57,410 --> 01:08:59,378
SHE WORKS ON EARLY STAGES OF

1817
01:08:59,378 --> 01:09:01,881
DRUG DEVELOPMENT FOR DIFFICULT

1818
01:09:01,881 --> 01:09:04,016
TARGETS IN ONCOLOGY, WE FOCUS ON

1819
01:09:04,016 --> 01:09:05,751
USING COMPUTATIONAL TOOLS TO

1820
01:09:05,751 --> 01:09:11,891
ADVANCE MODALITIES SUCH AS

1821
01:09:11,891 --> 01:09:12,558
TARGET PROTEIN DEGRADATIONS.

1822
01:09:12,558 --> 01:09:16,796
TAYLOR THE STAGE IS YOURS. 

1823
01:09:16,796 --> 01:09:17,663
>> THEY THINKS LEXY.

1824
01:09:17,663 --> 01:09:28,140
I WILL GO AHEAD AND SHARE MY

1825
01:09:29,141 --> 01:09:29,475
SCREEN.

1826
01:09:29,475 --> 01:09:33,012
CAN YOU SEE THE CORRECT SCREEN?

1827
01:09:33,012 --> 01:09:33,279
>> YES.

1828
01:09:33,279 --> 01:09:33,579
>> GREAT.

1829
01:09:33,579 --> 01:09:34,413
THANK YOU VERY MUCH FOR THE

1830
01:09:34,413 --> 01:09:36,382
INTRODUCTION AND THANKS TO THE

1831
01:09:36,382 --> 01:09:40,986
ORGANIZERS ARE HAVING ME HERE.

1832
01:09:40,986 --> 01:09:45,825
TODAY SO AS LEXY SAID I WORK AT

1833
01:09:45,825 --> 01:09:47,326
JANUARY &  JANUARY WITH THE

1834
01:09:47,326 --> 01:09:49,929
MEDICAL DISCOVERY GROUP IN

1835
01:09:49,929 --> 01:09:56,102
LA HOYA, AND WE WILL TALK ABOUT

1836
01:09:56,102 --> 01:09:56,569
UNDERSTANDING PROTAC

1837
01:09:56,569 --> 01:09:57,269
PERMIABILITY USING MODEL CITIZEN

1838
01:09:57,269 --> 01:09:57,903
LEAKULAR DISCIPLINARY MEDICARE

1839
01:09:57,903 --> 01:09:59,672
AND MEDICAIDICS IN SIMULATIONS

1840
01:09:59,672 --> 01:10:00,873
AND MODELS AND DEEP LEARNING.

1841
01:10:00,873 --> 01:10:10,649
SO AS A BRIEF BACKGROUND, SO

1842
01:10:10,649 --> 01:10:11,851
THEY ARE HETERO-- THEY HAVE 2

1843
01:10:11,851 --> 01:10:14,186
WAR HEADS AND A LINKER SO 1 WAR

1844
01:10:14,186 --> 01:10:16,122
HEAD WILL BIND THE PROTEIN OF

1845
01:10:16,122 --> 01:10:18,224
INTEREST, THE POI THERE AND THE

1846
01:10:18,224 --> 01:10:21,660
OTHER 1 BINDS AN E3 LIGASE, SO

1847
01:10:21,660 --> 01:10:23,162
THE E3 DISCIPLINARY LAS PINTAS

1848
01:10:23,162 --> 01:10:25,164
WILL TACK THE UBIQUITIN

1849
01:10:25,164 --> 01:10:26,665
TARGETING THE DEGRADATION BY THE

1850
01:10:26,665 --> 01:10:27,133
PROTEIN COMPLEXIOSOME.

1851
01:10:27,133 --> 01:10:30,269
SO THIS IS CONTRAST THE CLASSIC

1852
01:10:30,269 --> 01:10:31,437
SMALL MOLECULE INHIBITOR

1853
01:10:31,437 --> 01:10:33,272
APPROACH FOR TARGETING A PROTEIN

1854
01:10:33,272 --> 01:10:34,607
AND SIMPLY REMOVING ITSELF

1855
01:10:34,607 --> 01:10:42,114
FUNCTION, THIS 1 DEGRADE THE

1856
01:10:42,114 --> 01:10:42,648
ENTIRE PROTEIN ALTOGETHER.

1857
01:10:42,648 --> 01:10:47,219
SO AS YOU MIGHT GUESS, A COMMON

1858
01:10:47,219 --> 01:10:52,258
ISSUE WITH USING PROTAC'S DRUGS

1859
01:10:52,258 --> 01:10:54,193
WHICH IS DUE TO LOW VIABILITY

1860
01:10:54,193 --> 01:10:57,463
BUT THAT MAY BE PART TO THEIR

1861
01:10:57,463 --> 01:10:58,697
LOW PERMIABILITY.

1862
01:10:58,697 --> 01:11:00,900
SO I'M SHOWING A PROTAC TO

1863
01:11:00,900 --> 01:11:05,571
GREATER FIELD IN INDUSTRY, SO MY

1864
01:11:05,571 --> 01:11:07,373
FORMER COLLEAGUE LORI

1865
01:11:07,373 --> 01:11:08,207
[INDISCERNIBLE] DEMONSTRATED

1866
01:11:08,207 --> 01:11:09,175
THIS AT SEVERAL PHARMACEUTICAL

1867
01:11:09,175 --> 01:11:10,910
COMPANIES AND ASKED THEM TO

1868
01:11:10,910 --> 01:11:14,980
REPORT WHAT WAS THE MOST

1869
01:11:14,980 --> 01:11:17,816
SIGNIFICANT ADME OR PHYSIO

1870
01:11:17,816 --> 01:11:20,920
CHEMICAL PROPERTIES, FACING

1871
01:11:20,920 --> 01:11:22,087
DEVELOPMENT OF ORALLY

1872
01:11:22,087 --> 01:11:23,689
BIOAVAILABILITY DEGRADE ORS, SO

1873
01:11:23,689 --> 01:11:26,225
IN BLUE IS THE PRIMARY

1874
01:11:26,225 --> 01:11:26,825
CHALLENGE.

1875
01:11:26,825 --> 01:11:28,060
RED WAS SECONDARY CHALLENGE,

1876
01:11:28,060 --> 01:11:30,296
THIS WAS PEOPLE VOTING IN

1877
01:11:30,296 --> 01:11:33,599
SURVEY, PEOPLE AGREED IT WAS

1878
01:11:33,599 --> 01:11:34,233
PERMIABILITY, SOLUBILITY ALONG

1879
01:11:34,233 --> 01:11:35,201
WITH MODEL CITIZEN LEAKULAR

1880
01:11:35,201 --> 01:11:38,571
WEIGHT WHICH DOES CONTRIBUTE TO

1881
01:11:38,571 --> 01:11:40,439
PERMIA ABILITY AND SOLUABILITY

1882
01:11:40,439 --> 01:11:41,840
AND SECONDARY CHALLENGES, THESE

1883
01:11:41,840 --> 01:11:45,044
PROBLEMS ARE VERY COMPLEX, SO IT

1884
01:11:45,044 --> 01:11:48,480
SPANS EVERYTHING BUT INCLUDES

1885
01:11:48,480 --> 01:11:51,317
THINGS LIKE ABSORPTION, LIQUID

1886
01:11:51,317 --> 01:11:52,685
[INDISCERNIBLE] AND THINGS WE

1887
01:11:52,685 --> 01:11:54,720
WILL TALK ABOUT BRIEFLY.

1888
01:11:54,720 --> 01:11:57,623
SO AN EXAMPLE OF A PROTAC IS ON

1889
01:11:57,623 --> 01:12:00,559
THE RIGHT, IT'S A PART OF A

1890
01:12:00,559 --> 01:12:02,795
LARGER DATA SET PATENTED BY
J& 

1891
01:12:02,795 --> 01:12:04,763
J, AND BECAUSE IT'S PATENTED I

1892
01:12:04,763 --> 01:12:06,532
GET TO USE IT AS AN EXAMPLE

1893
01:12:06,532 --> 01:12:07,533
THROUGHOUT THIS TALK.

1894
01:12:07,533 --> 01:12:10,169
I WILL ALSO SPEAK ABOUT INTERNAL

1895
01:12:10,169 --> 01:12:11,870
DATA, I WON'T SHOW STRUCTURES

1896
01:12:11,870 --> 01:12:22,281
BUT I AM ALLOWED TO TALK

1897
01:12:30,022 --> 01:12:31,590
 ABOUT

1898
01:12:31,590 --> 01:12:31,757
IT.

1899
01:12:31,757 --> 01:12:34,526
THESE ARE CANINE KIDNEY CELLS

1900
01:12:34,526 --> 01:12:38,764
AND DRUGS AND PROBLEM IS MDCK

1901
01:12:38,764 --> 01:12:40,065
CELLS EXHIBIT ENDOGENOUS

1902
01:12:40,065 --> 01:12:42,601
ACTIVITY WHICH LEADS TO AN

1903
01:12:42,601 --> 01:12:43,602
UNDERESTIMATION OF PERMIABILITY.

1904
01:12:43,602 --> 01:12:44,937
SO, TO GET AROUND THIS, WHAT WE

1905
01:12:44,937 --> 01:12:50,509
DO IN MANY OTHER GROUPS DO, IS

1906
01:12:50,509 --> 01:12:55,414
WE USE MDCK CELLS TRANSPORTERS,

1907
01:12:55,414 --> 01:12:57,816
SO WHENEVER I TALK ABOUT

1908
01:12:57,816 --> 01:12:58,450
CORRELATIONS WITH PERMIABILITY

1909
01:12:58,450 --> 01:13:04,490
THIS IS THE ASSAY WE ARE GETTING

1910
01:13:04,490 --> 01:13:05,057
OUR DATA FROM.

1911
01:13:05,057 --> 01:13:07,326
AND WHEN WE USE THIS

1912
01:13:07,326 --> 01:13:08,560
EXPERIMENTAL PERMITIABILITY DATA

1913
01:13:08,560 --> 01:13:11,130
FROM THESE KNOCK OUT MDCK CELL

1914
01:13:11,130 --> 01:13:12,564
LINES, WE ARE ABLE TO TRAIN

1915
01:13:12,564 --> 01:13:14,033
MACHINE LEARNING MODELS THAT

1916
01:13:14,033 --> 01:13:16,568
CORRELATE PRETTY WELL WITH

1917
01:13:16,568 --> 01:13:18,537
EXPERIMENT, THIS IS -- SO, ON

1918
01:13:18,537 --> 01:13:22,107
THE LEFT-HAND SIDE OF THIS SLIDE

1919
01:13:22,107 --> 01:13:23,876
IN BLUE, YOU CAN SEE ALL THE

1920
01:13:23,876 --> 01:13:26,545
DATA FOR A SERIES OF PROTAC AND

1921
01:13:26,545 --> 01:13:28,113
ON THE RIGHT AND ORANGE IS JUST

1922
01:13:28,113 --> 01:13:29,548
THE SET OF COMPOUNDS THAT THE

1923
01:13:29,548 --> 01:13:31,383
MODEL HAD NEVER SEEN AND WAS

1924
01:13:31,383 --> 01:13:33,819
SYNTHESIZED AFTER THE MODEL WAS

1925
01:13:33,819 --> 01:13:34,053
CREATED.

1926
01:13:34,053 --> 01:13:37,156
SO, YOU KNOW THIS IS IN BLUE

1927
01:13:37,156 --> 01:13:39,358
ACTUALLY JUST THE TEST SET OF

1928
01:13:39,358 --> 01:13:42,461
ALL DATA SO 25% AND THE

1929
01:13:42,461 --> 01:13:47,266
STATISTICS FOR THAT RSQUARED,

1930
01:13:47,266 --> 01:13:48,267
.77, THE RANK ORDER IS QUITE

1931
01:13:48,267 --> 01:13:49,601
GOOD AND YOU CAN SEE THAT WE'RE

1932
01:13:49,601 --> 01:13:52,271
ABLE TO GET A PRETTY GOOD

1933
01:13:52,271 --> 01:13:54,640
UNDERSTANDING OF HOW THESE

1934
01:13:54,640 --> 01:13:56,775
PROTAC AT LEAST IN THIS SPECIFIC

1935
01:13:56,775 --> 01:13:58,811
SERIES, SO IT'S NOT NECESSARILY

1936
01:13:58,811 --> 01:14:00,746
GENERALIZABLE TO THIS EVER, BUT

1937
01:14:00,746 --> 01:14:01,747
IN THIS SERIES WE'RE ABLE TO GET

1938
01:14:01,747 --> 01:14:04,817
A GOOD IDEA OF WHAT THE

1939
01:14:04,817 --> 01:14:06,385
PERMIABILITY WILL BE WHICH IS

1940
01:14:06,385 --> 01:14:07,853
REALLY EXCITING AND IS VERY

1941
01:14:07,853 --> 01:14:09,521
HELPFUL WHEN WE HAVE LARGE

1942
01:14:09,521 --> 01:14:10,856
LIBRARIES OF COMPOUNDS THAT WE

1943
01:14:10,856 --> 01:14:15,661
NEED TO SELECT JUST A SET TO

1944
01:14:15,661 --> 01:14:15,928
MAKE.

1945
01:14:15,928 --> 01:14:17,663
HOWEVER, THESE ARE WHAT A JUST

1946
01:14:17,663 --> 01:14:19,932
TALKED ABOUT HEAR LIGAND BASED

1947
01:14:19,932 --> 01:14:22,568
MACHINE LEARNING MODELS, 2 D,

1948
01:14:22,568 --> 01:14:23,435
FINGERPRINT BASED MODELS, AND

1949
01:14:23,435 --> 01:14:26,004
THERE ARE ACTUALLY LESS USEFUL

1950
01:14:26,004 --> 01:14:27,172
FOR IESHES DENTIFICATION OF WHAT

1951
01:14:27,172 --> 01:14:29,174
ARE ACTUALLY THE KEY

1952
01:14:29,174 --> 01:14:29,975
CONTRIBUTORS TO PERMIABILITY,

1953
01:14:29,975 --> 01:14:32,044
ESPECIALLY IF THE CONTRIBUTORS

1954
01:14:32,044 --> 01:14:34,380
AREN'T JUST A SINGLE PROPERTY,

1955
01:14:34,380 --> 01:14:35,981
OR OBVIOUS DESCRIPTORS, AND

1956
01:14:35,981 --> 01:14:36,782
THAT'S LIEWCIALLY WHAT WE FIND

1957
01:14:36,782 --> 01:14:39,752
IS THE CASE, THIS IS A COMPLEX

1958
01:14:39,752 --> 01:14:43,822
ISSUE, AND YOU KNOW FROM THE

1959
01:14:43,822 --> 01:14:45,891
INDUSTRY PERSPECTIVES PAPER THEY

1960
01:14:45,891 --> 01:14:47,292
MENTIONED EARLIER, SOME FEATURES

1961
01:14:47,292 --> 01:14:49,128
THAT MAY BE ASSOCIATED WITH

1962
01:14:49,128 --> 01:14:50,028
PROTECT PERMIABILITY THAT WE

1963
01:14:50,028 --> 01:14:52,164
HAVE LOOKED AT AND DON'T

1964
01:14:52,164 --> 01:14:54,600
NECESSARILY CORRELATE PERFECTLY

1965
01:14:54,600 --> 01:14:58,504
WITH THE PROTAC PERMIABILITY MAY

1966
01:14:58,504 --> 01:15:01,140
CONTRIBUTE SO THINGS LIKE OF A

1967
01:15:01,140 --> 01:15:05,677
SURFACE AREA OF 100-200.

1968
01:15:05,677 --> 01:15:07,012
MOLECULAR WEIGHT, UNDER 900

1969
01:15:07,012 --> 01:15:09,148
WOULD BE IDEAL, DONORS AND

1970
01:15:09,148 --> 01:15:10,616
ACCEPT ACCEPTORS, AND THE

1971
01:15:10,616 --> 01:15:13,485
HYDROGEN BONDS, YOU CAN SEE THE

1972
01:15:13,485 --> 01:15:14,953
TABLE THERE, LESS THAN 14

1973
01:15:14,953 --> 01:15:17,022
HYDROGEN BOND RECEPTORS, AND

1974
01:15:17,022 --> 01:15:20,526
LESS THAN 500 HYDROGEN BOND

1975
01:15:20,526 --> 01:15:22,694
DONORS AND LIP O PHYLLISITY

1976
01:15:22,694 --> 01:15:29,902
PLOAS A ROLE AND A PH OF AROUND

1977
01:15:29,902 --> 01:15:31,170
7.4 WOULD BE GREAT.

1978
01:15:31,170 --> 01:15:34,473
BUT THE QUESTION IS DOES IT GO

1979
01:15:34,473 --> 01:15:36,542
BEYOND THE DESCRIPTORS AND IS

1980
01:15:36,542 --> 01:15:39,178
THERE A COMPONENT OF THE

1981
01:15:39,178 --> 01:15:41,346
RIGIDITY OR FLEXIBILITY OF THE

1982
01:15:41,346 --> 01:15:41,580
PROTAC.

1983
01:15:41,580 --> 01:15:43,649
I BRING IT UP BECAUSE 1 OF THE

1984
01:15:43,649 --> 01:15:45,918
WAYS THEY ARE ABLE TO XIPT THE

1985
01:15:45,918 --> 01:15:46,985
CELLS IS THAT THEY

1986
01:15:46,985 --> 01:15:47,352
[INDISCERNIBLE].

1987
01:15:47,352 --> 01:15:48,987
SO IN A MEMBRANE LIKE POLAR

1988
01:15:48,987 --> 01:15:54,026
VIERNLT, THEY'RE ON THE LEFT IN

1989
01:15:54,026 --> 01:15:55,461
RED, THEY ASSUME THIS COLLAPSED

1990
01:15:55,461 --> 01:15:57,429
KIND OF STATE, SO YOU KNOW CRUMP

1991
01:15:57,429 --> 01:16:00,566
PELLED UP AND IN A POLAR

1992
01:16:00,566 --> 01:16:02,201
ENVIRONMENT SUCH AS

1993
01:16:02,201 --> 01:16:05,938
INTRACELLULAR OR EXTRA CELLULAR,

1994
01:16:05,938 --> 01:16:07,940
THEY'RE MORE YELLOW, SO BEYOND

1995
01:16:07,940 --> 01:16:10,209
THE ABILITY THOUGH, TO

1996
01:16:10,209 --> 01:16:11,844
TRANSITION BETWEEN THE STATES

1997
01:16:11,844 --> 01:16:12,544
BASED ON THE SOFULLENT

1998
01:16:12,544 --> 01:16:15,214
ENVIRONMENT, I THINK THE RATE OF

1999
01:16:15,214 --> 01:16:16,114
TRANSITION PROBABLY MATTERS.

2000
01:16:16,114 --> 01:16:18,317
AND THAT MEANS LIKE IF A PROTAC

2001
01:16:18,317 --> 01:16:22,254
IS ABLE TO WICKLY ADAPT A

2002
01:16:22,254 --> 01:16:23,922
COLLAPSED CONFIRMATION, GET THE

2003
01:16:23,922 --> 01:16:26,592
MEMBRANE AND QUICKLY ADOPT AN

2004
01:16:26,592 --> 01:16:27,726
ELONGALTIED CONFIRMATION IN THE

2005
01:16:27,726 --> 01:16:29,027
RIGHT ENVIRONMENT, IT MAY LEAD

2006
01:16:29,027 --> 01:16:30,362
TO A BETTER CHANCE OF GETTING IN

2007
01:16:30,362 --> 01:16:31,730
AND OUT OF MEMBRANE, NOT GETTING

2008
01:16:31,730 --> 01:16:34,433
STUCK IN THE MEMBRANE AND ALSO

2009
01:16:34,433 --> 01:16:37,035
BEING ABLE TO ATTACH TO ITS E3

2010
01:16:37,035 --> 01:16:39,371
AND POI BINDING PARTNERS AND

2011
01:16:39,371 --> 01:16:43,308
ACTUALLY INITTIAIT THAT

2012
01:16:43,308 --> 01:16:44,276
DEGRADATION SEQUENCE.

2013
01:16:44,276 --> 01:16:47,412
SO TO ASSESS THIS HYPOTHESIS, WE

2014
01:16:47,412 --> 01:16:49,581
CREATED AN UNBIASED MOLECULAR

2015
01:16:49,581 --> 01:16:51,350
DYNAMIC SAMPLING WORK LOAD TO

2016
01:16:51,350 --> 01:16:52,885
DETERMINE THE KEY COMBINATION IS

2017
01:16:52,885 --> 01:16:55,387
IN DIFFERENT SOLVENTS SO WE USED

2018
01:16:55,387 --> 01:16:57,122
WATER, MEL AN OLDER PEOPLE AND

2019
01:16:57,122 --> 01:16:58,023
CHLOROFORM WHICH HAD

2020
01:16:58,023 --> 01:17:01,593
CONCENTRATES OF 80, 32, AND 5

2021
01:17:01,593 --> 01:17:02,661
JUST ABOUT.

2022
01:17:02,661 --> 01:17:04,830
AND CHLOROFORM WITH THE

2023
01:17:04,830 --> 01:17:07,833
DIALECTIC OF 5, AND CELL

2024
01:17:07,833 --> 01:17:10,269
MEMBRANE, AND WATER WORKS WITH

2025
01:17:10,269 --> 01:17:11,970
THE INTRACELLULAR AND EXTRA

2026
01:17:11,970 --> 01:17:14,940
CELLULAR ENVIRONMENT AND

2027
01:17:14,940 --> 01:17:16,308
METHANOL WAS SOMEWHERE IN

2028
01:17:16,308 --> 01:17:16,742
BETWEEN.

2029
01:17:16,742 --> 01:17:25,117
AND WE RAN DYNAM  DYNAMICS, ANDN

2030
01:17:25,117 --> 01:17:28,020
ABOUT 5 MICRO[INDISCERNIBLE] OF

2031
01:17:28,020 --> 01:17:30,255
MD BUT MD SIMULATIONS ARE GREAT,

2032
01:17:30,255 --> 01:17:33,392
THEY'RE FUN TO WATCH WITH SOME

2033
01:17:33,392 --> 01:17:37,162
POPCORN, BUT YOU KNOW IT'S AN

2034
01:17:37,162 --> 01:17:38,864
IMMENSE AMOUNT OF DISPARATE DATA

2035
01:17:38,864 --> 01:17:40,566
ESPECIALLY IF YOU SCALE UP

2036
01:17:40,566 --> 01:17:42,301
THROUGH OR 4 PROTAC, IF YOU ARE

2037
01:17:42,301 --> 01:17:44,670
LOOKING AT HUNDREDS THIS BECOMES

2038
01:17:44,670 --> 01:17:46,071
MORE COMPLICATED TO INTERPRET.

2039
01:17:46,071 --> 01:17:47,806
SO HOW DO WE CONDENSE THIS BIG

2040
01:17:47,806 --> 01:17:50,275
DATA DOWN TO A HUMAN

2041
01:17:50,275 --> 01:17:51,410
INTERPRETABLE FORMAT,

2042
01:17:51,410 --> 01:17:52,377
PARTICULARLY IN AN AUTOMATED

2043
01:17:52,377 --> 01:17:52,544
WAY.

2044
01:17:52,544 --> 01:17:54,780
SO THERE ARE MANY SUCH WAYS TO

2045
01:17:54,780 --> 01:17:57,149
DO THIS WONG IS TO USE THE

2046
01:17:57,149 --> 01:18:01,787
MARKER MODEL, IT'S A MARKER

2047
01:18:01,787 --> 01:18:05,524
MODEL THAT IDENTIFIES

2048
01:18:05,524 --> 01:18:06,291
KINETICALLY COMPOSITIONS AND THE

2049
01:18:06,291 --> 01:18:08,126
RATE BETWEEN THE STATES.

2050
01:18:08,126 --> 01:18:10,028
SO VERY BRIEFLY TO DO MD

2051
01:18:10,028 --> 01:18:11,029
SIMULATIONS, YOU CLUSTER AND USE

2052
01:18:11,029 --> 01:18:12,331
THE CLUSTER OF THE STATES SO

2053
01:18:12,331 --> 01:18:13,832
THIS SOUNDS LIKE A NORMAL

2054
01:18:13,832 --> 01:18:16,401
CLUSTERS WORK FLOW, WHAT IS MSM

2055
01:18:16,401 --> 01:18:18,203
IN THEORY ADD TO THIS, WELL IN

2056
01:18:18,203 --> 01:18:20,572
THE PROCESS OF MODEL REFINEMENT,

2057
01:18:20,572 --> 01:18:22,641
WE SELECT CLUSTERS THAT ARE MORE

2058
01:18:22,641 --> 01:18:24,242
SERIOUS STAKES, SO THEY'RE SO

2059
01:18:24,242 --> 01:18:25,777
STABLE THAT THEY MUST HAVE

2060
01:18:25,777 --> 01:18:29,281
FORGOTTEN WHERE THEY CAME FROM

2061
01:18:29,281 --> 01:18:30,849
BEFORE LEAVING.

2062
01:18:30,849 --> 01:18:34,586
SO WE SIGN THESE STATES.

2063
01:18:34,586 --> 01:18:37,389
AND THEN AFTER WE DEFINEITATES,

2064
01:18:37,389 --> 01:18:39,124
WE CAN ASSESS THE PROBABILITY OF

2065
01:18:39,124 --> 01:18:42,060
THE LIGAND STAYING IN THAT

2066
01:18:42,060 --> 01:18:42,761
COMPOSITIONAL STATE OR

2067
01:18:42,761 --> 01:18:44,062
TRANSITIONING TO ANOTHER STATE.

2068
01:18:44,062 --> 01:18:47,332
SO THIS -- THIS METHOD, YOU KNOW

2069
01:18:47,332 --> 01:18:48,834
A ALLOWS US TO IESHES DENTIFY

2070
01:18:48,834 --> 01:18:50,736
THE STATE OF OUR PROTAC AND

2071
01:18:50,736 --> 01:18:52,204
UNDERSTAND HOW WELL THE PROTAC

2072
01:18:52,204 --> 01:18:53,238
MOVES BETWEEN THOSE STATES.

2073
01:18:53,238 --> 01:18:55,007
SO THIS IS -- THIS IS THE KIND

2074
01:18:55,007 --> 01:18:57,042
OF THE PERFECT APPLICATION FOR

2075
01:18:57,042 --> 01:19:00,712
THIS QUESTION WE HAD BUT THE

2076
01:19:00,712 --> 01:19:05,884
ONLY ISSUE IS, YOU KNOW

2077
01:19:05,884 --> 01:19:07,419
CLASSICAL MARKER STATE MODELS

2078
01:19:07,419 --> 01:19:09,121
ARE WHEN MANY STATES ARE USED

2079
01:19:09,121 --> 01:19:11,189
BUT BECAUSE WE'RE ONLY

2080
01:19:11,189 --> 01:19:13,225
INTERESTED IN 2 STATES COLLAPSED

2081
01:19:13,225 --> 01:19:14,593
AND ELONGATED, WE USED

2082
01:19:14,593 --> 01:19:15,661
[INDISCERNIBLE] WHICH CAN

2083
01:19:15,661 --> 01:19:21,433
PRODUCE AN ACCURATE MODEL WITH

2084
01:19:21,433 --> 01:19:24,036
VERY FEW STATES SO VASM NETS

2085
01:19:24,036 --> 01:19:26,138
REPLACES AND OF COURSE MSCORE

2086
01:19:26,138 --> 01:19:28,140
SCREENING WITH A DEEP NEURAL

2087
01:19:28,140 --> 01:19:29,141
LEARNING NETWORK, WE DIDN'T COME

2088
01:19:29,141 --> 01:19:31,309
UP WITH ANY OF THIS ALTHOUGH THE

2089
01:19:31,309 --> 01:19:32,310
CITATIONS ARE AT THE BOTTOM,

2090
01:19:32,310 --> 01:19:36,014
THIS IS JUST WHAT WE OOH PLIED

2091
01:19:36,014 --> 01:19:39,084
TO THIS PROJECT. 

2092
01:19:39,084 --> 01:19:42,220
SO ONCE TRAJECTORY ARE

2093
01:19:42,220 --> 01:19:45,290
DISCREETIZED INTO METASTABLE

2094
01:19:45,290 --> 01:19:48,460
STATES USING MARKOV STATE MODEL

2095
01:19:48,460 --> 01:19:50,228
AND ANALYZED WITH NEURAL

2096
01:19:50,228 --> 01:19:51,630
NETWORKS IT PROVIDES 2 STATES

2097
01:19:51,630 --> 01:19:53,899
ELONGATED YOU CAN SEE COLLAPSE

2098
01:19:53,899 --> 01:19:55,967
IN THE SALMON COLOR AND THE

2099
01:19:55,967 --> 01:19:57,636
ELONGATED IN THE TEAL AND I WILL

2100
01:19:57,636 --> 01:20:00,172
EXPLAIN LATER ON HOW THE STATES

2101
01:20:00,172 --> 01:20:01,473
ARE IDENTIFIED AND WHAT

2102
01:20:01,473 --> 01:20:02,774
COLLECTIVE VARIABLES WERE USED

2103
01:20:02,774 --> 01:20:07,245
BUT THIS IS WHAT WE WANT TO GET

2104
01:20:07,245 --> 01:20:11,783
OUT OF THIS ANALYSIS.

2105
01:20:11,783 --> 01:20:14,453
SO WE SELECTED COMPOUNDS OF A

2106
01:20:14,453 --> 01:20:16,054
RANGE OF PERMIABILITY VALUES,

2107
01:20:16,054 --> 01:20:17,789
THE DATA AND EVERYTHING, I'M

2108
01:20:17,789 --> 01:20:19,925
SHOWING HERE IS ACTUALLY FROM

2109
01:20:19,925 --> 01:20:21,893
OUR INTERNAL DATA SET SO I CAN'T

2110
01:20:21,893 --> 01:20:23,061
SHOW THE ACTUAL STRUCTURES, YOU

2111
01:20:23,061 --> 01:20:28,834
CAN SEE A NUMBER OF PROTAC HERE

2112
01:20:28,834 --> 01:20:33,905
HAVE A VERY LOW MDCK KNOCK OUT

2113
01:20:33,905 --> 01:20:36,408
VALUE, THIS IS BY NO MEANS A

2114
01:20:36,408 --> 01:20:37,375
PERMEABLE COMPOUND BUT IT IS THE

2115
01:20:37,375 --> 01:20:39,644
RANGE AND THE RANGE WE ARE

2116
01:20:39,644 --> 01:20:40,011
WORKING WITH.

2117
01:20:40,011 --> 01:20:42,714
SO FIRST WE HAD TO DEFINE WHAT

2118
01:20:42,714 --> 01:20:43,749
COLLECTIVE VARIABLES WE WERE

2119
01:20:43,749 --> 01:20:45,617
BUILDING THE MODELS ON AT ALL SO

2120
01:20:45,617 --> 01:20:48,086
HOW DID WE DEFINE STATES, HOW DO

2121
01:20:48,086 --> 01:20:50,489
WE DEFINE COLLAPSE VERSUS

2122
01:20:50,489 --> 01:20:52,324
ELONGATED SO WE ALREADY TALKED

2123
01:20:52,324 --> 01:20:54,059
ABOUT THESE NUMBERS OF PROTAC,

2124
01:20:54,059 --> 01:20:55,393
THAT MIGHT CONTRIBUTE TO

2125
01:20:55,393 --> 01:20:58,396
PERMIABILITY OR VIABILITY BUT

2126
01:20:58,396 --> 01:20:59,064
OTHER CHARACTERISTICS CAN WE

2127
01:20:59,064 --> 01:21:03,935
PULL FROM THE SIMULATION?

2128
01:21:03,935 --> 01:21:06,438
SO EXAMPLES INCLUDE SAMPLES,

2129
01:21:06,438 --> 01:21:10,275
BOND OR RSD, OR WE SETTLED ON

2130
01:21:10,275 --> 01:21:12,577
END-TO-END DISTANCE AND THAT'S

2131
01:21:12,577 --> 01:21:14,179
JUST MEASURING THE SIMPLE

2132
01:21:14,179 --> 01:21:14,813
PROTAC.

2133
01:21:14,813 --> 01:21:16,481
SO SIMPLE METRIC.

2134
01:21:16,481 --> 01:21:19,017
WE TESTED PROBABLY 2 DOZEN OF

2135
01:21:19,017 --> 01:21:19,885
THESE DIFFERENT DESCRIPTORS IN

2136
01:21:19,885 --> 01:21:24,890
ORDER TO TRY TO FIND SOME SORT

2137
01:21:24,890 --> 01:21:27,159
OF CORRELATION THAT WAS

2138
01:21:27,159 --> 01:21:29,427
GENERALIZABLE ACROSS AS MANY

2139
01:21:29,427 --> 01:21:30,695
PROTACS AS WE COULD FIND.

2140
01:21:30,695 --> 01:21:33,965
AND WHAT WE FOUND WAS IN THAT

2141
01:21:33,965 --> 01:21:35,367
PERMIABILITYS IS HIGHER WHEN THE

2142
01:21:35,367 --> 01:21:36,968
END-TO-END DISTANCE IS LOWER.

2143
01:21:36,968 --> 01:21:39,137
SO THE PLOTS BELOW SHOWED 3

2144
01:21:39,137 --> 01:21:41,006
DIFFERENT COMPOUNDS, THEY'RE ALL

2145
01:21:41,006 --> 01:21:43,475
IN CHLOROFORM AND THESE ARE JUST

2146
01:21:43,475 --> 01:21:44,910
MD SIMULATION DATA AND THE 1 ON

2147
01:21:44,910 --> 01:21:48,180
THE LEFT HAS LOWEST AVERAGE

2148
01:21:48,180 --> 01:21:50,816
END-TO-END DISTANCE AND HAS THE

2149
01:21:50,816 --> 01:21:52,017
HIGHEST PERMIABILITY SO

2150
01:21:52,017 --> 01:21:53,718
18.5-ANGSTROMS AND 10 TO THE

2151
01:21:53,718 --> 01:21:58,023
SIXTH, SO YOU WILL NOTICE THAT

2152
01:21:58,023 --> 01:21:59,224
THE DIFFERENCES, THE DISTANCES

2153
01:21:59,224 --> 01:22:01,026
ARE NOT VERY DIFFERENT, RIGHT?

2154
01:22:01,026 --> 01:22:03,261
SO THE LOWEST AND 18 AND A HALF,

2155
01:22:03,261 --> 01:22:04,696
THE HIGHEST IS 20.3, LIKE THIS

2156
01:22:04,696 --> 01:22:08,099
IS ALMOST IN THE NOISE HOWEVER,

2157
01:22:08,099 --> 01:22:10,202
THE COOL PART ABOUT THESE 3

2158
01:22:10,202 --> 01:22:14,539
PROTAC IS THEY ALSO HAD VERY

2159
01:22:14,539 --> 01:22:15,674
SIMILAR PERMEABLE VALUES, SO

2160
01:22:15,674 --> 01:22:18,243
EVEN A SLIGHT CHANGE IN

2161
01:22:18,243 --> 01:22:19,244
COLLAPSABILITY MAY AFFECT

2162
01:22:19,244 --> 01:22:22,347
PERMABILITY, SO THIS IS WHAT WE

2163
01:22:22,347 --> 01:22:24,950
STARTED WITH AND WHEN WE LOOKED

2164
01:22:24,950 --> 01:22:26,451
-- WHEN WE   PERFORM THIS

2165
01:22:26,451 --> 01:22:28,119
ANALYSIS AND WHAT THE SPECIFIC

2166
01:22:28,119 --> 01:22:30,288
EXAMPLES PAIRS OF HIGH AND LOW

2167
01:22:30,288 --> 01:22:31,556
PERMEABLE COMPOUNDS WE DO SEE

2168
01:22:31,556 --> 01:22:33,091
SOME DIFFERENCES SO THIS WHAT I

2169
01:22:33,091 --> 01:22:34,693
SHOWED PREVIOUSLY WAS JUST THE

2170
01:22:34,693 --> 01:22:36,695
MD SIMULATION DATA AND LOOKING

2171
01:22:36,695 --> 01:22:38,363
AT 1 DESCRIPTOR, NOW WHAT WE'VE

2172
01:22:38,363 --> 01:22:41,166
DONE IS WE'VE DONE THIS MSN WORK

2173
01:22:41,166 --> 01:22:44,336
FLOW AND WE HAVE THE RESULTS,

2174
01:22:44,336 --> 01:22:46,638
SO, I'M SHOWING HERE A WONDERFUL

2175
01:22:46,638 --> 01:22:48,673
GREEN BLOB WHICH UNFORTUNATELY I

2176
01:22:48,673 --> 01:22:50,008
CAN'T SHOW THE EXACT STRUCTURE

2177
01:22:50,008 --> 01:22:52,410
OF, BUT IT HAS A HIGHER, YOU

2178
01:22:52,410 --> 01:22:55,914
KNOW IT HAS THE -- 1 OF OUR

2179
01:22:55,914 --> 01:22:58,116
HIGHEST PERMIABILITYS AND WE

2180
01:22:58,116 --> 01:23:00,285
FIND THAT IN CHLOROFORM.

2181
01:23:00,285 --> 01:23:01,586
THERE'S A SLIGHTLY HIGHER

2182
01:23:01,586 --> 01:23:02,787
PERCENTAGE OF COLLAPSE

2183
01:23:02,787 --> 01:23:04,823
CONFIRMATION, BY SLIGHTLY HIGHER

2184
01:23:04,823 --> 01:23:06,958
I MEAN 51%, SO IT'S ABOUT EVEN

2185
01:23:06,958 --> 01:23:08,927
BETWEEN COLLAPSE AND SALMON AND

2186
01:23:08,927 --> 01:23:10,262
ELONGATED IN BLUE.

2187
01:23:10,262 --> 01:23:11,763
AND REMEMBER, CHLOROFORM IS

2188
01:23:11,763 --> 01:23:14,266
APPROXIMATING THE MEMBRANE HERE.

2189
01:23:14,266 --> 01:23:17,736
IN CONTRAST WHEN WE LOOK AT

2190
01:23:17,736 --> 01:23:20,305
COMPOUND 2 WHICH IS NOTABLY

2191
01:23:20,305 --> 01:23:21,840
SIMILAR TO COMPOUND 1 EXCEPT FOR

2192
01:23:21,840 --> 01:23:22,741
THIS 1 R-GROUP.

2193
01:23:22,741 --> 01:23:25,243
COMPOUND 2 HAS MUCH LOWER

2194
01:23:25,243 --> 01:23:26,578
PERMIABILITY AND IN CONTRAST

2195
01:23:26,578 --> 01:23:29,214
ONLY 43% OF THE CONFIRMATIONS

2196
01:23:29,214 --> 01:23:29,547
ARE COLLAPSED.

2197
01:23:29,547 --> 01:23:32,784
SO, IT'S SAYING THAT IN THE

2198
01:23:32,784 --> 01:23:34,653
MEMBRANE, COMPOUND 2 WHICH HAS A

2199
01:23:34,653 --> 01:23:35,820
LOWER PERMIABILITY ISN'T FORMING

2200
01:23:35,820 --> 01:23:38,857
AS MANY OF THOSE COLLAPSE

2201
01:23:38,857 --> 01:23:44,796
CONFIRMATIONS AS THE MORE

2202
01:23:44,796 --> 01:23:45,330
PERMEABLE COMPOUND 1.

2203
01:23:45,330 --> 01:23:50,101
SO WE ALSO KNOW THAT INTERNAL

2204
01:23:50,101 --> 01:23:51,369
[INDISCERNIBLE] HYDROGEN BONDS

2205
01:23:51,369 --> 01:23:52,904
ARE ASSOCIATE WIDE INCREASE IN

2206
01:23:52,904 --> 01:23:53,238
PERMIABILITY.

2207
01:23:53,238 --> 01:23:56,641
SO THIS PAPER SHOWS THAT IN THE

2208
01:23:56,641 --> 01:23:58,043
SOLVENTS 2 INTERIN LECULAR

2209
01:23:58,043 --> 01:23:59,177
HYDROGEN BONDS FORM AND ALLOW

2210
01:23:59,177 --> 01:24:02,213
FOR A MORE COLLAPSED STATE WITH

2211
01:24:02,213 --> 01:24:04,082
A LOWER 3D POLAR SURFACA AREA.

2212
01:24:04,082 --> 01:24:06,451
SO WHEN WE LOOKED AT THE

2213
01:24:06,451 --> 01:24:08,453
FORMATION OF HYDROGEN BONDS

2214
01:24:08,453 --> 01:24:10,255
WITHIN THE METASTABLE STATES OF

2215
01:24:10,255 --> 01:24:11,890
THE PERMEABLE AND LESS PERMEABLE

2216
01:24:11,890 --> 01:24:13,224
COMPOUND THAT I JUST TALKED

2217
01:24:13,224 --> 01:24:17,162
ABOUT, WE DID FIND THAT THE MORE

2218
01:24:17,162 --> 01:24:18,697
PERMEABLE COMPOUND HAS MORE

2219
01:24:18,697 --> 01:24:20,098
HYDROGEN BONDS FORM NOTHING ITS

2220
01:24:20,098 --> 01:24:20,598
COLLAPSED STATE.

2221
01:24:20,598 --> 01:24:22,734
SO THAT MEANS THAT THE MORE

2222
01:24:22,734 --> 01:24:24,736
PERMEABLE COMPOUND IS ABLE TO

2223
01:24:24,736 --> 01:24:26,938
FORM THOSE INTRA MOLECULAR

2224
01:24:26,938 --> 01:24:28,473
HYDROGEN BONDS LEADING TO A MORE

2225
01:24:28,473 --> 01:24:30,909
COLLAPSED STATE WHERE THE LESS

2226
01:24:30,909 --> 01:24:38,316
PERMEABLE COP POUND ISN'T ABLE

2227
01:24:38,316 --> 01:24:38,516
TO.

2228
01:24:38,516 --> 01:24:40,485
AND AGAIN HERE ARE MORE BLOBS

2229
01:24:40,485 --> 01:24:42,520
THAT DON'T MEAN ANYTHING EXCEPT

2230
01:24:42,520 --> 01:24:43,488
THAT THESE REPRESENT THE SEND

2231
01:24:43,488 --> 01:24:45,390
ROADWAYS WE ARE SEEING IN SOME

2232
01:24:45,390 --> 01:24:50,996
OF THESE METASTABLE STATES FROM

2233
01:24:50,996 --> 01:24:52,297
THE MARKOV AND STATE ANALYSIS,

2234
01:24:52,297 --> 01:24:54,699
AND WE SEE THAT KEY HYDROGEN

2235
01:24:54,699 --> 01:24:55,967
BONDS ARE FORMING IN THE

2236
01:24:55,967 --> 01:24:57,836
COLLAPSED STATES SO THIS DOESN'T

2237
01:24:57,836 --> 01:24:58,670
-- THIS ISN'T LIKE REVOLUTIONARY

2238
01:24:58,670 --> 01:25:01,740
RIGHT HERE AND WE'RE SEEING THAT

2239
01:25:01,740 --> 01:25:04,075
THE WE ARE SEEING HYDROGEN BONDS

2240
01:25:04,075 --> 01:25:07,112
FORM WHEN WE LOOK AT THE

2241
01:25:07,112 --> 01:25:09,581
COLLAPSED FORM OF THE PROTACS.

2242
01:25:09,581 --> 01:25:12,183
AND THEN WE WORKED WITH OUR

2243
01:25:12,183 --> 01:25:15,920
EXPERIMENTAL COLLEAGUES WHERE WE

2244
01:25:15,920 --> 01:25:20,692
EVALUATED THE SAME COMPOUNDS IN

2245
01:25:20,692 --> 01:25:22,160
WATER AND CHLOROFORM, WE FOUND

2246
01:25:22,160 --> 01:25:24,963
THAT IN DSO, THERE ARE FEWER

2247
01:25:24,963 --> 01:25:25,864
HYDROGEN BONDS FORMED, WHICH IS

2248
01:25:25,864 --> 01:25:27,832
NOT SURPRISING BUT IT DOES MAKE

2249
01:25:27,832 --> 01:25:29,234
US MORE CONFIDENT IN THE

2250
01:25:29,234 --> 01:25:30,135
MODELING RESULTS WHICH SHOWED

2251
01:25:30,135 --> 01:25:31,503
THE SAME THING.

2252
01:25:31,503 --> 01:25:34,205
INCLUDING ACTUALLY THE SAME

2253
01:25:34,205 --> 01:25:36,074
INTRA MOLECULAR HYDROGEN BONDS,

2254
01:25:36,074 --> 01:25:37,509
SO I WILL WALK THROUGH THIS

2255
01:25:37,509 --> 01:25:39,577
FIGURE WITH YOU, AGAIN I HAD TO

2256
01:25:39,577 --> 01:25:40,612
REMOVAL THE COMPOUND INFORMATION

2257
01:25:40,612 --> 01:25:42,447
SO I SKETCHED IT AS BEST AS I

2258
01:25:42,447 --> 01:25:48,553
COULD AND THE ABCD JUST REFER TO

2259
01:25:48,553 --> 01:25:51,389
SPECIFIC ATOMS AND THE NUMBER

2260
01:25:51,389 --> 01:25:52,624
THERE ARE TEMPERATURE CO

2261
01:25:52,624 --> 01:25:54,392
EFFICIENT AND GREEN MEANS MORE

2262
01:25:54,392 --> 01:25:55,560
POSITIVE, RED MEANS MORE

2263
01:25:55,560 --> 01:25:57,862
NEGATIVE AND THE MORE POSITIVE

2264
01:25:57,862 --> 01:25:58,696
TEMPERATURE CO EFFICIENT IS

2265
01:25:58,696 --> 01:26:00,899
ASSOCIATE WIDE A STRONGER

2266
01:26:00,899 --> 01:26:02,434
MOLECULAR HYDROGEN BOND OR

2267
01:26:02,434 --> 01:26:04,769
INCREASED SHIELDING, SO ON THE

2268
01:26:04,769 --> 01:26:07,405
TOP THERE IN THE -- WITH THE

2269
01:26:07,405 --> 01:26:13,578
NUMBER, THE NMR RESULTS IN DMSO,

2270
01:26:13,578 --> 01:26:19,984
AND CDCL IS BOTTOM RIGHT IS

2271
01:26:19,984 --> 01:26:22,187
MD-SIMULATION CHLOROFORM

2272
01:26:22,187 --> 01:26:23,755
COLLAPSED SIN THROID SO YOU CAN

2273
01:26:23,755 --> 01:26:27,325
SEE THAT THE MOLECULAR BONDS

2274
01:26:27,325 --> 01:26:30,662
FROM NMR, ARE ARE IN THE

2275
01:26:30,662 --> 01:26:31,729
ENSEMBLES AND THERE ARE MORE

2276
01:26:31,729 --> 01:26:34,666
BONDS IN CHLOROFORM AND CDCL3

2277
01:26:34,666 --> 01:26:35,867
THAN IN WATER SORE BOTH IN

2278
01:26:35,867 --> 01:26:39,471
EXPERIMENT AND THE MSM

2279
01:26:39,471 --> 01:26:39,771
ENSEMBLES.

2280
01:26:39,771 --> 01:26:41,506
AND THIS GAVE US MORE CONFIDENCE

2281
01:26:41,506 --> 01:26:44,375
THAT WHAT WE'RE MODELING IS

2282
01:26:44,375 --> 01:26:46,578
AGREEING WITH EXPERIMENT MEANING

2283
01:26:46,578 --> 01:26:48,546
WE CAN USE THIS INFORMATION

2284
01:26:48,546 --> 01:26:50,048
WITHOUT USING NMR DATA IN THE

2285
01:26:50,048 --> 01:26:51,616
FUTURE AND WE CAN JUST USE

2286
01:26:51,616 --> 01:26:52,951
MODELING AND HAVE GOOD

2287
01:26:52,951 --> 01:26:53,785
CONFIDENCE IN THIS.

2288
01:26:53,785 --> 01:26:55,019
AND THIS GENERALIZED TO A NUMBER

2289
01:26:55,019 --> 01:26:56,721
OF THE COMPOUNDS WE ARE LOOKING

2290
01:26:56,721 --> 01:26:57,856
AT, IT WASN'T JUST THIS 1

2291
01:26:57,856 --> 01:27:00,391
EXAMPLE BUT I FIGURED, YOU ALL

2292
01:27:00,391 --> 01:27:07,832
DIDN'T NEED TO SEE MORE BLOBS OF

2293
01:27:07,832 --> 01:27:08,166
COMPOUNDS.

2294
01:27:08,166 --> 01:27:10,034
OKAY, SO, THIS ANALYSIS WAS  

2295
01:27:10,034 --> 01:27:11,903
PERFORMED ACTUALLY ON SHOR

2296
01:27:11,903 --> 01:27:13,204
SHORTER MICROSECOND LONG

2297
01:27:13,204 --> 01:27:14,472
SIMMULESS AND CLUSTERED WITH,

2298
01:27:14,472 --> 01:27:16,274
YOU KNOW JUST STANDARD

2299
01:27:16,274 --> 01:27:17,475
CLUSTERING METHODS, BUT

2300
01:27:17,475 --> 01:27:19,577
INTERESTINGLY, WE DIDN'T SEE ANY

2301
01:27:19,577 --> 01:27:20,411
CORRELATION OF PERMIABILITY, AND

2302
01:27:20,411 --> 01:27:24,782
WE DIDN'T SEE ANY OF THE INTRA

2303
01:27:24,782 --> 01:27:26,217
MOLECULAR HYDROGEN BONDS

2304
01:27:26,217 --> 01:27:28,153
SUPPORTING THE NMR RESULTS, AND

2305
01:27:28,153 --> 01:27:30,288
SURE THOSE INTRA MOLECULAR

2306
01:27:30,288 --> 01:27:31,589
HYDROGEN BONDS APPEARED AND IT

2307
01:27:31,589 --> 01:27:33,791
WAS KIND OF -- DIDN'T CORRELATE

2308
01:27:33,791 --> 01:27:37,228
WITH ANYTHING, WE DIDN'T PULL

2309
01:27:37,228 --> 01:27:39,430
OUT ANY CORRELATION AND SEEMS

2310
01:27:39,430 --> 01:27:42,000
SAMPLING WITH SAMPLING AND

2311
01:27:42,000 --> 01:27:44,068
LONGER DETAILS, AND ALSO

2312
01:27:44,068 --> 01:27:46,504
INTERPRETING SIMULATION DATA

2313
01:27:46,504 --> 01:27:48,106
WITH MSNs, PROVIDED MORE

2314
01:27:48,106 --> 01:27:49,674
INFORMATION FOR ANALYSIS AND

2315
01:27:49,674 --> 01:27:50,742
COMPOUND DESIGN, SO IT SEEMED

2316
01:27:50,742 --> 01:27:56,915
PRETTY NECESSARY TO BE ABLE TO

2317
01:27:56,915 --> 01:27:58,449
INTERPRET THAT DATA.

2318
01:27:58,449 --> 01:27:59,617
SO FULL, OUR ONGOING EFFORTS

2319
01:27:59,617 --> 01:28:01,419
RIGHT NOW ARE THAT WE'RE USING A

2320
01:28:01,419 --> 01:28:07,325
DATA SET OF THE 135 IRAC PROTAC

2321
01:28:07,325 --> 01:28:11,062
AND DOING THIS WORK FLOW ON ALL

2322
01:28:11,062 --> 01:28:13,331
OF THEM AND YOU KNOW THAT'S

2323
01:28:13,331 --> 01:28:15,466
GOING TO BE PUBLICLY SHAREABLE

2324
01:28:15,466 --> 01:28:16,668
WHICH IS EXCITING BUT WE'RE ALSO

2325
01:28:16,668 --> 01:28:18,603
DOING THE SAME THING ON THE

2326
01:28:18,603 --> 01:28:21,639
INTERNAL SIMILAR NUMBER OF

2327
01:28:21,639 --> 01:28:22,707
PROTACS AND THE NICE PART ABOUT

2328
01:28:22,707 --> 01:28:24,842
THAT IS WE CAN ASSESS IT ON 2

2329
01:28:24,842 --> 01:28:26,211
VERY DIFFERENT SERIES AND SEE

2330
01:28:26,211 --> 01:28:27,512
FOR VERY DIFFERENT TARGETS ANDEE

2331
01:28:27,512 --> 01:28:31,549
IF THIS IS GENERALIZABLE.

2332
01:28:31,549 --> 01:28:33,751
NOTABLY WE'RE ALSO WORKING ON

2333
01:28:33,751 --> 01:28:34,719
IMPROVING THE PERMIABILITY

2334
01:28:34,719 --> 01:28:36,154
SCORE, SO I DIDN'T TALK TOO MUCH

2335
01:28:36,154 --> 01:28:37,989
ABOUT THAT TODAY, BUT WHAT I

2336
01:28:37,989 --> 01:28:40,325
MEANT BY PERMIABILITY CORE IS WE

2337
01:28:40,325 --> 01:28:43,628
LOOKED AT THESE RATIOS OF

2338
01:28:43,628 --> 01:28:46,598
CONFIRMATIONS AND PERCENTAGES OF

2339
01:28:46,598 --> 01:28:47,865
ELONGATED VERSUS COLLAPSE AND

2340
01:28:47,865 --> 01:28:50,268
KIND OF A RATIO OF RATIOS, THIS

2341
01:28:50,268 --> 01:28:52,437
IS GREAT, IT WILL BE NICE IF IT

2342
01:28:52,437 --> 01:28:54,839
WAS SIMPLE, IT'S PROBABLY NOT

2343
01:28:54,839 --> 01:28:56,674
THAT SIMPLE, IT'S PROBABLY MUCH,

2344
01:28:56,674 --> 01:28:58,109
MUCH, MORE COMPLEX, AND SO, YOU

2345
01:28:58,109 --> 01:28:58,943
KNOW THE FIRST THINK THIS WE

2346
01:28:58,943 --> 01:29:01,980
NEED TO LOCK AT IS, ARE 2 STATES

2347
01:29:01,980 --> 01:29:04,182
REALLY ENOUGH, COLLAPSED VERSUS

2348
01:29:04,182 --> 01:29:04,515
ELONGATED.

2349
01:29:04,515 --> 01:29:05,883
MY GUESS IS NO, BUT WE'RE

2350
01:29:05,883 --> 01:29:07,285
STARTING TO WITH THE 2 STATE

2351
01:29:07,285 --> 01:29:09,120
THEORY AND SEE HOW FAR WE CAN

2352
01:29:09,120 --> 01:29:10,355
GET WITH THIS.

2353
01:29:10,355 --> 01:29:11,689
WE'RE ALSO DOING COMPARISON OF

2354
01:29:11,689 --> 01:29:13,458
ALL THESE MATCHED PAIRS WITH

2355
01:29:13,458 --> 01:29:17,028
HIGH AND LOW PERMIABILITY, I

2356
01:29:17,028 --> 01:29:17,929
JUST MENTIONED 1 TODAY, AND LIKE

2357
01:29:17,929 --> 01:29:19,130
I SAID, YOU KNOW WE'RE

2358
01:29:19,130 --> 01:29:21,299
INTERESTED IN DOING SOME

2359
01:29:21,299 --> 01:29:25,370
GENERALIZABILITY ACROSS OTHER

2360
01:29:25,370 --> 01:29:26,037
SERIES AS WELL.

2361
01:29:26,037 --> 01:29:27,338
AND ANOTHER IMPORTANT ASPECT

2362
01:29:27,338 --> 01:29:29,607
THAT I DIDN'T TOUCH ON TOO MUCH

2363
01:29:29,607 --> 01:29:31,442
TODAY IS THE REASON WE'RE SO

2364
01:29:31,442 --> 01:29:34,078
INTERESTED IN PERMIABILITY IS

2365
01:29:34,078 --> 01:29:35,013
BECAUSE OF THE BIOAVAILABILITY

2366
01:29:35,013 --> 01:29:39,150
ULTIMATELY, THAT'S WHAT THE GOAL

2367
01:29:39,150 --> 01:29:41,686
IS AND WE KNOW JUST BECAUSE A

2368
01:29:41,686 --> 01:29:43,087
PROTAC DIDN'T HAVE HIGH

2369
01:29:43,087 --> 01:29:44,756
PERMIABILITY, MAYBE IT EVEN HAS

2370
01:29:44,756 --> 01:29:47,925
LOW PERMIABILITY, WE DO GET

2371
01:29:47,925 --> 01:29:50,128
COMPOUNDS WITH LOW PERMIABILITY,

2372
01:29:50,128 --> 01:29:50,895
TO HAVE BIOAVAILABILITY.

2373
01:29:50,895 --> 01:29:52,096
THIS IS A COMPLEX ISSUE,

2374
01:29:52,096 --> 01:29:53,731
HOWEVER, IF YOU GET INTO A

2375
01:29:53,731 --> 01:29:55,900
PERMEABLE SPACE, IT CAN USUALLY

2376
01:29:55,900 --> 01:29:58,603
HELP GET INTO A HIGHER

2377
01:29:58,603 --> 01:29:59,070
BIOAVAILABILITY SPACE?

2378
01:29:59,070 --> 01:30:01,639
SO HOW DO WE USE ALL OF THIS

2379
01:30:01,639 --> 01:30:04,642
INFORMATION TO DESIGN PROTAC

2380
01:30:04,642 --> 01:30:05,543
WITH HIGHER PERMIABILITY, DO YOU

2381
01:30:05,543 --> 01:30:09,447
WANT TO DESIGN AN OPTIMAL INTRA

2382
01:30:09,447 --> 01:30:10,648
MOLECULAR HYDROGEN BONDS AND ON

2383
01:30:10,648 --> 01:30:13,351
THAT NOTE, WE ONLY TALKED ABOUT

2384
01:30:13,351 --> 01:30:15,153
INTRA MOLECULAR HYDROGEN BONDS

2385
01:30:15,153 --> 01:30:17,021
TODAY, BUT THERE'S SO MUCH DATA

2386
01:30:17,021 --> 01:30:19,857
IN THE MARKERS THAT WE CAN USE

2387
01:30:19,857 --> 01:30:24,829
TO EXPLORE THESE PERMEABLE AND

2388
01:30:24,829 --> 01:30:26,197
IMPERMEABLE PROTACS, WE'RE JUST

2389
01:30:26,197 --> 01:30:27,165
SCRATCHING THE SURFACE HERE, SO

2390
01:30:27,165 --> 01:30:30,168
WITH THAT, I WANT TO THANK ALL

2391
01:30:30,168 --> 01:30:36,574
OF MY COLLABORATORS, J&  J,
AND

2392
01:30:36,574 --> 01:30:39,277
IN PARTICULAR I WANT TO THANK

2393
01:30:39,277 --> 01:30:40,511
[INDISCERNIBLE] HE'S A CURRENT

2394
01:30:40,511 --> 01:30:42,413
RESEARCH FELLOW AT THE FLAT IRON

2395
01:30:42,413 --> 01:30:44,582
INSTITUTE, HE WAS A Y&  J
SUMMER

2396
01:30:44,582 --> 01:30:47,618
INTERN AND HE DID HIS Ph.D. AT

2397
01:30:47,618 --> 01:30:49,787
UCSD WITH [INDISCERNIBLE], AND

2398
01:30:49,787 --> 01:30:54,092
HE IS STILL WORKING WITH US AND

2399
01:30:54,092 --> 01:30:55,893
WE HAVE SOME HOPEFULLY RESECTION

2400
01:30:55,893 --> 01:30:57,762
SIGHTING RESULTS COMING OUT IN

2401
01:30:57,762 --> 01:30:59,163
THE NEXT SEVERAL MONTHS, SO,

2402
01:30:59,163 --> 01:31:03,434
WITH THAT I WOULD LIKE TO THANK

2403
01:31:03,434 --> 01:31:06,104
YOU FOR YOUR TEAPGZ AND I CAN

2404
01:31:06,104 --> 01:31:06,537
TAKE ANY QUESTIONS.

2405
01:31:06,537 --> 01:31:07,071
>> THANK YOU VERY MUCH.

2406
01:31:07,071 --> 01:31:11,075
I THINK THIS IS GREAT

2407
01:31:11,075 --> 01:31:13,911
PRESENTATION AND REALLY GOOD

2408
01:31:13,911 --> 01:31:15,012
REVIEW OF THE PRODUCT.

2409
01:31:15,012 --> 01:31:16,681
SO WE HAVE SEVERAL QUESTIONS IN

2410
01:31:16,681 --> 01:31:21,018
THE CHAT, LET ME START WITH 1,

2411
01:31:21,018 --> 01:31:22,854
THE FIRST 1.

2412
01:31:22,854 --> 01:31:24,789
HAVE YOU TRIED DIRECTLY

2413
01:31:24,789 --> 01:31:25,623
SIMULATING THE PASSAGE OF

2414
01:31:25,623 --> 01:31:34,332
PRODUCTS FROM THE MEMBRANE.

2415
01:31:34,332 --> 01:31:34,866
>> YES.

2416
01:31:34,866 --> 01:31:35,299
NO, WE HAVEN'T.

2417
01:31:35,299 --> 01:31:37,568
BUT WE THOUGHT ABOUT IT THOUGH,

2418
01:31:37,568 --> 01:31:38,736
WE THOUGHT ABOUT PULLING A

2419
01:31:38,736 --> 01:31:39,937
PROTAC OR A MEMBRANE FOR

2420
01:31:39,937 --> 01:31:41,439
EXAMPLE, BUT WE HAVEN'T DONE IT,

2421
01:31:41,439 --> 01:31:42,540
I'M CURIOUS TO DO IT.

2422
01:31:42,540 --> 01:31:47,745
I THINK IT WOULD BE HARDER TO DO

2423
01:31:47,745 --> 01:31:48,780
ON A LARGER SCALE.

2424
01:31:48,780 --> 01:31:51,015
I WAS HOPING TO SCALE THIS UP TO

2425
01:31:51,015 --> 01:31:52,150
HUNDREDS OF PROTACS, I DON'T

2426
01:31:52,150 --> 01:31:54,786
KNOW HOW I WOULD SCALE THE POLE

2427
01:31:54,786 --> 01:31:58,990
ANCHOR BUT I GUESS WE COULD.

2428
01:31:58,990 --> 01:32:01,426
YEAH, SO, -- THAT'S A GREAT

2429
01:32:01,426 --> 01:32:01,626
IDEA.

2430
01:32:01,626 --> 01:32:04,595
WE HAVE DONE SOME OTHER WORK

2431
01:32:04,595 --> 01:32:06,831
WITH OTHER TOOLS IN MEMBRANES,

2432
01:32:06,831 --> 01:32:12,870
THAT DIDN'T REALLY WORK OUT TOO

2433
01:32:12,870 --> 01:32:13,104
WELL.

2434
01:32:13,104 --> 01:32:15,973
NNOW FOR ANOTHER QUESTION,

2435
01:32:15,973 --> 01:32:17,775
[INDISCERNIBLE] FOR THE

2436
01:32:17,775 --> 01:32:19,477
DIFFERENT TETRAMERS, DO YOU HAVE

2437
01:32:19,477 --> 01:32:20,545
THIS IDEA?

2438
01:32:20,545 --> 01:32:25,750
>> COULD YOU REPEAT THAT?

2439
01:32:25,750 --> 01:32:30,054
OH, I SEE.

2440
01:32:30,054 --> 01:32:34,759
THAT'S A GREAT QUESTION.

2441
01:32:34,759 --> 01:32:41,966
WE WE HAVE THOUGHT ABOUT THIS.

2442
01:32:41,966 --> 01:32:45,069
WE'VE ALSO IGNORED IT TO SOME

2443
01:32:45,069 --> 01:32:49,240
EXTENT, SO THAT'S DEFINITELY A

2444
01:32:49,240 --> 01:32:50,007
SCAPEGOAT ANSWER.

2445
01:32:50,007 --> 01:32:58,382
BUT NO, IT'S A GOOD QUESTION.

2446
01:32:58,382 --> 01:33:01,285
WE BASICALLY HAVEN'T ADDRESSED

2447
01:33:01,285 --> 01:33:01,586
IT.

2448
01:33:01,586 --> 01:33:02,854
WE UNDERSTAND IT'S AN ISSUE AND

2449
01:33:02,854 --> 01:33:03,921
SOMETHING TO CONSIDER BUT WE

2450
01:33:03,921 --> 01:33:07,158
HAVE NOT WORKED ON THAT

2451
01:33:07,158 --> 01:33:08,860
PARTICULAR ISSUE.

2452
01:33:08,860 --> 01:33:13,464
>> FOR THE TETRAMERS.

2453
01:33:13,464 --> 01:33:13,898
>> EXACTLY.

2454
01:33:13,898 --> 01:33:15,433
>> THE OTHER QUESTION IS
WHAT IS

2455
01:33:15,433 --> 01:33:19,103
THE DISTANCE IN, HAVE YOU DONE

2456
01:33:19,103 --> 01:33:21,205
STUDIES WITH MOLECULAR

2457
01:33:21,205 --> 01:33:22,340
FLEXIBILITY DESCRIPTORS TO

2458
01:33:22,340 --> 01:33:25,309
CURATE YOUR RESULS, HAVE YOU

2459
01:33:25,309 --> 01:33:26,911
TRIED ANY MEDICAL FLEXIBILITY

2460
01:33:26,911 --> 01:33:30,882
DESCRIPTORS TO MAKE A

2461
01:33:30,882 --> 01:33:35,920
PREDICTION?

2462
01:33:35,920 --> 01:33:37,255
>> SO MOLECULAR
FLEXIBILITY

2463
01:33:37,255 --> 01:33:38,623
DESCRIPTORS, ARE WE TALKING

2464
01:33:38,623 --> 01:33:40,358
ABOUT RSD OR RADIO GENERATION OR

2465
01:33:40,358 --> 01:33:41,559
THESE SORTS OF THINGS?

2466
01:33:41,559 --> 01:33:43,694
>> HAVE YOU TRIED ANY OF
THOSE?

2467
01:33:43,694 --> 01:33:47,832
>> YEAH, WE TRIED ALL OF
THEM.

2468
01:33:47,832 --> 01:33:51,269
WE TRIED PROBABLY 2 DOZEN

2469
01:33:51,269 --> 01:33:52,503
DIFFERENT MOLECULAR FLEXIBILITY

2470
01:33:52,503 --> 01:33:53,137
DESCRIPTORS, BASICALLY ANYTHING

2471
01:33:53,137 --> 01:33:55,039
WE COULD THINK OF AND THIS IS

2472
01:33:55,039 --> 01:33:55,973
LIKE INITIAL WORK, RIGHT WHERE

2473
01:33:55,973 --> 01:33:58,175
WE'RE TRYING TO CORRELATE

2474
01:33:58,175 --> 01:33:59,644
BASICALLY ANYTHING, ANY

2475
01:33:59,644 --> 01:34:03,247
DECRYPTER WITH -- TO SEE IF IT

2476
01:34:03,247 --> 01:34:04,649
WOULD CORRELATE, AND THEN THE

2477
01:34:04,649 --> 01:34:10,521
MOST RELIABLE ACROSS ALL THE

2478
01:34:10,521 --> 01:34:10,788
PROTACS.

2479
01:34:10,788 --> 01:34:11,122
>> GOTCHA.

2480
01:34:11,122 --> 01:34:14,091
ONE MORE QUESTION, HOW COULD 1

2481
01:34:14,091 --> 01:34:16,527
DETERMINE IF THE MD SIMULATION

2482
01:34:16,527 --> 01:34:18,863
[INDISCERNIBLE] ENOUGH FOR THE

2483
01:34:18,863 --> 01:34:19,130
ANALYSIS?

2484
01:34:19,130 --> 01:34:20,831
>> THAT'S A GREAT
QUESTION.

2485
01:34:20,831 --> 01:34:21,699
DEPENDS ON YOUR SYSTEM.

2486
01:34:21,699 --> 01:34:23,668
AND YOU KNOW I THINK THE

2487
01:34:23,668 --> 01:34:25,169
GENERALLY, THE MORE DATA THE

2488
01:34:25,169 --> 01:34:29,340
BETTER, I'M NOT NECESSARILY THE

2489
01:34:29,340 --> 01:34:36,914
MSM EXPERTS AND THERE ARE MANY

2490
01:34:36,914 --> 01:34:37,949
WAYS TO ANALYZE THAT.

2491
01:34:37,949 --> 01:34:39,550
FOR THIS PARTICULAR SYSTEM, SO

2492
01:34:39,550 --> 01:34:42,153
WE WORKED WITH MSMs IN GRAD

2493
01:34:42,153 --> 01:34:45,489
SCHOOL AND LARGER PROTEINS FOR

2494
01:34:45,489 --> 01:34:47,992
THIS SMALLER SYSTEM WHICH IS

2495
01:34:47,992 --> 01:34:50,895
REALLY JUST, YOU KNOW, THESE

2496
01:34:50,895 --> 01:34:53,397
AREN'T HUGE, THEY'RE LARGE SMALL

2497
01:34:53,397 --> 01:34:54,732
MOLECULES BUT THEY'RE NOT GIANT.

2498
01:34:54,732 --> 01:34:56,367
IT SEEMS THAT A COUPLE

2499
01:34:56,367 --> 01:34:57,969
MICROSECONDS IS GOOD ENOUGH FOR

2500
01:34:57,969 --> 01:35:02,206
THESE, BUT LIKE I SAID, SO THIS

2501
01:35:02,206 --> 01:35:04,542
IS 2 DIFFERENT QUESTIONS, SORRY

2502
01:35:04,542 --> 01:35:05,476
I'M GETTING INTO 2 DIFFERENT

2503
01:35:05,476 --> 01:35:06,611
THINGS, WOKNOW IS HOW LONG THESE

2504
01:35:06,611 --> 01:35:08,379
CAN BE FOR THE SAME AMOUNT OF

2505
01:35:08,379 --> 01:35:09,480
THE BUILDING, JUST TOTALLY

2506
01:35:09,480 --> 01:35:11,415
DEPENDS ON THE SYSTEM, RIGHT?

2507
01:35:11,415 --> 01:35:12,516
FOR THIS PARTICULAR SYSTEM WHICH

2508
01:35:12,516 --> 01:35:13,818
I'M GOING TO ANSWER THE QUESTION

2509
01:35:13,818 --> 01:35:16,187
FOR, IT SEEMS TO BE THAT WE NEED

2510
01:35:16,187 --> 01:35:17,989
MORE THAN A MICROSECOND EACH

2511
01:35:17,989 --> 01:35:18,889
THOUGH THAT SEEMS QUITE LONG TO

2512
01:35:18,889 --> 01:35:20,424
ME BUT IT SEEMS TO GET ANY OF

2513
01:35:20,424 --> 01:35:21,759
THE NRGZ OUT THAT WE WANT TO,

2514
01:35:21,759 --> 01:35:26,597
YOU DO NEED MORE THAN A

2515
01:35:26,597 --> 01:35:27,031
MICROSECOND.

2516
01:35:27,031 --> 01:35:27,365
>> GOTCHA.

2517
01:35:27,365 --> 01:35:29,767
YEAH 1 QUESTION I WANT TO ASK

2518
01:35:29,767 --> 01:35:32,136
YOU, IT'S A BIT OFFTOPPIC BUT

2519
01:35:32,136 --> 01:35:33,004
THE PRODUCT HAS DIFFERENT

2520
01:35:33,004 --> 01:35:34,271
LINKERS AND DIFFERENT SIZE OF

2521
01:35:34,271 --> 01:35:35,940
LINKERS HAVE YOU LOOKEDDA THE

2522
01:35:35,940 --> 01:35:37,508
THAT PROBLEM, HAVE YOU USED KIND

2523
01:35:37,508 --> 01:35:40,811
OF LIKE MACHINE LEARNING FOR

2524
01:35:40,811 --> 01:35:42,279
INSILICO MINUTES DO YOU PREDICT

2525
01:35:42,279 --> 01:35:43,581
THAT BECAUSE SOMETIMES AFFINITY

2526
01:35:43,581 --> 01:35:46,417
IS [INDISCERNIBLE] AS WELL AS

2527
01:35:46,417 --> 01:35:46,817
ACUITY.

2528
01:35:46,817 --> 01:35:49,153
NYEAH, THE LINKER QUESTION,

2529
01:35:49,153 --> 01:35:50,721
YEAH, LINKERS MAKE A HUGE

2530
01:35:50,721 --> 01:35:51,322
DIFFERENCE, YOU'RE ABSOLUTELY

2531
01:35:51,322 --> 01:35:53,524
RIGHT, WE HAVE LOOKED AT IT IN

2532
01:35:53,524 --> 01:35:56,327
TERPS OF WE HAVE VERY RIGID

2533
01:35:56,327 --> 01:35:57,928
LINKERS AND FLEXIBLE LINKERS

2534
01:35:57,928 --> 01:35:59,163
THAT ARE THE SAME LENGTH SO

2535
01:35:59,163 --> 01:36:00,564
WE'RE TRYING TO CONTROL FOR

2536
01:36:00,564 --> 01:36:02,767
LENGTH OF LINKER BY LOOKING AT

2537
01:36:02,767 --> 01:36:03,834
RIGIDITY AND FLEXIBILITY BETWEEN

2538
01:36:03,834 --> 01:36:07,538
THE 2 AND YEAH, THE FLEXIBILITY

2539
01:36:07,538 --> 01:36:16,347
LINKERS ALLOWED FOR MORE COMPACT

2540
01:36:16,347 --> 01:36:19,016
PROTACS AND THEY CAN FOLD UP AND

2541
01:36:19,016 --> 01:36:20,451
THE REGULAR LINKERS CAN'T.

2542
01:36:20,451 --> 01:36:23,320
WE LOOKEDDA THE LINKER LENGTH,

2543
01:36:23,320 --> 01:36:25,489
TYPE, YEAH, THATIA A -- BUT YUF

2544
01:36:25,489 --> 01:36:28,659
BECAUSE A LINKER IS VERY

2545
01:36:28,659 --> 01:36:30,094
FLEXIBILITY DOESN'T NECESSARILY

2546
01:36:30,094 --> 01:36:33,631
MEAN IT WILL BE THE MOST

2547
01:36:33,631 --> 01:36:36,801
PERMIABILITY EITHER.

2548
01:36:36,801 --> 01:36:40,771
NOKAY.

2549
01:36:40,771 --> 01:36:42,540
SOUNDS GOOD.

2550
01:36:42,540 --> 01:36:43,507
SOMEONE ASKED ABOUT THE

2551
01:36:43,507 --> 01:36:45,876
[INDISCERNIBLE] OF YOUR MACHINE

2552
01:36:45,876 --> 01:36:47,478
LEARNING MODELS, IT'S POINTED

2553
01:36:47,478 --> 01:36:48,646
OUT [INDISCERNIBLE] AND THEY ASK

2554
01:36:48,646 --> 01:36:52,583
DID YOU TRY TO IMPROVE THAT

2555
01:36:52,583 --> 01:36:54,719
BEFORE TRY THE SIMULATION TO GET

2556
01:36:54,719 --> 01:36:54,952
INSIDE.

2557
01:36:54,952 --> 01:36:58,189
SO BASIC LYE YEAH, IT'S A MORE

2558
01:36:58,189 --> 01:37:03,160
GENERAL QUESTION WHEN YOU DO

2559
01:37:03,160 --> 01:37:04,729
MACHINE LEARNING AND THE SAME

2560
01:37:04,729 --> 01:37:05,830
PIPELINE, HAVE YOU TRIED TO

2561
01:37:05,830 --> 01:37:09,467
IMPROVE THE MACHINE LEARNING OR

2562
01:37:09,467 --> 01:37:11,669
IT'S TOTALLY FINE, LIKE WHATEVER

2563
01:37:11,669 --> 01:37:13,370
THE ACCEPTABLE METRICS FOR YOU,

2564
01:37:13,370 --> 01:37:16,540
IN IT CASE, IN TERMS OF

2565
01:37:16,540 --> 01:37:16,907
[INDISCERNIBLE].

2566
01:37:16,907 --> 01:37:18,008
>> YEAH, WE ARE ALWAYS
TRYING TO

2567
01:37:18,008 --> 01:37:22,513
IMPROVE THE MACHINE LEARNING

2568
01:37:22,513 --> 01:37:26,617
MODELS FOR PERMIABILITY WITH

2569
01:37:26,617 --> 01:37:33,324
THIS PARTICULAR 2 D MODEL THAT

2570
01:37:33,324 --> 01:37:37,528
THE R-SQUARED SCORE IS PRETTY

2571
01:37:37,528 --> 01:37:37,728
GOOD.

2572
01:37:37,728 --> 01:37:42,199
WE -- IT'S GOOD ENOUGH FOR OUR

2573
01:37:42,199 --> 01:37:46,003
--  WE HAVE A CUT-OFF BASICALLY

2574
01:37:46,003 --> 01:37:48,606
IF IT'S PREDICTED TO BE ABOUT A

2575
01:37:48,606 --> 01:37:50,741
CERTAIN KIND OF VALUE THEN WE

2576
01:37:50,741 --> 01:37:52,109
MOVE IT FORWARD IF IT'S BELOW,

2577
01:37:52,109 --> 01:37:53,844
WE GET RID OF IT AND WE DON'T

2578
01:37:53,844 --> 01:37:55,279
MOVE IT FORWARD AND WE

2579
01:37:55,279 --> 01:38:02,186
CONTINUOUSLY CHALLENGE THE

2580
01:38:02,186 --> 01:38:02,920
MODEL, TOO.

2581
01:38:02,920 --> 01:38:04,922
SO THIS MODEL'S PRETTY GOOD IS

2582
01:38:04,922 --> 01:38:06,390
BASICALLY WHAT I'M SAYING

2583
01:38:06,390 --> 01:38:07,191
HOWEVER, IF YOU THROW IN

2584
01:38:07,191 --> 01:38:10,561
SOMETHING THAT THE MODEL HAS

2585
01:38:10,561 --> 01:38:12,062
NEVER SEEP BEFORE, SOMETHING

2586
01:38:12,062 --> 01:38:13,297
VERY DIFFERENT, VERY HIGH

2587
01:38:13,297 --> 01:38:15,065
PERMIABILITY OR SOMETHING AND

2588
01:38:15,065 --> 01:38:17,334
MAYBE A STRUCTURE BASED CHANGE,

2589
01:38:17,334 --> 01:38:18,002
THAT DOESN'T SEEM LIKE IT'S

2590
01:38:18,002 --> 01:38:22,439
GOING TO BE A HUGE DIFFERENCE,

2591
01:38:22,439 --> 01:38:23,707
THE MODEL WON'T BE ABLE TO

2592
01:38:23,707 --> 01:38:25,476
PREDICT THAT VERY WELL AND WE'VE

2593
01:38:25,476 --> 01:38:26,944
SEEN THAT HAPPEN BEFORE AND

2594
01:38:26,944 --> 01:38:29,713
THAT'S BASICALLY WHY WE PIVOTED

2595
01:38:29,713 --> 01:38:32,116
TO ALSO USING THESE PHYSICS

2596
01:38:32,116 --> 01:38:33,717
BASED METHODS BECAUSE YOU JUST

2597
01:38:33,717 --> 01:38:39,223
CAN'T CAPTURE IT ALL WITH 2 D

2598
01:38:39,223 --> 01:38:40,357
INFORMATION.

2599
01:38:40,357 --> 01:38:42,359
NSOUNDS GOOD, THANK YOU VERY

2600
01:38:42,359 --> 01:38:44,361
MUCH AGAIN BRYN FOR THE GREAT

2601
01:38:44,361 --> 01:38:45,062
TALK AND PRESENTATION.

2602
01:38:45,062 --> 01:38:52,002
>> THANK YOU AND THANKS
FOR THE

2603
01:38:52,002 --> 01:38:53,237
GREAT QUESTIONS EVERYBODY.

2604
01:38:53,237 --> 01:38:58,409
NOW LET'S MEET SHERYL

2605
01:38:58,409 --> 01:38:59,710
ARROWSMITH, SENIOR SCIENCE

2606
01:38:59,710 --> 01:39:00,778
PROFESSOR MEDICAL BIOPHYSICS AT

2607
01:39:00,778 --> 01:39:02,713
THE UNIVERSITY OF TR RONT O AND

2608
01:39:02,713 --> 01:39:04,915
CHIEF SCIENTIST OF THE

2609
01:39:04,915 --> 01:39:05,749
STRUCTURAL GENOMICS CONSORTIUM

2610
01:39:05,749 --> 01:39:07,818
AT THE UNIVERSITY OF TORONTO.

2611
01:39:07,818 --> 01:39:10,588
SHE HAS PUBLISHED OVER 300

2612
01:39:10,588 --> 01:39:12,957
RESEARCH ARTICLES AND MEASURED

2613
01:39:12,957 --> 01:39:14,692
GUY ANALYTICS AS BEING AMONG THE

2614
01:39:14,692 --> 01:39:18,329
AWARDS TOP 1% OF HIGHLY CITED

2615
01:39:18,329 --> 01:39:21,031
SCIENTISTS IN 2018, 2019, 2022,

2616
01:39:21,031 --> 01:39:23,267
AND 2323.

2617
01:39:23,267 --> 01:39:25,803
SHE WAS ELECTED AS AAAS FELLOW

2618
01:39:25,803 --> 01:39:29,006
2015 AND FELLOW OF ROYAL SOCIETY

2619
01:39:29,006 --> 01:39:31,575
OF CANADA 2020.

2620
01:39:31,575 --> 01:39:33,410
SHERYL, PLEASE, THE STAGE IS

2621
01:39:33,410 --> 01:39:33,811
YOURS.

2622
01:39:33,811 --> 01:39:34,211
>> THANK YOU.

2623
01:39:34,211 --> 01:39:44,188
CAN EVERYBODY SEE AND HEAR?

2624
01:39:44,188 --> 01:39:44,855
>> YES,.

2625
01:39:44,855 --> 01:39:45,189
>> GREAT.

2626
01:39:45,189 --> 01:39:46,323
SO I'M VERY PLEASED TO TALK TO

2627
01:39:46,323 --> 01:39:48,559
YOU ABOUT THE WORK WE'RE DOING

2628
01:39:48,559 --> 01:39:52,596
ON DATA GENERATION IN SUPPORT OF

2629
01:39:52,596 --> 01:39:53,931
KOOSMUTATIONAL DRUG DISCOVERY.

2630
01:39:53,931 --> 01:39:58,903
SO THE SGC IS A PUBLIC

2631
01:39:58,903 --> 01:39:59,536
PARTNERSHIP OF DATA SCIENTISTS

2632
01:39:59,536 --> 01:40:02,239
AND A NUMBER OF PHARMACEUTICAL

2633
01:40:02,239 --> 01:40:03,173
COMPANIES AND LISTED BELOW HERE

2634
01:40:03,173 --> 01:40:07,845
AND ALL HAVE A COMMON INTEREST

2635
01:40:07,845 --> 01:40:09,480
AND PRETTY COMPETITIVE RELATED

2636
01:40:09,480 --> 01:40:11,115
TO RESEARCH DRUG DORPH.

2637
01:40:11,115 --> 01:40:16,287
SO IN THE CONTEXT OF

2638
01:40:16,287 --> 01:40:18,188
COMPUTATIONAL WORK AND THE POWER

2639
01:40:18,188 --> 01:40:20,457
OF ARTIFICIAL INTELIENCE AND

2640
01:40:20,457 --> 01:40:21,926
MACHINE LEARNING, I THINK THIS

2641
01:40:21,926 --> 01:40:25,396
YEAR'S NOBLE PRIZE IS IN PHYSICS

2642
01:40:25,396 --> 01:40:26,897
AND CHEMISTRY HIGHLIGHT THE

2643
01:40:26,897 --> 01:40:35,973
EXCITEMENT AND THE POWER OF

2644
01:40:35,973 --> 01:40:36,307
THESE METHODS.

2645
01:40:36,307 --> 01:40:39,310
IN PARTICULARLY THE AWARD IN

2646
01:40:39,310 --> 01:40:40,177
CHEMISTRY WAS AWARDING TO 50

2647
01:40:40,177 --> 01:40:41,045
YEAR-OLD CHALLENGE WHICH WAS

2648
01:40:41,045 --> 01:40:45,983
BASED ON THE DNA SEQUENCE OF

2649
01:40:45,983 --> 01:40:47,484
ENCODING A CHAIN AND PREDICTING

2650
01:40:47,484 --> 01:40:51,755
THE 3D STRUCTURE AND IT WAS IN

2651
01:40:51,755 --> 01:40:55,426
2021 THE GROUPS FROM DEEP MIND

2652
01:40:55,426 --> 01:41:00,230
AND DAVID BAKERS WORK IN PROTEIN

2653
01:41:00,230 --> 01:41:04,068
DESEEN SORT OF SIMULTANEOUSLY

2654
01:41:04,068 --> 01:41:07,271
SHOWED THAT USING PHYSICS BASED

2655
01:41:07,271 --> 01:41:09,573
AND AI DEEP LEARNING METHODS

2656
01:41:09,573 --> 01:41:12,609
THAT THIS COULD BE DONE.

2657
01:41:12,609 --> 01:41:14,211
SO, I THINK WE CAN TAKE SOME

2658
01:41:14,211 --> 01:41:15,980
LESSONS FROM THIS AND THIS IS

2659
01:41:15,980 --> 01:41:19,016
WHAT MY TALK, THE THEME OF MY

2660
01:41:19,016 --> 01:41:20,584
TALK WILL BE, SO WE THINK A

2661
01:41:20,584 --> 01:41:23,087
CHALLENGE IS GREAT AS THE

2662
01:41:23,087 --> 01:41:24,922
PROTEIN FOLDING PROBLEM IS THE

2663
01:41:24,922 --> 01:41:27,524
ABILITY TO PRODICT DRUG LIKE

2664
01:41:27,524 --> 01:41:29,293
COMPOUNDS THAT COMBINE TO END

2665
01:41:29,293 --> 01:41:31,261
MODULATE A GIVEN PROTEIN.

2666
01:41:31,261 --> 01:41:33,831
SO WE'RE DEFINITELY NOT THERE

2667
01:41:33,831 --> 01:41:33,998
YET.

2668
01:41:33,998 --> 01:41:37,534
THERE ARE SOME HINTS THAT EARLY

2669
01:41:37,534 --> 01:41:40,471
COMPUTATIONAL METHODS CAN

2670
01:41:40,471 --> 01:41:41,372
PREDICT SOME MOLECULES THAT BIND

2671
01:41:41,372 --> 01:41:44,274
TO I GIVEN TARGET BUT IT'S FAR

2672
01:41:44,274 --> 01:41:46,377
FROMENTIOUS SPECIALLY FOR

2673
01:41:46,377 --> 01:41:48,078
UNPRECEDENTED TARGETS THAT HAVE

2674
01:41:48,078 --> 01:41:49,613
NEVER BEEN LIGANDED BEFORE.

2675
01:41:49,613 --> 01:41:50,681
IT'S VERY FAR FROM SUCCESSFUL

2676
01:41:50,681 --> 01:41:51,215
RIGHT NOW.

2677
01:41:51,215 --> 01:41:53,017
SO HOW DO WE GET TO THIS POINT,

2678
01:41:53,017 --> 01:41:55,285
WE'RE AT THIS EARLY STAGE OF

2679
01:41:55,285 --> 01:41:59,023
DRUG DORPHY CAN BE LARGELY

2680
01:41:59,023 --> 01:42:00,357
COMPUTATIONAL AND TAKE SOME OF

2681
01:42:00,357 --> 01:42:02,426
THE BURDEN OFF OF THE EXPENSIVE

2682
01:42:02,426 --> 01:42:03,627
AND SLOW METHODS OF DRUG

2683
01:42:03,627 --> 01:42:05,763
DISCOVERY THAT HAVE TO BE DONE

2684
01:42:05,763 --> 01:42:06,196
IN THE LAB.

2685
01:42:06,196 --> 01:42:11,068
SO WE'RE GOING TO APPLY LESSONS

2686
01:42:11,068 --> 01:42:13,570
LEARNED FROM THE STRUCTURAL

2687
01:42:13,570 --> 01:42:15,439
BIOLOGY COMMUNITY THAT HELP TO

2688
01:42:15,439 --> 01:42:17,474
PROMOTE THE SUCCESS, I WILL TALK

2689
01:42:17,474 --> 01:42:20,744
ABOUT PILOT PROGGENTS THAT WE'VE

2690
01:42:20,744 --> 01:42:23,547
BEEN DOING TO ASSESS AI MACHINE

2691
01:42:23,547 --> 01:42:25,949
LEARNING ASSISTED PROTEIN

2692
01:42:25,949 --> 01:42:26,483
COMPOUND SCREENING MLGTS.

2693
01:42:26,483 --> 01:42:29,753
I'LL TALK ABOUT HOW TO ENABLE

2694
01:42:29,753 --> 01:42:36,160
THESE METHODS WITH OPEN CATAAND

2695
01:42:36,160 --> 01:42:37,928
PARTICULARLY HOW TO ARCHIVE,

2696
01:42:37,928 --> 01:42:39,730
STORE AND DISSEMINATE DATA AND

2697
01:42:39,730 --> 01:42:40,597
ALSO COMMUNITY BENCHMARK PRONG

2698
01:42:40,597 --> 01:42:43,467
ECTOMYOSINS WHICH ARE THE OTHER

2699
01:42:43,467 --> 01:42:45,436
-- PROJECTS WHICH ARE THE OTHER

2700
01:42:45,436 --> 01:42:47,304
KEY METHOD REQUIRED FOR KNOWING

2701
01:42:47,304 --> 01:42:54,845
WHEN YOU'VE BEEN SUCCESSFUL.

2702
01:42:54,845 --> 01:42:56,613
AND THEN IN THE END HOW THE

2703
01:42:56,613 --> 01:42:58,115
COMMUNITY CAN WORK TOWARDS THESE

2704
01:42:58,115 --> 01:42:59,850
GOALS MUCH SO WHAT DOES SUCCESS

2705
01:42:59,850 --> 01:43:01,552
LOOK LIKE, SO THE PROTEIN

2706
01:43:01,552 --> 01:43:02,586
FOLDING COMMUNITY KNEW THEY HAD

2707
01:43:02,586 --> 01:43:08,659
YOU KNOW METHODS THAT REALLY

2708
01:43:08,659 --> 01:43:10,527
WORKED EVENTUALLY AFTER 14

2709
01:43:10,527 --> 01:43:16,400
BIANNUAL COMMUNITY WIDE

2710
01:43:16,400 --> 01:43:18,135
COMPETITIONS CALLED CASP, THEY

2711
01:43:18,135 --> 01:43:19,736
FINALLY SHOWED THAT ALPHA FOLD 2

2712
01:43:19,736 --> 01:43:22,673
WAS VERY HIGHLY SUCCESSFUL IN

2713
01:43:22,673 --> 01:43:23,474
PREDICTING THE 3 DIMENSIONAL

2714
01:43:23,474 --> 01:43:33,283
FOLD OF THE PROTEIN.

2715
01:43:33,283 --> 01:43:37,721
AND IT WAS EXPERIMENTAL DAT AND

2716
01:43:37,721 --> 01:43:39,890
THE PROTEIN DATA BANK, THE PDB

2717
01:43:39,890 --> 01:43:41,892
AND TAKEN--THEY'S WHAT ERCH HAS

2718
01:43:41,892 --> 01:43:43,260
BEEN USING OVER MANY YEAR ANDS

2719
01:43:43,260 --> 01:43:49,967
FINALLY THERE WAS ENOUGH DATA,

2720
01:43:49,967 --> 01:43:52,936
THERE WAS ENOUGH 3 STRUCTURES IN

2721
01:43:52,936 --> 01:43:54,338
THE PDB ANDLET ABILITIES OF AI

2722
01:43:54,338 --> 01:43:55,439
COULD MAKE USE OF THAT EMPLOY SO

2723
01:43:55,439 --> 01:43:57,241
LET ME JUST GIVE YOU A LITTLE

2724
01:43:57,241 --> 01:44:00,744
BIT OF BACKGROUND ABOUT THE SGC,

2725
01:44:00,744 --> 01:44:05,082
AS OUR NAME SUGJEFFS STRUCTURAL

2726
01:44:05,082 --> 01:44:06,817
GENOMICS CONSORTIUM, WE STARTED

2727
01:44:06,817 --> 01:44:08,519
AS AN ORGANIZATIONS PIEZATION,

2728
01:44:08,519 --> 01:44:10,220
WE WERE AMONG THE LARNLEST

2729
01:44:10,220 --> 01:44:12,222
CONTRIBUTORS TO THE PDB FOR A

2730
01:44:12,222 --> 01:44:14,291
NUMBER OF YEARS AND ALSO

2731
01:44:14,291 --> 01:44:16,426
GREATEST SINGLE TRIBUTER TO THE

2732
01:44:16,426 --> 01:44:18,462
CASP COMPETITIONS WITH OVER A

2733
01:44:18,462 --> 01:44:23,367
HUNDRED PROTEINS IN THE MID2000S

2734
01:44:23,367 --> 01:44:24,868
EMPLOY AND AROUND 2010 OR SO, WE

2735
01:44:24,868 --> 01:44:27,471
STARTED USING THE PROTEINS THAT

2736
01:44:27,471 --> 01:44:30,741
WE MADE TO IDENTIFY BIOACTIVE

2737
01:44:30,741 --> 01:44:32,709
SMALL MOLECULES THAT BIND AND

2738
01:44:32,709 --> 01:44:35,812
MODULATE THESE PROTEINS WE CALL

2739
01:44:35,812 --> 01:44:38,182
THEM CHEMICAL PROBES AND

2740
01:44:38,182 --> 01:44:39,116
MORRECENTLY WE'VE STARTED TO

2741
01:44:39,116 --> 01:44:41,518
APPLY MACHINE LEARNING METHODS

2742
01:44:41,518 --> 01:44:45,255
TO TRY TO ACCELERATE THIS ENTIRE

2743
01:44:45,255 --> 01:44:48,458
PROCESS BECAUSE OF THE VALUE WE

2744
01:44:48,458 --> 01:44:50,494
RECOGNIZED OF CHEMICAL PROBES TO

2745
01:44:50,494 --> 01:44:51,128
THE RESEARCH COMMUNITY.

2746
01:44:51,128 --> 01:44:52,996
NOW ALL THIS WORK IS IN THE

2747
01:44:52,996 --> 01:45:00,070
PUBLIC DOMAIN, EVEN THOUGH WE

2748
01:45:00,070 --> 01:45:01,138
ARE CO FUNDED AND COLLABORATE

2749
01:45:01,138 --> 01:45:02,973
WITH A NUMBER OF PHARMACEUTICAL

2750
01:45:02,973 --> 01:45:06,109
COMPANIES AND SO THIS IS PRETTY

2751
01:45:06,109 --> 01:45:07,144
COMPETITIVE RESEARCH, AND WE'RE

2752
01:45:07,144 --> 01:45:08,779
ENDING UP OVER THE NEXT 10 YEARS

2753
01:45:08,779 --> 01:45:11,748
WE'RE AIMING TO DO A LOT MORE OF

2754
01:45:11,748 --> 01:45:15,786
THIS, WE CAN LEVERAGE MACHINE

2755
01:45:15,786 --> 01:45:17,354
LEARNING AND GLOBAL

2756
01:45:17,354 --> 01:45:18,655
COMPUTATIONAL COMMUNITY TO

2757
01:45:18,655 --> 01:45:21,358
PARTNER WITH THE DATA THAT WE

2758
01:45:21,358 --> 01:45:22,826
GENERATE, TO DEVELOP METHODS

2759
01:45:22,826 --> 01:45:25,229
THAT WILL ACCELERATE DRUG DORPHY

2760
01:45:25,229 --> 01:45:27,731
AND TO IDENTIFY A MODULATOR FOR

2761
01:45:27,731 --> 01:45:28,532
MOST HUMAN PROTEINS.

2762
01:45:28,532 --> 01:45:35,806
THAT'S THIS PROGEC CALLED TARGET

2763
01:45:35,806 --> 01:45:36,306
2035.

2764
01:45:36,306 --> 01:45:37,307
SO THE IDEAL MOLECULE WE WOULD

2765
01:45:37,307 --> 01:45:40,277
LIKE TO MAKE FOR EACH HUMAN

2766
01:45:40,277 --> 01:45:42,946
PROTEIN IS A PROBE OR SMALL LIKE

2767
01:45:42,946 --> 01:45:43,914
MOLECULE THAT SELECTIVELY

2768
01:45:43,914 --> 01:45:46,216
MODULATES THE ACTIVITY OF A

2769
01:45:46,216 --> 01:45:46,950
SPECIFIC PROTEIN.

2770
01:45:46,950 --> 01:45:48,719
THE CONCEPT IS PUBLISHED IN THIS

2771
01:45:48,719 --> 01:45:51,555
PAPER A NUMBER OF YEARS AGO,

2772
01:45:51,555 --> 01:45:56,293
THESE ARE HIGHLY, HIGHLY USEFUL

2773
01:45:56,293 --> 01:45:58,295
TOOLS, OFTEN THE TYPE OF TOOL 1

2774
01:45:58,295 --> 01:45:59,596
MAKES INITIALLY WHEN YOU ARE A

2775
01:45:59,596 --> 01:46:03,333
TARGET OF INTEREST TO INSURE

2776
01:46:03,333 --> 01:46:04,167
THAT PHARMACOLOGICAL INHIBITION

2777
01:46:04,167 --> 01:46:07,671
ARE NOW KNOCKED DOWN OR CHEMICAL

2778
01:46:07,671 --> 01:46:09,840
KNOCK DOWN WITH A PROTAC

2779
01:46:09,840 --> 01:46:11,375
DEGREATER GIVES YOU THE

2780
01:46:11,375 --> 01:46:19,916
PHENOTYPE THAT 1'S INTERESTED IN

2781
01:46:19,916 --> 01:46:22,653
EMPLOY SO WE MADE AND

2782
01:46:22,653 --> 01:46:23,954
COLLABORATED WITH THE

2783
01:46:23,954 --> 01:46:24,721
PHARMACEUTICAL INDUSTRY TO

2784
01:46:24,721 --> 01:46:27,791
COLLABORATE OVER 200 CHEMICAL

2785
01:46:27,791 --> 01:46:28,091
PROBES NOW.

2786
01:46:28,091 --> 01:46:30,327
AND IN THE LAST 10 YEARS AND SO

2787
01:46:30,327 --> 01:46:31,928
SOME OF THESE ARE PUBLISHED HERE

2788
01:46:31,928 --> 01:46:33,897
IN THESE PAPERINGS SO WE MADE A

2789
01:46:33,897 --> 01:46:35,098
NUMBER OF THEM OURSELVES, IN

2790
01:46:35,098 --> 01:46:36,867
COLLABORATION WITH BIG FARM ABIG

2791
01:46:36,867 --> 01:46:38,969
PHARMA DONATED A NUMBER OF

2792
01:46:38,969 --> 01:46:47,511
CHEMICAL PROBES, IT'S A NICE

2793
01:46:47,511 --> 01:46:50,013
KINASE, SOMEWHAT SELECTIVE SET

2794
01:46:50,013 --> 01:46:51,615
OF KINASE INHIBITORS WITH THE

2795
01:46:51,615 --> 01:46:54,318
GENOMICS SET SO THESE ARE ALL

2796
01:46:54,318 --> 01:46:55,519
PUBLICLY AVAILABLE COMPOUNDS AND

2797
01:46:55,519 --> 01:46:57,654
DATA THAT GOES WITH THEM.

2798
01:46:57,654 --> 01:47:02,259
AND THEY'VE BEEN USED BY

2799
01:47:02,259 --> 01:47:03,293
THOUSANDS OF PEOPLE AND

2800
01:47:03,293 --> 01:47:07,731
THOUSANDS OF PAPERS BY NOW.

2801
01:47:07,731 --> 01:47:08,832
SO NEVERTHELESS, IT'S STILL A

2802
01:47:08,832 --> 01:47:10,334
VERY SMALL NUMBER OF HUMAN

2803
01:47:10,334 --> 01:47:12,369
PROTEINS, AND SO THERE'S THIS

2804
01:47:12,369 --> 01:47:13,970
DARK PROTEOME, IF WE -- THIS IS

2805
01:47:13,970 --> 01:47:15,572
A NUMBER OF PUBLISHED PAPERS

2806
01:47:15,572 --> 01:47:21,044
GIVEN FOR EACH OF THE ROUGHLY

2807
01:47:21,044 --> 01:47:21,478
2000 HUMAN PROTEINS.

2808
01:47:21,478 --> 01:47:23,246
SO MOST OF THESE ARE

2809
01:47:23,246 --> 01:47:24,014
UNDERSTUDIED WITH NO

2810
01:47:24,014 --> 01:47:24,748
PUBLICATIONS ON THEM.

2811
01:47:24,748 --> 01:47:26,583
WE CAN PREDICT THEIR STRUCTURES

2812
01:47:26,583 --> 01:47:28,685
NOW BUT WE STILL DON'T KNOW WHAT

2813
01:47:28,685 --> 01:47:32,789
MANY OF THEM DO FUNCTIONALLY.

2814
01:47:32,789 --> 01:47:34,691
UNBIASED GENOMICS STUDIES SUCH

2815
01:47:34,691 --> 01:47:36,193
AS GWAS STUDIES, ET CETERA SHOW

2816
01:47:36,193 --> 01:47:37,160
MOST HUMAN PROTEINS HAVE SOME

2817
01:47:37,160 --> 01:47:39,796
KIND OF LINK TO A TEASE.

2818
01:47:39,796 --> 01:47:41,531
SO THERE IS A RATIONAL FOR

2819
01:47:41,531 --> 01:47:44,067
TRYING TO FIND A MODULATOR OR A

2820
01:47:44,067 --> 01:47:46,870
LIGAND TO MOST HUMAN PROTEINS IF

2821
01:47:46,870 --> 01:47:47,437
WE CAN.

2822
01:47:47,437 --> 01:47:51,241
AND THAT'S WHAT THE OBJECTIVE OF

2823
01:47:51,241 --> 01:47:52,676
THIS TARGET 2035 IS SHOWN HERE

2824
01:47:52,676 --> 01:47:54,344
SO ALL I WILL TALK ABOUT TODAY

2825
01:47:54,344 --> 01:47:56,780
IS WAYS WE CAN ACCELERATE

2826
01:47:56,780 --> 01:47:59,349
GETTING TO THIS POINT AND THESE

2827
01:47:59,349 --> 01:48:02,252
ARE JUST THE, AGAIN, THE USEFUL,

2828
01:48:02,252 --> 01:48:03,653
YOU KNOW THE GREAT NEWS IT WOULD

2829
01:48:03,653 --> 01:48:06,423
BE IF YOU HAD A MODULATOR OF

2830
01:48:06,423 --> 01:48:09,226
EVERY HUMAN PROTEIN MUCH THE

2831
01:48:09,226 --> 01:48:12,729
PROBLEM IS, THAT EACH CHEMICAL

2832
01:48:12,729 --> 01:48:14,931
PROBE A POTENT SELECTIVE IN CELL

2833
01:48:14,931 --> 01:48:16,967
MOLECULE COSTS ON THE ORDER OF 2

2834
01:48:16,967 --> 01:48:18,168
TO $5 MILLION EACH.

2835
01:48:18,168 --> 01:48:20,971
TAKES YEARS TO DEVELOP, AND WE

2836
01:48:20,971 --> 01:48:24,674
WILL JUST NEVER GET TO THE POINT

2837
01:48:24,674 --> 01:48:28,612
OF HAVING 1 FOR MOST HUMAN

2838
01:48:28,612 --> 01:48:30,514
PROTEINS, UNLESS WE LEVERAGED

2839
01:48:30,514 --> 01:48:32,916
THE POWER OF MACHINE LEARNING

2840
01:48:32,916 --> 01:48:35,419
AND NEW TECHNOLOGIES IN A

2841
01:48:35,419 --> 01:48:43,927
MASSIVE WAY AND DO THIS IN A

2842
01:48:43,927 --> 01:48:45,395
GLOBALLY GENERATED MANNER AND DO

2843
01:48:45,395 --> 01:48:47,030
THIS WITH ALL THE TECHNOLOGIES

2844
01:48:47,030 --> 01:48:47,431
TOGETHER.

2845
01:48:47,431 --> 01:48:48,098
THIS HAS BEEN COMMENT OFFICE OF

2846
01:48:48,098 --> 01:48:52,369
DIVERSITY IN A NUMBER OF PAPERS

2847
01:48:52,369 --> 01:48:54,404
RECENTLY ABOUT THE POTENTIAL OF

2848
01:48:54,404 --> 01:48:56,940
AI MACHINE LEARNING TO

2849
01:48:56,940 --> 01:48:58,141
ACCELERATE DRUG DISCOVERY,

2850
01:48:58,141 --> 01:48:59,443
HERE'S QUOTES WITH UNDERLYING

2851
01:48:59,443 --> 01:49:01,378
KEY PARTS THAT PARTICULARLY WE

2852
01:49:01,378 --> 01:49:07,417
ARE TRYING TO ADDRESS WHERE,

2853
01:49:07,417 --> 01:49:10,053
WHERE PREDICTIONS, COMPUTATIONAL

2854
01:49:10,053 --> 01:49:10,720
PREDICTIONS ARE MADE ABOUT

2855
01:49:10,720 --> 01:49:12,122
COMPOUNDS THAT ARE MADE TO A

2856
01:49:12,122 --> 01:49:14,257
TARGET NEED TO BE VALIDATED IN

2857
01:49:14,257 --> 01:49:14,858
AN EXPERIMENTAL SETTING,

2858
01:49:14,858 --> 01:49:16,960
SOMEBODY HAS TO DO THIS WHICH IS

2859
01:49:16,960 --> 01:49:21,631
AN EXPENSIVE UNDERTAKING AND A

2860
01:49:21,631 --> 01:49:23,700
LOT OF THIS, A LOT OF THE DRUG

2861
01:49:23,700 --> 01:49:24,634
DORPHY AND LIGAND DISCOVERY

2862
01:49:24,634 --> 01:49:25,969
TAKES PLACE IN INDUSTRY AND AS

2863
01:49:25,969 --> 01:49:28,238
WE JUST HEARD, THERE'S A LOT OF

2864
01:49:28,238 --> 01:49:30,974
KNOWLEDGE IN INDUSTRY, SO IF

2865
01:49:30,974 --> 01:49:32,042
INDUSTRY AND ACADEMIA CAN WORK

2866
01:49:32,042 --> 01:49:38,515
TOGETHER, I THINK WE WILL GET

2867
01:49:38,515 --> 01:49:39,049
THERE FASTER.

2868
01:49:39,049 --> 01:49:41,585
SO WE HAVE TOWARDS THAT GOAL,

2869
01:49:41,585 --> 01:49:43,520
WE'VE CARRIED OUT A NUMBER OF

2870
01:49:43,520 --> 01:49:44,821
RECENT PILOT PROJECTS ON

2871
01:49:44,821 --> 01:49:50,727
EXPERIMENT AL HIT FINDING.

2872
01:49:50,727 --> 01:49:52,162
THE IDEA IS TO FIND A

2873
01:49:52,162 --> 01:49:55,131
METHODOLOGY THAT WILL BE OF

2874
01:49:55,131 --> 01:50:00,570
GENERIC USE, IF YOU WILL, SO,

2875
01:50:00,570 --> 01:50:10,547
WE'VE FOCUSED ON THESE

2876
01:50:10,547 --> 01:50:12,649
BIOPHYSICAL BINDING ASSAYS AX

2877
01:50:12,649 --> 01:50:13,583
FINNITY AFFLICTION, AND WE'RE

2878
01:50:13,583 --> 01:50:20,790
TRYING TO GET AWAY FROM AND MORE

2879
01:50:20,790 --> 01:50:24,895
EXPENSIVE AND MORE TECHNICALLY

2880
01:50:24,895 --> 01:50:27,264
DEMANDING ASSAYS SUCH AS EXTRA

2881
01:50:27,264 --> 01:50:28,965
FRAGMENT SCREENING KLF-TWO WORKS

2882
01:50:28,965 --> 01:50:31,601
AND IS QUITE USEFUL BUT REQUIRES

2883
01:50:31,601 --> 01:50:34,571
SPECIAL SET UP AND MATERIALS,

2884
01:50:34,571 --> 01:50:39,009
AND HITS ARE VERY WEAK OR YOU

2885
01:50:39,009 --> 01:50:44,014
KNOW SPECIFIC FUNCTIONAL ASSAYS

2886
01:50:44,014 --> 01:50:45,315
FOR EXAMPLE WERE PLANT BASED

2887
01:50:45,315 --> 01:50:47,617
METHODS THAT REQUIRE ALSO A LOT

2888
01:50:47,617 --> 01:50:49,920
OF TWEAKING AND MAYBE DIFFERENT

2889
01:50:49,920 --> 01:50:50,720
FOR DIFFERENT PROTEINS.

2890
01:50:50,720 --> 01:50:52,822
SO THE METHODS SHOWN HERE, I

2891
01:50:52,822 --> 01:50:54,791
WILL WALK THROUGH BRIEFLY, EACH

2892
01:50:54,791 --> 01:50:57,127
OF THESE, WE'RE COLLABORATING

2893
01:50:57,127 --> 01:51:00,964
WITH INDUSTRY TO DEVELOP AND

2894
01:51:00,964 --> 01:51:02,399
DEMONSTRATE AND APPLIES TO APPLY

2895
01:51:02,399 --> 01:51:04,434
THEM AT SCALE ON A LOT OF

2896
01:51:04,434 --> 01:51:10,106
PROTEINS OVER THE NEXT 5-10

2897
01:51:10,106 --> 01:51:10,307
YEARS.

2898
01:51:10,307 --> 01:51:13,143
SO PILOT 1 IS DNA ENCODED

2899
01:51:13,143 --> 01:51:14,077
LIBRARY WITH SELECTION AND

2900
01:51:14,077 --> 01:51:16,746
MACHINE LEARNING ANALYSIS THAT

2901
01:51:16,746 --> 01:51:19,716
WOULD ENABLE 1 TO PREDICT

2902
01:51:19,716 --> 01:51:20,650
COMMERCIALLY AVAILABLE

2903
01:51:20,650 --> 01:51:22,352
COMPOUNDS, SO, THAT 1 COULD

2904
01:51:22,352 --> 01:51:23,820
PURCHASE THAT ARE PREDICTED TO

2905
01:51:23,820 --> 01:51:26,623
BIND THE PROTEIN OF INTEREST AS

2906
01:51:26,623 --> 01:51:28,291
OPPOSE TO SINGICIZING EXACTLY

2907
01:51:28,291 --> 01:51:29,526
THE MOLECULE OFF DNA THAT WAS

2908
01:51:29,526 --> 01:51:32,862
SELECTED BY THE PROTEIN BECAUSE

2909
01:51:32,862 --> 01:51:35,065
THE LATTER IS FAIRLY EXPENSIVE

2910
01:51:35,065 --> 01:51:35,966
AND RESOURCE INTENSIVE.

2911
01:51:35,966 --> 01:51:38,301
WE APPLIED THIS TO THIS LARGE

2912
01:51:38,301 --> 01:51:40,870
PROTEIN FAMILY CALLED WD40

2913
01:51:40,870 --> 01:51:43,473
REPEAT PROTEINS, THESE ARE

2914
01:51:43,473 --> 01:51:46,643
PROTEIN-PROTEIN INTERACTION 

2915
01:51:46,643 --> 01:51:48,178
DOMAINS AND IN THE CENTER OF

2916
01:51:48,178 --> 01:51:50,880
THIS DENDRITIC CELL RAGRAM IS A

2917
01:51:50,880 --> 01:51:54,751
STRUCTURE OF 1 WDR PROTEIN THAT

2918
01:51:54,751 --> 01:51:56,453
ARE DATA PROTELLER PROTEINS

2919
01:51:56,453 --> 01:51:58,221
OFTEN BINDING A PEPTIDE LIGAND

2920
01:51:58,221 --> 01:52:00,657
OF ANOTHER PROTEIN IN THE

2921
01:52:00,657 --> 01:52:02,258
CENTRAL POCKET HERE.

2922
01:52:02,258 --> 01:52:03,560
WE CLONED EXPRESSED PURIFIED,

2923
01:52:03,560 --> 01:52:05,362
SOLVED A NUMBER OF STRUCTURES AS

2924
01:52:05,362 --> 01:52:08,064
SHOWN ON HERE AND THEN WE TOOK

2925
01:52:08,064 --> 01:52:11,334
16 OF THESE WDR PROTEINS AND 2

2926
01:52:11,334 --> 01:52:12,969
NONWDR PROTEINS SO THAT OUR

2927
01:52:12,969 --> 01:52:15,839
METHOD WAS NOT -- SO WE KNOW OUR

2928
01:52:15,839 --> 01:52:17,107
METHOD WAS NOT BIASED TOWARDS

2929
01:52:17,107 --> 01:52:25,081
JUST THIS 1 PROTEIN FAMILY.

2930
01:52:25,081 --> 01:52:26,216
WE COLLABORATED WITH X-SCN OR

2931
01:52:26,216 --> 01:52:27,617
THE LIBRARY IN THE FIELD AND

2932
01:52:27,617 --> 01:52:30,453
THEY DID LIBRARY SELECTIONS, WE

2933
01:52:30,453 --> 01:52:39,229
THEN COLLABORATED WITH COMPANY

2934
01:52:39,229 --> 01:52:43,233
CALLED ZEBI-AI AT THE TIME TO

2935
01:52:43,233 --> 01:52:46,603
USE MACHINE LEARNING TO ANALYZE

2936
01:52:46,603 --> 01:52:48,905
THE DATA, THE SELECTION DATA AND

2937
01:52:48,905 --> 01:52:49,939
PREDICT COMPOUNDS THAT WOULD

2938
01:52:49,939 --> 01:52:53,276
BIND TO OUR 18 TARGETS.

2939
01:52:53,276 --> 01:52:56,413
WE TESTED THOSE 18 TARGETS USING

2940
01:52:56,413 --> 01:52:58,615
A VARIETY OF BIOPHYSICAL METHODS

2941
01:52:58,615 --> 01:53:02,819
SHOWN HERE, EACH WAS TESTED BY

2942
01:53:02,819 --> 01:53:04,621
SURFACE PLASMA AND RESONANCE SO

2943
01:53:04,621 --> 01:53:08,191
THE SPR AND IN A SECOND, TO CALL

2944
01:53:08,191 --> 01:53:09,459
A COMPOUND, CONFIRM THE COMPOUND

2945
01:53:09,459 --> 01:53:13,363
IS NOT A BINDER, WE WANTED 2

2946
01:53:13,363 --> 01:53:14,230
BIOPHYSICAL METHODS TO BE SURE

2947
01:53:14,230 --> 01:53:15,331
IT WAS REALLY BINDING.

2948
01:53:15,331 --> 01:53:24,140
SO HERE'S THE RESULTS.

2949
01:53:24,140 --> 01:53:27,977
AT LEAST A THIRD OF THE TARGETS

2950
01:53:27,977 --> 01:53:30,380
SHOWING THE TARGETS IN BIND.

2951
01:53:30,380 --> 01:53:32,015
A COUPLE OTHER WEAKER BINDERS

2952
01:53:32,015 --> 01:53:34,751
AND OTHER PROTEINS DID NOT

2953
01:53:34,751 --> 01:53:41,124
RETURN A VALIDATED BINDER.

2954
01:53:41,124 --> 01:53:45,361
BUT, ALTHOUGH THESE WERE WEAKISH

2955
01:53:45,361 --> 01:53:48,832
BINDERS IN THE MICROMOLAR RANGE,

2956
01:53:48,832 --> 01:53:53,737
THEY'RE ALL NEW ADDRESSABLE

2957
01:53:53,737 --> 01:53:56,840
COMPOUNDS AND IT WILL BE IN THE

2958
01:53:56,840 --> 01:53:58,742
PAPER COMING OUT IN THE NEXT

2959
01:53:58,742 --> 01:54:00,844
JAMA AND IN THE BI ONY ARCHIVE.

2960
01:54:00,844 --> 01:54:05,982
SO ENCOURAGED BY THE SUCCESS LED

2961
01:54:05,982 --> 01:54:09,085
BY RAFAEL [INDISCERNIBLE] AT SBC

2962
01:54:09,085 --> 01:54:10,820
AT UNC CHAPEL HILL IS LEADING

2963
01:54:10,820 --> 01:54:18,461
OUR DEVELOPMENT OF AN OPEN

2964
01:54:18,461 --> 01:54:19,796
DELL ML COLLABORATIVE NETWORK IN

2965
01:54:19,796 --> 01:54:21,464
WHICH WE'RE DOING DELATWAL

2966
01:54:21,464 --> 01:54:22,198
SELECTIONS AND CORRELATING THE

2967
01:54:22,198 --> 01:54:28,071
DATA IN  THE FORM OF FINGERPRINT

2968
01:54:28,071 --> 01:54:32,976
SELECTED MOLECULES AND IN A

2969
01:54:32,976 --> 01:54:35,712
DATABASE CALLED ARCHECK, I WILL

2970
01:54:35,712 --> 01:54:36,913
MENTION IN A MOMENT, BUT THIS

2971
01:54:36,913 --> 01:54:39,315
WILL BE AVAILABLE FOR ANYONE TO

2972
01:54:39,315 --> 01:54:41,284
DO MACHINE LEARNING AND PREDICT

2973
01:54:41,284 --> 01:54:42,252
COMPOUNDS THAT BIND TO THE

2974
01:54:42,252 --> 01:54:45,388
PROTEIN OF INTEREST AND ALSO

2975
01:54:45,388 --> 01:54:46,489
SHARE THESE MODELS FOR THE

2976
01:54:46,489 --> 01:54:47,690
COMMUNITY TO USE, AND SO, SOME

2977
01:54:47,690 --> 01:54:51,661
OF THE GROUPS THAT ARE DOING THE

2978
01:54:51,661 --> 01:54:56,032
DELL SCREENING ARE SHOWN ALONG

2979
01:54:56,032 --> 01:54:56,466
THE BOTTOM HERE.

2980
01:54:56,466 --> 01:54:59,869
SO THAT WILL BE GENERATING

2981
01:54:59,869 --> 01:55:00,737
PROTEIN SMALL MOLECULE

2982
01:55:00,737 --> 01:55:03,206
INTERACTION DATA FROM DELL ML,

2983
01:55:03,206 --> 01:55:07,343
THE SECOND PILOT WAS, AFFINITY

2984
01:55:07,343 --> 01:55:09,379
SELECTION MASS SPECTROMETRY,

2985
01:55:09,379 --> 01:55:11,514
EXPERIMENTS THAT ARE LARGELY

2986
01:55:11,514 --> 01:55:14,851
USED IN INDUSTRY, BUT NOT VERY

2987
01:55:14,851 --> 01:55:20,056
MUCH IN THE ACADEMIA.

2988
01:55:20,056 --> 01:55:23,626
OUR COLLEAGUE HUI PENG IS AN

2989
01:55:23,626 --> 01:55:25,428
EXPERT IN MASS SPECTROMETRY AT

2990
01:55:25,428 --> 01:55:27,063
THE UNIVERSITY OF TORONTO AND

2991
01:55:27,063 --> 01:55:30,767
USED IT IN THE FIELD OF SCIENCE

2992
01:55:30,767 --> 01:55:37,540
AND COLLABORATING THEY LEVON

2993
01:55:37,540 --> 01:55:41,377
HALABELIAN, AT TORONTO, THEY

2994
01:55:41,377 --> 01:55:42,512
WORKED TOGETHER TO BUILT THIS TO

2995
01:55:42,512 --> 01:55:44,113
BE ABLE TO SCREEN PROTEINS IN A

2996
01:55:44,113 --> 01:55:47,851
NUMBER OF DAYS AGAINST A NUMBER

2997
01:55:47,851 --> 01:55:48,751
OF COMPOUNDS.

2998
01:55:48,751 --> 01:55:51,120
WE'VE BUILT SEVERAL LIBRARIES, A

2999
01:55:51,120 --> 01:55:53,223
LARGE LIBRARY FROM

3000
01:55:53,223 --> 01:55:54,591
[INDISCERNIBLE] DIVERSITY

3001
01:55:54,591 --> 01:55:56,659
LIBRARY AND SMALLER LIBRARIES

3002
01:55:56,659 --> 01:56:00,396
FOR JUST SORT OF LIGAND --

3003
01:56:00,396 --> 01:56:02,832
ASSESSING LIGAND ABILITY OF

3004
01:56:02,832 --> 01:56:03,466
PROTEINS FOR EXAMPLE.

3005
01:56:03,466 --> 01:56:05,735
NOW INTERESTING THING ABOUT THIS

3006
01:56:05,735 --> 01:56:11,674
METHOD IS IT CAN BE -- IF THE --

3007
01:56:11,674 --> 01:56:13,943
IF THE ISOLATION OF THE COMPOUND

3008
01:56:13,943 --> 01:56:15,278
THAT BINDS TO YOUR PROTEIN OF

3009
01:56:15,278 --> 01:56:21,818
INTEREST IS DONE THROUGH A

3010
01:56:21,818 --> 01:56:25,388
COLUMN, WE CAN SLEKD AN ANTIMDR

3011
01:56:25,388 --> 01:56:26,422
BIEN XG THAT WAS QUITE

3012
01:56:26,422 --> 01:56:28,291
INTERESTING AND THESE ARE

3013
01:56:28,291 --> 01:56:29,425
EXAMPLES FROM KNOWN PROTEINS AND

3014
01:56:29,425 --> 01:56:33,162
COMPOUND PAIRS THAT WE IMU ONLY

3015
01:56:33,162 --> 01:56:34,731
1 WOULD BIND TO THE OTHER.

3016
01:56:34,731 --> 01:56:36,466
SO THIS WAS VERY EXCITING AS

3017
01:56:36,466 --> 01:56:39,102
WELL, AND IT TURNS OUT THAT IT

3018
01:56:39,102 --> 01:56:41,871
SEEMS THAT WHEN 1 -- WHEN YOU

3019
01:56:41,871 --> 01:56:47,810
GET JUST 1 PEAK OUT OF A

3020
01:56:47,810 --> 01:56:48,411
[INDISCERNIBLE] THAT WAS

3021
01:56:48,411 --> 01:56:50,046
SCREENED WITH THE PROTEIN, THAT

3022
01:56:50,046 --> 01:56:53,216
THERE'S A HIGHER PERCENTAGE OF

3023
01:56:53,216 --> 01:56:54,150
THAT COMPOUND VALIDATING OF

3024
01:56:54,150 --> 01:56:56,719
COURSE THIS IS A KREEN SO NOT

3025
01:56:56,719 --> 01:56:57,720
EVERY COMPOUND VALIDATES.

3026
01:56:57,720 --> 01:56:59,656
WE APPLIED THIS TO OVER 30

3027
01:56:59,656 --> 01:57:03,927
TARGETS SHOWN HERE WITH THE

3028
01:57:03,927 --> 01:57:06,863
VARIABLE LIGAND ABILITY SCORES

3029
01:57:06,863 --> 01:57:08,464
AS SHOWN, SOME OF THE STATISTICS

3030
01:57:08,464 --> 01:57:11,334
ARE SHOWN AT THE BOTTOM AND

3031
01:57:11,334 --> 01:57:14,137
ABOUT ALMOST A THIRD TO A HALF

3032
01:57:14,137 --> 01:57:21,544
OF THESE GAVE HITS AND MANY OF

3033
01:57:21,544 --> 01:57:23,179
THOSE WERE VALIDATED BY

3034
01:57:23,179 --> 01:57:23,646
BIOPHYSICAL METHODS.

3035
01:57:23,646 --> 01:57:26,649
SO HOW DO WE SCORE, SCALE,

3036
01:57:26,649 --> 01:57:27,450
DISSEMINATE THESE LARGE DATA SET

3037
01:57:27,450 --> 01:57:29,519
ANDS HOW DO WE MAKE THEM

3038
01:57:29,519 --> 01:57:32,689
ACCESSIBILITY TO THE COMMUNITY

3039
01:57:32,689 --> 01:57:39,429
TO USE.

3040
01:57:39,429 --> 01:57:40,830
SO BENJAMIN CANES IS A SCIENTIST

3041
01:57:40,830 --> 01:57:42,065
AT THE UNIVERSITY OF TR RONT O

3042
01:57:42,065 --> 01:57:43,366
AND HE THINKS A LOT ABOUT THIS

3043
01:57:43,366 --> 01:57:43,700
AREA.

3044
01:57:43,700 --> 01:57:45,969
WE WANT TO MAKE THE DATA

3045
01:57:45,969 --> 01:57:47,103
AVAILABLE TO EVERYONE AND WE

3046
01:57:47,103 --> 01:57:48,938
WANT PEOPLE TO USE IT AND THEN

3047
01:57:48,938 --> 01:57:53,643
GIVE BACK THEIR MODELS TO THE

3048
01:57:53,643 --> 01:57:56,512
COMMUNITY TO USE, AND SO, BENIA

3049
01:57:56,512 --> 01:57:59,415
MIN CAME UP WITH THE CONCEPT OF

3050
01:57:59,415 --> 01:58:01,384
ARTIFICIAL INTELLIGENCE READY

3051
01:58:01,384 --> 01:58:03,319
CHEMICAL KNOWLEDGE BASE OR AIR

3052
01:58:03,319 --> 01:58:08,257
CHECK, THIS IS A DATABASE WHOSE

3053
01:58:08,257 --> 01:58:10,059
PROTOTYPE IS HOSTED AT GOOGLE

3054
01:58:10,059 --> 01:58:12,628
CLOUD,OOSE AVAILABLE WITH NO

3055
01:58:12,628 --> 01:58:14,197
RESTRICTIONS AND THE USE OF THE

3056
01:58:14,197 --> 01:58:16,799
DATA IF YOU AGREEN CELLS TO,

3057
01:58:16,799 --> 01:58:18,968
JUST A FEW BASIC POINTS, THE

3058
01:58:18,968 --> 01:58:20,436
DATA IS FORMATTED FOR DATA

3059
01:58:20,436 --> 01:58:22,038
SCIENTISTS WHICH IS THE KEY

3060
01:58:22,038 --> 01:58:22,271
THING.

3061
01:58:22,271 --> 01:58:26,943
SO THE DATA IS UNIFORM, ALWAYS

3062
01:58:26,943 --> 01:58:29,946
STORED IN THE SAME WAY AND TELL

3063
01:58:29,946 --> 01:58:31,748
BE PRIMARILY TO TYPES OF DATA AT

3064
01:58:31,748 --> 01:58:35,985
LEAST IN THE BEGINNING TO

3065
01:58:35,985 --> 01:58:38,187
DELL ML AND DELL SELECTIONS AND

3066
01:58:38,187 --> 01:58:39,756
ASMS DAILY BASIS THEA AND WE

3067
01:58:39,756 --> 01:58:41,457
HIGHLY ENCOURAGE, DON'T REQUIRE

3068
01:58:41,457 --> 01:58:44,227
BUT WE HIGHLY ENCOURAGE, USERS

3069
01:58:44,227 --> 01:58:48,464
TO MAKE THEIR MACHINE LEARNING

3070
01:58:48,464 --> 01:58:52,301
MODELS USING THIS DATA OPEN

3071
01:58:52,301 --> 01:58:55,738
SOURCE AND TO EVEN DEPOSIT IT IN

3072
01:58:55,738 --> 01:58:56,539
THE DATABASE.

3073
01:58:56,539 --> 01:58:59,375
SO, THE AIR CHECK TEAM IS IN THE

3074
01:58:59,375 --> 01:59:06,582
LOWER LEFT HERE, THIS A SCREEN

3075
01:59:06,582 --> 01:59:09,552
SHOT OF THE FROM THE WEBPAGE AND

3076
01:59:09,552 --> 01:59:11,187
ALEX'S GROUP FROM UNC HAS USED A

3077
01:59:11,187 --> 01:59:14,757
NUMBER OF THESE DATA SETS TO

3078
01:59:14,757 --> 01:59:17,560
SUCCESSFULLY CREATE AN ML MODEL

3079
01:59:17,560 --> 01:59:19,629
THAT PREDICTS A COMPOUND THAT IS

3080
01:59:19,629 --> 01:59:24,400
VALIDATED TO BIND, TO TARGET A

3081
01:59:24,400 --> 01:59:25,635
COUPLE TARGETS OF INTEREST, AND

3082
01:59:25,635 --> 01:59:33,910
SOME OF THE TARGETS ARE SHOWN

3083
01:59:33,910 --> 01:59:36,212
HERE AND THIS IS KIND OF

3084
01:59:36,212 --> 01:59:38,081
ENCOURAGING THAT WE CAN TRY TO

3085
01:59:38,081 --> 01:59:40,149
FOSTER THIS ACTIVITY IN A SORT

3086
01:59:40,149 --> 01:59:42,685
OF GLOBAL NETWORK AMONGST THE

3087
01:59:42,685 --> 01:59:49,058
COMMUNITY. 

3088
01:59:49,058 --> 01:59:55,765
THE THIRD PILOT, WAS 1 OF PURELY

3089
01:59:55,765 --> 01:59:56,232
COMPUTATIONAL METHODS.

3090
01:59:56,232 --> 01:59:58,201
WE WERE ASKED BY COMPANIES EARLY

3091
01:59:58,201 --> 02:00:00,436
ON WHETHER WE COULD TEST THEIR

3092
02:00:00,436 --> 02:00:01,404
PREDICS OF COMPOUNDS THAT WERE

3093
02:00:01,404 --> 02:00:03,039
BOUND TO A GIVEN PROTEIN, WE

3094
02:00:03,039 --> 02:00:07,176
WORK WITH ADAM WISE AND SICLICA,

3095
02:00:07,176 --> 02:00:08,978
AND AGAIN ABOUT HALF OR A THIRD

3096
02:00:08,978 --> 02:00:11,747
OF THE CASES THEY IEE, AUDIENCE

3097
02:00:11,747 --> 02:00:12,582
DENTIFIED COMPOUNDS THAT CONFIRM

3098
02:00:12,582 --> 02:00:13,749
OUR PROTEIN OF INTEREST.

3099
02:00:13,749 --> 02:00:16,619
NOW THESE HITS ARE WEAK, THEY'RE

3100
02:00:16,619 --> 02:00:20,356
QUITE WEAK IN FACT, BUT THEY

3101
02:00:20,356 --> 02:00:22,225
WERE MEASURABLE.

3102
02:00:22,225 --> 02:00:25,895
AND THEY WERE ALSO MORE POTENT

3103
02:00:25,895 --> 02:00:28,598
THAN FRAGMENTS TEND TO BE AND

3104
02:00:28,598 --> 02:00:30,533
THEY WERE PROGRESSABLE AND WERE

3105
02:00:30,533 --> 02:00:33,202
FIRST IN CLASS FOR THEIR TARGETS

3106
02:00:33,202 --> 02:00:37,306
SO THIS ALSO LED TO THE, YOU

3107
02:00:37,306 --> 02:00:38,508
KNOW, IT'S ENCOURAGING, I THINK,

3108
02:00:38,508 --> 02:00:39,609
IT WAS ENCOURAGING FOR THE FIELD

3109
02:00:39,609 --> 02:00:41,744
AS I SAID IN THE BEGENERATEDDING

3110
02:00:41,744 --> 02:00:43,246
WE'RE STILL NOT THERE, AND THERE

3111
02:00:43,246 --> 02:00:45,047
IS A LOT OF HYPE IN THE FIELD

3112
02:00:45,047 --> 02:00:46,582
ABOUT AI AND MACHINE LEARNING

3113
02:00:46,582 --> 02:00:49,719
AND DRUG DISCOVERY, SO HOW DO WE

3114
02:00:49,719 --> 02:00:51,254
KNOW WHO TO WORK WITH IN THIS

3115
02:00:51,254 --> 02:00:53,422
AREA AND HOW DO WE KNOW WHEN OUR

3116
02:00:53,422 --> 02:00:55,758
METHODS OR COLLEAGUES METHODS

3117
02:00:55,758 --> 02:00:59,162
ARE REALLY WORKING.

3118
02:00:59,162 --> 02:01:03,733
AND FROM THE CASP AND PROTEIN

3119
02:01:03,733 --> 02:01:11,140
FOLDING FIELD, THAT IS A

3120
02:01:11,140 --> 02:01:13,075
BENCHMARKING EXERCISES, SO FROM

3121
02:01:13,075 --> 02:01:15,978
THE PROTEIN TRUCTURE PREDICTION,

3122
02:01:15,978 --> 02:01:19,749
BENCHMARKING EXERCISE CALLED

3123
02:01:19,749 --> 02:01:22,218
CASPS, CRITICAL ASSESSMENT OF

3124
02:01:22,218 --> 02:01:26,622
STRUCTURE PREDICTION, WE MODELS

3125
02:01:26,622 --> 02:01:27,857
AND MATT [INDISCERNIBLE] IN OUR

3126
02:01:27,857 --> 02:01:30,259
GROUP COLLABORATING WITH JOHN

3127
02:01:30,259 --> 02:01:31,327
[INDISCERNIBLE] WHO HAS

3128
02:01:31,327 --> 02:01:33,262
DEVELOPED AND RUN THE CASP

3129
02:01:33,262 --> 02:01:34,497
COMPETITION FOR MANY, MANY

3130
02:01:34,497 --> 02:01:39,101
YEARS, THEY WORK TOGETHER TO

3131
02:01:39,101 --> 02:01:40,703
CREATE THE CACHE CHALLENGE WHICH

3132
02:01:40,703 --> 02:01:42,038
IS A PROTECTIVE CHALLENGE IN

3133
02:01:42,038 --> 02:01:44,073
WHICH A TARGET IS ANNOUNCED, THE

3134
02:01:44,073 --> 02:01:45,074
COMPUTATIONAL COMMUNITY MAKES

3135
02:01:45,074 --> 02:01:46,108
PREDICTIONS AND IS ABLE TO

3136
02:01:46,108 --> 02:01:48,578
PREDICT UP TO A HUNDRED

3137
02:01:48,578 --> 02:01:51,647
COMPOUNDS AND THEN WE WILL TEST

3138
02:01:51,647 --> 02:01:52,448
THEM EXPERIMENTALLY AND SEE

3139
02:01:52,448 --> 02:02:01,991
WHICH METHOD HAS WORKED THE

3140
02:02:01,991 --> 02:02:02,191
BEST.

3141
02:02:02,191 --> 02:02:04,627
SO THERE'S 6 CACHE COMPETITIONS

3142
02:02:04,627 --> 02:02:05,061
SO FAR.

3143
02:02:05,061 --> 02:02:07,196
TWO HAVE COMPLETED AND THE DATA

3144
02:02:07,196 --> 02:02:10,800
IS PUBLICLY RELEASED, THAT WAS

3145
02:02:10,800 --> 02:02:15,371
THE 2 PARKINSON'S TARGET AND

3146
02:02:15,371 --> 02:02:16,906
SARS-COV-2 XP13 HELIX CASE, AND

3147
02:02:16,906 --> 02:02:18,207
YOU CAN FIND THOSE ON THE

3148
02:02:18,207 --> 02:02:20,376
WEBSITE AND A NUMBER ARE UNDER

3149
02:02:20,376 --> 02:02:22,144
WAY, WE'RE -- THE COMPOUNDS ARE

3150
02:02:22,144 --> 02:02:23,813
BEING TESTED OR PREDICTIONS ARE

3151
02:02:23,813 --> 02:02:25,915
BEING CARRIED OUT RIGHT NOW.

3152
02:02:25,915 --> 02:02:29,051
SO THIS IS A SYSTEMATIC

3153
02:02:29,051 --> 02:02:32,989
SELF-CONSISTENT ANALYSIS FOR

3154
02:02:32,989 --> 02:02:34,323
EACH PROTEIN.

3155
02:02:34,323 --> 02:02:37,593
AND THE -- IT'S A LOT OF

3156
02:02:37,593 --> 02:02:38,394
EXPERIMENTAL VALIDATION NEEDED

3157
02:02:38,394 --> 02:02:42,798
FOR THIS BUT WE THINK THAT IT'S

3158
02:02:42,798 --> 02:02:45,234
USEFUL TO MONITOR OVER TIME HOW

3159
02:02:45,234 --> 02:02:45,868
WELL THE COMPUTATIONAL COMMUNITY

3160
02:02:45,868 --> 02:02:51,407
IS GOING TO BE ABLE TO PREDICT

3161
02:02:51,407 --> 02:02:52,041
COMPOUNDS THAT BIND.

3162
02:02:52,041 --> 02:02:53,809
SO I WILL END IT NOW WITH JUST

3163
02:02:53,809 --> 02:02:58,080
AN OPEN CALL TO JOIN US IN THIS

3164
02:02:58,080 --> 02:02:58,314
EFFORT.

3165
02:02:58,314 --> 02:03:01,584
TARGET 2035 IS MEANT TO BE A

3166
02:03:01,584 --> 02:03:05,054
GLOBAL COLLABORATIVE EFFORT,

3167
02:03:05,054 --> 02:03:06,222
PUBLIC-PRIVATE PARTNERSHIP, IT

3168
02:03:06,222 --> 02:03:10,026
IS AN OPEN SCIENCE PARTNERSHIP,

3169
02:03:10,026 --> 02:03:11,661
AND WE ENCOURAGE ANYONE WHO WAS

3170
02:03:11,661 --> 02:03:13,296
INTERESTED TO REACH OUT.

3171
02:03:13,296 --> 02:03:16,132
SO 1 KEY AREA IS AN OPEN CALL

3172
02:03:16,132 --> 02:03:18,868
FOR PROTEIN TARGETS,

3173
02:03:18,868 --> 02:03:20,970
SPECIFICALLY THE PHYSICAL

3174
02:03:20,970 --> 02:03:22,638
RECOMBIN ANT PROTEIN THAT IS IN

3175
02:03:22,638 --> 02:03:26,142
A GOOD STATE AND IF GROUPS CAN

3176
02:03:26,142 --> 02:03:28,144
PROVIDE US WITH PROTEINS, WE

3177
02:03:28,144 --> 02:03:33,382
WILL DO THE SCREENS BY ASMS OR

3178
02:03:33,382 --> 02:03:34,950
DELL-ML, WE WILL PUT THE DATA IN

3179
02:03:34,950 --> 02:03:39,422
AIR CHECK AND WE WILL VALIDATE

3180
02:03:39,422 --> 02:03:41,557
OR CONFIRM WHETHER OUR PREDICTED

3181
02:03:41,557 --> 02:03:42,725
HITS ARE ACTUALLY BINDING TO THE

3182
02:03:42,725 --> 02:03:47,096
PROTEIN OF INTEREST AND THE

3183
02:03:47,096 --> 02:03:48,864
RESULTS, THE ONLY CAVEAT PERHAPS

3184
02:03:48,864 --> 02:03:50,966
IF YOU VIEW IT AS A CAVEAT IS

3185
02:03:50,966 --> 02:03:52,802
ALL THE RESULTS WILL GO IN THE

3186
02:03:52,802 --> 02:03:54,136
PUBLIC DOMAIN WITH NO

3187
02:03:54,136 --> 02:03:54,804
RESTRICTIONS ON THEIR USE AND IF

3188
02:03:54,804 --> 02:03:56,238
YOU HAVE A PROTEIN, YOU WOULD

3189
02:03:56,238 --> 02:04:03,012
LIKE PUT THROUGH THIS PROCESS,

3190
02:04:03,012 --> 02:04:04,480
NICOLE BURGESSBROWN WHO'S ENAIL

3191
02:04:04,480 --> 02:04:06,382
IS HERE CAN COORDINATE THAT.

3192
02:04:06,382 --> 02:04:07,483
AND IN AN AREA WHERE WE WOULD

3193
02:04:07,483 --> 02:04:16,926
LOVE TO HAVE A COLLABORATORS AND

3194
02:04:16,926 --> 02:04:21,630
AND PARTICIPANTS IS CALLED MAIN

3195
02:04:21,630 --> 02:04:24,100
FRAME, AND WE ARE WORKING ON

3196
02:04:24,100 --> 02:04:25,468
DISCOVERING MANY PROTEINS THAT

3197
02:04:25,468 --> 02:04:29,338
ARE IN THE DARK PROTEOME, A

3198
02:04:29,338 --> 02:04:30,673
NETWORK OF MACHINE SCIENTISTS,

3199
02:04:30,673 --> 02:04:35,678
WE WILL BE RUNNING BENCHMARKING

3200
02:04:35,678 --> 02:04:37,046
CHALLENGES AND EXPERIMENTALLY

3201
02:04:37,046 --> 02:04:41,350
TEST COMPOUNDS, THAT ARE

3202
02:04:41,350 --> 02:04:44,120
PREDICTED FOR FREE AND TOP LABS

3203
02:04:44,120 --> 02:04:46,188
THAT ARE SUCCESSFUL WILL BE

3204
02:04:46,188 --> 02:04:47,189
INVITED TO DEDICATED CONFERENCE

3205
02:04:47,189 --> 02:04:55,865
TO DISCUSS THIS, AND AGAIN, THE

3206
02:04:55,865 --> 02:04:59,101
CONTACTS HERE ARE

3207
02:04:59,101 --> 02:04:59,902
[INDISCERNIBLE], AND MATTHIEU,

3208
02:04:59,902 --> 02:05:02,505
AND ALEX, AND I WANT TO END HERE

3209
02:05:02,505 --> 02:05:03,873
WITH FUTURE DISCOVERY AND WHAT

3210
02:05:03,873 --> 02:05:05,274
WE THINK THEY LOOK LIKE EMPLOY

3211
02:05:05,274 --> 02:05:06,809
TARGETS WE THINK THE WHOLE HUMAN

3212
02:05:06,809 --> 02:05:12,615
PROTEOME IS A GOOD TARGET AND WE

3213
02:05:12,615 --> 02:05:14,683
WANT TO SWITCH FROM THE BURDEN

3214
02:05:14,683 --> 02:05:17,620
OF HAVING TO DO ALL THE INITIAL

3215
02:05:17,620 --> 02:05:18,821
PROTEINS, SMALL MOLECULE

3216
02:05:18,821 --> 02:05:21,190
INTERACTION WORK IN EXPERIMENTAL

3217
02:05:21,190 --> 02:05:23,459
METHODS TO A LARGELY INSILICO

3218
02:05:23,459 --> 02:05:25,461
METHOD AND WE THINK THAT BY

3219
02:05:25,461 --> 02:05:27,563
HAVING ENOUGH DATA AND

3220
02:05:27,563 --> 02:05:29,298
BENCHMARKING TO DEVELOP MACHINE

3221
02:05:29,298 --> 02:05:30,566
LEARNING METHODS, WE WILL

3222
02:05:30,566 --> 02:05:31,834
ACCELERATE THAT, AND THEN I

3223
02:05:31,834 --> 02:05:33,669
THINK NEXT WE'RE GOING TO HEAR

3224
02:05:33,669 --> 02:05:37,373
ABOUT THESE EXCITING AREA OF YOU

3225
02:05:37,373 --> 02:05:39,208
KNOW AUTOMATION AND FURTHER

3226
02:05:39,208 --> 02:05:41,610
DEVELOPMENT OF COMPOUNDS FOR

3227
02:05:41,610 --> 02:05:43,913
DRUG DISCOVERY, GUIDED BY AI IS

3228
02:05:43,913 --> 02:05:46,015
AUTOMATION AND WE THINK THAT

3229
02:05:46,015 --> 02:05:47,883
WILL ALSO BE IMPORTANT, THE

3230
02:05:47,883 --> 02:05:50,319
ACCELERATION CONSORTIUM HERE IN

3231
02:05:50,319 --> 02:05:53,722
TORONTO HAS A LOT OF ACTIVITY IN

3232
02:05:53,722 --> 02:05:53,989
THAT AREA.

3233
02:05:53,989 --> 02:05:56,258
AND THEN, YOU KNOW A PUBLIC

3234
02:05:56,258 --> 02:05:58,127
DOMAIN TOOL FOR EACH HUMAN

3235
02:05:58,127 --> 02:06:00,329
PROTEIN COULD THEN LEAD WHOEVER

3236
02:06:00,329 --> 02:06:02,531
TO MAKE THE DRUGS THAT WE NEED

3237
02:06:02,531 --> 02:06:05,167
FROM IT, SO MANY UNMET HUMAN

3238
02:06:05,167 --> 02:06:05,468
CONDITIONS.

3239
02:06:05,468 --> 02:06:08,971
SO I WOULD LIKE TO THANK, ALL MY

3240
02:06:08,971 --> 02:06:10,139
COLLEAGUES AT THE SBC TR RONT O

3241
02:06:10,139 --> 02:06:14,510
AND ALSO THE OTHER SBC SITES AND

3242
02:06:14,510 --> 02:06:16,645
IN PARTICULAR WHAT I SPOKE ABOUT

3243
02:06:16,645 --> 02:06:19,215
TODAY ARE PARTICULARLY

3244
02:06:19,215 --> 02:06:20,316
VALIDATING PREDICTED HITS AND

3245
02:06:20,316 --> 02:06:22,084
SGC TR RONT O ARE THE FOLKS

3246
02:06:22,084 --> 02:06:32,628
LISTED HERE ON THE LEFT AND THE

3247
02:06:32,995 --> 02:06:34,196
[INDISCERNIBLE], CACHE, AND THE

3248
02:06:34,196 --> 02:06:37,233
FOLKS ARE LISTED HERE AND THE

3249
02:06:37,233 --> 02:06:39,535
ASMS PLATFORM AND THE FUNDERS.

3250
02:06:39,535 --> 02:06:41,704
HAPPY TO TAKE QUESTIONS.

3251
02:06:41,704 --> 02:06:42,605
>> GREAT PRESENTATION.

3252
02:06:42,605 --> 02:06:43,372
COUPLE MINUTES LEFT FOR THE

3253
02:06:43,372 --> 02:06:45,374
QUESTION, SO LET ME ASK FROM THE

3254
02:06:45,374 --> 02:06:49,578
CHART, SO WE HAVE 1 SITUATION,

3255
02:06:49,578 --> 02:06:51,947
YES, TYPICALLY THE POOLS FOR THE

3256
02:06:51,947 --> 02:06:53,349
ABILITY LIMITS THE EFFECTIVE

3257
02:06:53,349 --> 02:06:54,283
CONCENTRATION OF EACH COMPOUND.

3258
02:06:54,283 --> 02:06:58,554
WHAT IS THE POST SIZE FOR YOUR

3259
02:06:58,554 --> 02:06:58,854
ASMS WORK.

3260
02:06:58,854 --> 02:07:02,458
I BELIEVE THEY ARE USING POOLS

3261
02:07:02,458 --> 02:07:05,394
OFIFY HUNDRED COMPOUNDS.

3262
02:07:05,394 --> 02:07:05,661
>> 500.

3263
02:07:05,661 --> 02:07:07,129
>> ALL RIGHT.

3264
02:07:07,129 --> 02:07:08,731
SO WHILE PEOPLE TYPING THE

3265
02:07:08,731 --> 02:07:10,599
QUESTION, LET ME ASK 1.

3266
02:07:10,599 --> 02:07:12,101
YOU MENTIONED THE DEFINITION OF

3267
02:07:12,101 --> 02:07:14,069
THE CHEMICAL PROBLEM ON THE

3268
02:07:14,069 --> 02:07:15,971
SLIDE WHICH IS GREAT BUT LIKE

3269
02:07:15,971 --> 02:07:18,207
FROM YOUR EXPERIENCE, WHAT WOULD

3270
02:07:18,207 --> 02:07:19,708
LIKE AND FROM YOUR OPINION, I

3271
02:07:19,708 --> 02:07:21,677
WOULD SAY WHAT WOULD YOU

3272
02:07:21,677 --> 02:07:23,879
CONSIDER THE REQUIREMENTS TO BE

3273
02:07:23,879 --> 02:07:26,549
A PROBE, IN VIVO ACTIVITY, CELL

3274
02:07:26,549 --> 02:07:28,384
ACTIVITY, AND TARGET ENGAGEMENT

3275
02:07:28,384 --> 02:07:30,185
AND WHAT THE FINISH WILL GIVE.

3276
02:07:30,185 --> 02:07:33,789
>> IN THE FIRST INSTANCE,
CELL

3277
02:07:33,789 --> 02:07:34,924
ACTIVITY, DEFINITELY TARGET

3278
02:07:34,924 --> 02:07:36,625
ENGAGEMENT, AND WE DO MEASURE

3279
02:07:36,625 --> 02:07:38,627
THAT, BEFORE WE CALL SOMETHING A

3280
02:07:38,627 --> 02:07:39,562
CHEMICAL PROBE, WE SHOW THAT IT

3281
02:07:39,562 --> 02:07:43,899
GETS INTO THE CELL ENGAGING THE

3282
02:07:43,899 --> 02:07:46,769
TARGET, DOES NOT HAVE

3283
02:07:46,769 --> 02:07:47,369
UNINHIBITORS TENDED OFF TARGET

3284
02:07:47,369 --> 02:07:49,038
CENTER FOR EXCELLENCE ON AGINGS,

3285
02:07:49,038 --> 02:07:50,172
SO WE TESTED AGAINST A NUMBER OF

3286
02:07:50,172 --> 02:07:51,707
CELL LINES AND MAKE SURE IT'S

3287
02:07:51,707 --> 02:07:55,811
NOT JUST A TOXIC COMPOUND AND BY

3288
02:07:55,811 --> 02:07:57,413
O PHYSICALLY CHARACTERIZE THAT

3289
02:07:57,413 --> 02:07:59,148
THE COMPOUND INTERACTS WITH THE

3290
02:07:59,148 --> 02:08:01,283
PROTEIN OF INTEREST USUALLY WITH

3291
02:08:01,283 --> 02:08:05,621
A RECOMBIN ANT PROTEIN AS WELL.

3292
02:08:05,621 --> 02:08:07,222
>> GOT IT AND VERY LAST

3293
02:08:07,222 --> 02:08:09,658
QUESTION, VERY QUICK, YOU

3294
02:08:09,658 --> 02:08:10,993
MENTIONED THE NAME CODE

3295
02:08:10,993 --> 02:08:13,596
ELABORATE, WHAT IS YOUR

3296
02:08:13,596 --> 02:08:16,332
EXPERIENCE IN REPRODUCE ABILITY

3297
02:08:16,332 --> 02:08:18,934
OF THE [INDISCERNIBLE] LAB

3298
02:08:18,934 --> 02:08:20,302
RAKERY RESULS.

3299
02:08:20,302 --> 02:08:21,136
>> REPRODUCIBILITY?

3300
02:08:21,136 --> 02:08:22,972
>> YES, SO HOW IS IT

3301
02:08:22,972 --> 02:08:25,140
REPRODUCIBLE BECAUSE I THINK WE

3302
02:08:25,140 --> 02:08:26,942
TOUCH BASE PARTLY AS THE

3303
02:08:26,942 --> 02:08:28,577
EFFORTS, SO IF YOU HAVE ANY

3304
02:08:28,577 --> 02:08:30,245
INSIGHT CAN YOU COMMENT ON

3305
02:08:30,245 --> 02:08:30,446
THAT,.

3306
02:08:30,446 --> 02:08:32,581
>> SO IF YOU DO THE SAME
ASSAY

3307
02:08:32,581 --> 02:08:32,781
TWICE.

3308
02:08:32,781 --> 02:08:34,750
>> YES, BASICALLY IF YOU
RUN IT

3309
02:08:34,750 --> 02:08:37,119
A COUPLE TIMES, AND THE

3310
02:08:37,119 --> 02:08:39,121
OFFSEASON, THE ACTIVITY AND

3311
02:08:39,121 --> 02:08:41,790
WHAT'S THE KEY RATE?

3312
02:08:41,790 --> 02:08:43,258
>> YES, SO, DON'T ACTUALLY
-- I

3313
02:08:43,258 --> 02:08:46,061
DON'T ACTUALLY KNOW THE ANSWER

3314
02:08:46,061 --> 02:08:47,296
TO THAT QUESTION IN THAT CASE

3315
02:08:47,296 --> 02:08:49,765
BECAUSE WE'RE NOT DOING THE OFF

3316
02:08:49,765 --> 02:08:52,835
DNA, RIGHT IN SO, WE'RE DODGING

3317
02:08:52,835 --> 02:08:54,903
THAT ISSUE, AND GOING STRAIGHT

3318
02:08:54,903 --> 02:08:59,808
TO A MACHINE LEARNING MODEL FROM

3319
02:08:59,808 --> 02:09:01,410
THE SELECTION DATA AND SAY WHAT

3320
02:09:01,410 --> 02:09:04,079
COULD WE BY THAT LOOKS LIKE IT

3321
02:09:04,079 --> 02:09:05,314
WOULD BIND BASED ON THE

3322
02:09:05,314 --> 02:09:06,682
COMPOUNDS THAT BOUND WITH DNA

3323
02:09:06,682 --> 02:09:08,651
ATTACHED TO THEM AND IT'S NOT A

3324
02:09:08,651 --> 02:09:09,785
HUNDRED PERCENT, I MEAN AND WE

3325
02:09:09,785 --> 02:09:12,921
DON'T KNOW HOW MUCH OF IT IS THE

3326
02:09:12,921 --> 02:09:14,390
MODELS WHICH ARE STILL BEING

3327
02:09:14,390 --> 02:09:16,859
DEVELOPED AND IMPROVED ON, OR

3328
02:09:16,859 --> 02:09:17,593
WHETHER THERE'S SOMETHING, YOU

3329
02:09:17,593 --> 02:09:19,728
KNOW THERE'S SOME ISSUE WITH THE

3330
02:09:19,728 --> 02:09:24,733
DNA BEING THERE OR THAT MAY ALSO

3331
02:09:24,733 --> 02:09:25,868
CONFOUND THANK YOU SO MUCH 1

3332
02:09:25,868 --> 02:09:27,603
MORE TIME ONCE AGAIN, THAT'S A

3333
02:09:27,603 --> 02:09:30,239
GREAT PRESENTATION AND LET ME,

3334
02:09:30,239 --> 02:09:35,511
MOVE FORWARD TO OUR NEXT

3335
02:09:35,511 --> 02:09:39,548
SPEAKER, LET ME INTRODUCE

3336
02:09:39,548 --> 02:09:41,984
MILAD ABOLHASANI, AND HE IS WITH

3337
02:09:41,984 --> 02:09:42,818
THE UNIVERSITY FACULTY SCHOLAR

3338
02:09:42,818 --> 02:09:45,688
IN THE DEPARTMENT OF CHEMICAL

3339
02:09:45,688 --> 02:09:47,489
AND BIOMOLECULAR ENGINEERING IN

3340
02:09:47,489 --> 02:09:49,324
AT NORTH CAROLINA STATE

3341
02:09:49,324 --> 02:09:49,625
UNIVERSITY.

3342
02:09:49,625 --> 02:09:51,060
HE RECEIVED HIS Ph.D. FROM THE

3343
02:09:51,060 --> 02:09:55,964
UNIVERSITY OF TORONTO IN 2014,

3344
02:09:55,964 --> 02:09:58,734
AT NCITATE UNIVERSITY HE LEADS A

3345
02:09:58,734 --> 02:10:00,936
DIVERSE RESEARCH GROUP FOR THE

3346
02:10:00,936 --> 02:10:03,105
STUDY OF SELF-PRIMING COLLABS,

3347
02:10:03,105 --> 02:10:04,440
HE RECEIVED ENORMOUS AMOUNT OF

3348
02:10:04,440 --> 02:10:09,678
THE AWARDS INCLUDING THE AWARD,

3349
02:10:09,678 --> 02:10:10,979
IACL 3555, [INDISCERNIBLE] AWARD

3350
02:10:10,979 --> 02:10:12,748
FROM MACHINE LEARNING IN

3351
02:10:12,748 --> 02:10:17,219
CHEMICAL SCIENCE ENGINEERING AND

3352
02:10:17,219 --> 02:10:17,519
MANY HOURS.

3353
02:10:17,519 --> 02:10:20,489
SO MILID, THE STAGE IS YOURS.

3354
02:10:20,489 --> 02:10:21,690
>> THANK YOU LEXY FOR THE
KIND

3355
02:10:21,690 --> 02:10:23,058
INTRODUCTION AND THANK YOU

3356
02:10:23,058 --> 02:10:24,026
ORGANIZERS FOR THE INVITATION TO

3357
02:10:24,026 --> 02:10:25,594
TALK ABOUT SOME OF THE WORK I'VE

3358
02:10:25,594 --> 02:10:27,596
BEEN DOING IN MY LAB FOR THE

3359
02:10:27,596 --> 02:10:32,935
PAST 7, 8 YEARS TO ACTUALLY WAY

3360
02:10:32,935 --> 02:10:36,805
THE APPROACH AND READING HUMAN

3361
02:10:36,805 --> 02:10:37,606
KNOWLEDGE, AI AND ROBOTICS, SO

3362
02:10:37,606 --> 02:10:41,043
WE ARE LOOKING FOR A NEW

3363
02:10:41,043 --> 02:10:44,513
MATERIAL MOLECULE, NO LONGER

3364
02:10:44,513 --> 02:10:45,347
DEALING WITH THIS UNIVERSE, WE

3365
02:10:45,347 --> 02:10:47,349
WILL HAVE A MAP OF THE UNIVERSE,

3366
02:10:47,349 --> 02:10:48,650
WE KNOW THE TARGET PROPERS OF

3367
02:10:48,650 --> 02:10:50,018
THE MATERIAL WE'RE LOOKING FOR,

3368
02:10:50,018 --> 02:10:51,887
WE DON'T KNOW HOW TO PFAFFIGATE

3369
02:10:51,887 --> 02:10:53,889
IT, OR FIND IT, WE DON'T EVEN

3370
02:10:53,889 --> 02:10:55,057
KNOW WHAT THAT MOLECULE IS, SO

3371
02:10:55,057 --> 02:10:57,493
MOST OF THE TIME WE'RE JUST

3372
02:10:57,493 --> 02:10:59,528
BLINDLY SEARCHING THROUGH THIS

3373
02:10:59,528 --> 02:11:02,097
SPACE WITH THE AEROSPACE

3374
02:11:02,097 --> 02:11:03,832
TECHNIQUE, SO THAT'S 50% OF THE

3375
02:11:03,832 --> 02:11:06,802
REASON WHY, THE MOLECULAR DORPHY

3376
02:11:06,802 --> 02:11:08,670
EFFORTS ARE [INDISCERNIBLE]

3377
02:11:08,670 --> 02:11:09,238
EXPENSIVE.

3378
02:11:09,238 --> 02:11:10,606
THE OTHER 50% OF THE PROBLEM IS

3379
02:11:10,606 --> 02:11:12,107
THE WAY WE NAVIGATE THIS

3380
02:11:12,107 --> 02:11:14,042
CHEMICAL UNIVERSE, THIS IS THE

3381
02:11:14,042 --> 02:11:17,513
WAY THE COMMERCIAL APPROACH

3382
02:11:17,513 --> 02:11:18,714
[INDISCERNIBLE] SCIENCES HOW YOU

3383
02:11:18,714 --> 02:11:21,316
TRAIN SCIENTISTS IN THE LAB,

3384
02:11:21,316 --> 02:11:23,118
THEY SPEND 70-80% OF THE THEIR

3385
02:11:23,118 --> 02:11:25,320
TIME DOING TASKS, PIPETTING,

3386
02:11:25,320 --> 02:11:27,756
READING SAMPLES, RUNNING

3387
02:11:27,756 --> 02:11:29,691
REACTIONS AND CHARACTERIZING

3388
02:11:29,691 --> 02:11:31,760
MAYBE 10 OR 20 SAMPLES A DAY, AT

3389
02:11:31,760 --> 02:11:32,795
THE END OF THE DAY ISSUE THEY GO

3390
02:11:32,795 --> 02:11:33,829
BACK TO THEIR OFFICE AND LOOK

3391
02:11:33,829 --> 02:11:35,931
THEA THE DATA AND USE THEIR

3392
02:11:35,931 --> 02:11:37,099
INTUITION TO DECIDE WHAT IS THE

3393
02:11:37,099 --> 02:11:39,101
NEXT EXPERIMENT TO RUN, SO THIS

3394
02:11:39,101 --> 02:11:40,502
IS OUR CONVENTIONAL CLOSED LOOP

3395
02:11:40,502 --> 02:11:43,038
SCIENCE AND IT'S VERY SLOW AND

3396
02:11:43,038 --> 02:11:43,472
RESOURCE INTENSIVE.

3397
02:11:43,472 --> 02:11:46,341
NOW IMAGINE, DEALING WITH HIGH

3398
02:11:46,341 --> 02:11:47,543
DIMENSIONAL PROBLEMS, 10, 20

3399
02:11:47,543 --> 02:11:48,343
DIFFERENT PARAMETERS YOU'RE

3400
02:11:48,343 --> 02:11:50,112
DEALING WITH, THE APPROACH THAT

3401
02:11:50,112 --> 02:11:51,680
WE TYPICALLY TAKE IS 1 PARAMETER

3402
02:11:51,680 --> 02:11:53,549
AT A TIME CAN 1 EXPERIMENT AT A

3403
02:11:53,549 --> 02:11:54,383
TIME.

3404
02:11:54,383 --> 02:11:58,620
SO THIS IS THE OTHER 50% OF THE

3405
02:11:58,620 --> 02:11:59,955
PROBLEM, MAIN SYMMETRY, IS OF

3406
02:11:59,955 --> 02:12:01,657
COURSE IS VERY, VERY EXPENSE AND

3407
02:12:01,657 --> 02:12:02,257
I HAVE SLOW.

3408
02:12:02,257 --> 02:12:03,292
SO LOOKINGA THE THIS PICTURE

3409
02:12:03,292 --> 02:12:05,594
THERE'S A LOT OF ROOM FOR

3410
02:12:05,594 --> 02:12:07,262
IMPROVEMENT, FOR EXAMPLE, WE CAN

3411
02:12:07,262 --> 02:12:08,664
LEVERAGE THE ADVANCEMENTS IN

3412
02:12:08,664 --> 02:12:10,032
ROBOTICS AND LAB AUTOMATION TO

3413
02:12:10,032 --> 02:12:11,200
BRING OUR HUMAN SCIENTISTS OUT

3414
02:12:11,200 --> 02:12:14,069
OF THE LAB, NOT OUT OF THE LOOP

3415
02:12:14,069 --> 02:12:15,804
OF EXPERIMENTAL SCIENCES, AND

3416
02:12:15,804 --> 02:12:18,240
THEN, HAVE THIS END-TO-END

3417
02:12:18,240 --> 02:12:19,208
AUTOMATED EXPERIMENTAL WORK

3418
02:12:19,208 --> 02:12:21,310
FLOW, INCREASE THROUGH PUT AND

3419
02:12:21,310 --> 02:12:22,144
REPRODUCIBILITY AND RELIABILITY

3420
02:12:22,144 --> 02:12:25,848
OF EXPERIMENTS AND THEN USE

3421
02:12:25,848 --> 02:12:26,915
RECENT ADVANCE WANTS IN MACHINE

3422
02:12:26,915 --> 02:12:28,817
LEARNING AND AI TO BETTER GUIDE

3423
02:12:28,817 --> 02:12:31,019
THE DECISION MAKING PROGRESS,

3424
02:12:31,019 --> 02:12:33,488
BETTER SAMPLE THE INFINITE

3425
02:12:33,488 --> 02:12:34,990
CHEMICAL UNIVERSE AND HAVE MORE

3426
02:12:34,990 --> 02:12:36,491
INFORMED SAMPLING OF THAT

3427
02:12:36,491 --> 02:12:38,427
REACTION IN SPACE.

3428
02:12:38,427 --> 02:12:40,495
SO WE INTEGRATE ROBOTICS

3429
02:12:40,495 --> 02:12:41,864
AUTOMATION WITH AN AI AGENT, YOU

3430
02:12:41,864 --> 02:12:43,999
CREATE WHAT WE CALL A

3431
02:12:43,999 --> 02:12:44,933
SELF-THRIVING LAB.

3432
02:12:44,933 --> 02:12:49,705
SO NOW WHAT YOU GET OUT OF THE

3433
02:12:49,705 --> 02:12:50,405
SELF-DRIVING LAB, IS

3434
02:12:50,405 --> 02:12:51,273
ACCELERATION, CONVENTIONAL STAT

3435
02:12:51,273 --> 02:12:53,742
EDGYS AND WE CALL THIS A ROBOTIC

3436
02:12:53,742 --> 02:12:56,478
CO PILOT, LIKE THE AI CO PILOT,

3437
02:12:56,478 --> 02:12:59,648
IT'S ROBOTIC CO PILOT, BRIDGES

3438
02:12:59,648 --> 02:13:03,218
THE GAP BETWEEN THE PHYSICAL

3439
02:13:03,218 --> 02:13:03,819
WORK.

3440
02:13:03,819 --> 02:13:08,023
BUT MOST LABS ARE MOST

3441
02:13:08,023 --> 02:13:08,957
SUSTAINABLE, SO REDUCE YOUR

3442
02:13:08,957 --> 02:13:11,126
FOOTPRINT OF THE DISCOVERY AND

3443
02:13:11,126 --> 02:13:13,195
IF YOU REDUCE YOUR PROCESSES,

3444
02:13:13,195 --> 02:13:15,063
CAN YOU REDUCE THAT TIME FROM

3445
02:13:15,063 --> 02:13:16,231
LAB TO MARKET.

3446
02:13:16,231 --> 02:13:18,033
SO DEPENDING ON HOW YOU

3447
02:13:18,033 --> 02:13:19,868
INTEGRATE HUMAN KNOWLEDGE OF

3448
02:13:19,868 --> 02:13:22,437
THESE ROBOTICS AND AUTOMATION,

3449
02:13:22,437 --> 02:13:24,907
IT CAN HAVE DIFFERENT HUMAN AI

3450
02:13:24,907 --> 02:13:26,441
ROBOT COLLABORATIONS IN YOUR

3451
02:13:26,441 --> 02:13:27,976
EXPERIMENTAL DISCOVERY WORK

3452
02:13:27,976 --> 02:13:38,086
FLOW, YOU CAN START FROM THE

3453
02:13:38,086 --> 02:13:44,259
PEACEFUL OPERATION WITH THE NEW

3454
02:13:44,259 --> 02:13:47,362
IRNLT VENTION AND HYPOTHESIS AND

3455
02:13:47,362 --> 02:13:48,597
EVOLVES, OVER TIME AND WHERE WE

3456
02:13:48,597 --> 02:13:50,766
ARE NOW, I'M GOING TO TALK ABOUT

3457
02:13:50,766 --> 02:13:53,468
TODAY IS THIS CLUES OF

3458
02:13:53,468 --> 02:13:54,436
AUTONOMOUS LABS THAT THE HUMAN

3459
02:13:54,436 --> 02:13:56,405
SCIENTIST IS STILL THE DRIVER

3460
02:13:56,405 --> 02:13:57,706
AND THE SELF-DRIVING LAB IS THE

3461
02:13:57,706 --> 02:13:59,675
CO PILOT THAT BRIDGES THE GAP

3462
02:13:59,675 --> 02:14:00,776
BETWEEN THE DIGITAL WORLD AND

3463
02:14:00,776 --> 02:14:01,677
THE PHYSICAL WORLD.

3464
02:14:01,677 --> 02:14:04,246
SO THE DIGITAL WORLD IS YOUR

3465
02:14:04,246 --> 02:14:06,415
CHEMICAL FORMATSICS AND THAT YOU

3466
02:14:06,415 --> 02:14:10,752
CAN DO INSILICO SCREENING AND

3467
02:14:10,752 --> 02:14:11,753
MAYBE OPTIMIZATION BUT THAT

3468
02:14:11,753 --> 02:14:14,222
INFORMATION IS STILL UNDER

3469
02:14:14,222 --> 02:14:14,957
DIGITAL AWARD.

3470
02:14:14,957 --> 02:14:20,429
SO THIS THE SELF-DRIVING  LAB

3471
02:14:20,429 --> 02:14:22,764
TAKES THAT AND TESTS THAT IN A

3472
02:14:22,764 --> 02:14:25,200
CLOSED LOOP FORMAT, AND VERIFYS

3473
02:14:25,200 --> 02:14:26,835
THE HYPOTHESIS AND MAYBE REJECTS

3474
02:14:26,835 --> 02:14:28,837
IT OR COMES UP WITH A NEW

3475
02:14:28,837 --> 02:14:30,339
HYPOTHESIS AND ONCE YOU GATE AT

3476
02:14:30,339 --> 02:14:32,040
THE END OF THE DAY THERE'S A LOT

3477
02:14:32,040 --> 02:14:34,977
OF BENEFITS WITH THE ROBOTIC

3478
02:14:34,977 --> 02:14:37,312
PART AND REDUCE TIME SOLUTION

3479
02:14:37,312 --> 02:14:38,680
BUT HUNDRED TO 1000 TIME GOING

3480
02:14:38,680 --> 02:14:41,483
FROM 10 PLUS YEARS AND WITH THE

3481
02:14:41,483 --> 02:14:42,484
MOLECULAR DISCOVERY DOWN TO A

3482
02:14:42,484 --> 02:14:47,522
FEW MONTH ANDS GOING FROM A

3483
02:14:47,522 --> 02:14:49,124
HUNDRED MILLION DOLLARS TO LESS

3484
02:14:49,124 --> 02:14:50,192
THAN A MILLION DOLLARS DOWN FOR

3485
02:14:50,192 --> 02:14:53,428
YOUR DISCOVERY AND THEN WITH THE

3486
02:14:53,428 --> 02:14:55,297
ROBOTIC CO PILOT, YOU CAN HAVE

3487
02:14:55,297 --> 02:15:05,774
THE PRECURSOR   FORMULATION.

3488
02:15:10,479 --> 02:15:11,680
SO JUST DOING WHAT WE'VE DON

3489
02:15:11,680 --> 02:15:13,849
DOING FOR THE PAST CENTURY USING

3490
02:15:13,849 --> 02:15:16,318
FLAS AND HAVING A LOT OF LACE IN

3491
02:15:16,318 --> 02:15:17,519
THE CHEMICALS WE ARE USING THIS

3492
02:15:17,519 --> 02:15:21,256
IS AN OPPORTUNITY TO THINK ABOUT

3493
02:15:21,256 --> 02:15:22,057
[INDISCERNIBLE], CHEMICAL

3494
02:15:22,057 --> 02:15:23,258
COMSUMPTION TRYING TO REENGINEER

3495
02:15:23,258 --> 02:15:25,594
DESIGN OF OUR AUTOMATION

3496
02:15:25,594 --> 02:15:27,929
PLATFORMS FOR THE MARKER DORPHY

3497
02:15:27,929 --> 02:15:30,032
WITH THIS PERSPECTIVE OF

3498
02:15:30,032 --> 02:15:30,766
NEUTRALIZING THE CHEMICAL

3499
02:15:30,766 --> 02:15:38,440
COMSUMPTION SO WE CAN MINIMIZE

3500
02:15:38,440 --> 02:15:40,542
THE [INDISCERNIBLE] AH

3501
02:15:40,542 --> 02:15:41,143
TONMATIONS.

3502
02:15:41,143 --> 02:15:42,644
NOT EVERY AUTOMATION SOLUTION IS

3503
02:15:42,644 --> 02:15:42,944
PRECISE.

3504
02:15:42,944 --> 02:15:44,679
WE HAVE TO MAKE SURE THE DATA WE

3505
02:15:44,679 --> 02:15:47,749
ARE GENERATING THROUGH THIS

3506
02:15:47,749 --> 02:15:48,350
AUTOMATION TECHNOLOGIES ARE

3507
02:15:48,350 --> 02:15:49,184
PRECISELY REPRODUCIBLE BECAUSE

3508
02:15:49,184 --> 02:15:50,685
THAT'S THE DATA WE WILL PROVIDE

3509
02:15:50,685 --> 02:15:52,888
TO THE AI AGENT.

3510
02:15:52,888 --> 02:15:54,156
THE AI AGENTS TRAIN UNDER THE

3511
02:15:54,156 --> 02:15:55,690
DATA YOU ARE GENERATING IN THE

3512
02:15:55,690 --> 02:15:59,261
LAB AND IF YOU HAVE NOISY DATA,

3513
02:15:59,261 --> 02:16:00,362
INSTEAD OF ACCELERATING SCIENCE,

3514
02:16:00,362 --> 02:16:02,197
YOU END UP DECELERATING YOUR

3515
02:16:02,197 --> 02:16:03,832
CONSCIENCE AND YOU HAVE TO THINK

3516
02:16:03,832 --> 02:16:05,367
ABOUT WHAT ARE THE MOST

3517
02:16:05,367 --> 02:16:06,001
APPROPRIATE TECHNIQUES THAT

3518
02:16:06,001 --> 02:16:07,602
PROVIDE THROUGH PUT THAT IS

3519
02:16:07,602 --> 02:16:08,770
NEEDED TO PROVIDE DAILY BASIS

3520
02:16:08,770 --> 02:16:10,605
THEA FOR THAT AI AGENT.

3521
02:16:10,605 --> 02:16:11,907
AND AT THE END OF THE DAY, WE

3522
02:16:11,907 --> 02:16:14,209
WHAT WE WANT TO HAVE IS THE

3523
02:16:14,209 --> 02:16:14,843
RELIABLE EXPERIMENTAL DATA AND

3524
02:16:14,843 --> 02:16:17,446
WE HAVE TO THINK ABOUT HOW DO WE

3525
02:16:17,446 --> 02:16:18,346
ACTUALLY TRANSLATE KNOWLEDGE

3526
02:16:18,346 --> 02:16:20,082
BETWEEN DIFFERENT MODALITIES OF

3527
02:16:20,082 --> 02:16:21,650
RUNNING REACTIONS, DIFFERENT

3528
02:16:21,650 --> 02:16:23,118
TYPES OF HARDWARE, DIFFERENT

3529
02:16:23,118 --> 02:16:26,721
ROBOTICS, YOU CAN RUN THE SAME

3530
02:16:26,721 --> 02:16:29,691
ORGANIC SYNTHESIS USING

3531
02:16:29,691 --> 02:16:30,659
[INDISCERNIBLE] FLASK OR

3532
02:16:30,659 --> 02:16:32,894
[INDISCERNIBLE], OR NEUTRALIZED

3533
02:16:32,894 --> 02:16:33,628
BATCH REACTORS.

3534
02:16:33,628 --> 02:16:35,163
NOW DEPENDING ON THE ENVIRONMENT

3535
02:16:35,163 --> 02:16:37,399
REAC, YOU MIGHT GET DIFFERENT

3536
02:16:37,399 --> 02:16:42,637
RESULTS, SO  HOW DO YOU AUGMENT

3537
02:16:42,637 --> 02:16:43,705
THESE DATA, FROM 1 LAB TO

3538
02:16:43,705 --> 02:16:44,272
ANOTHER 1.

3539
02:16:44,272 --> 02:16:46,575
SO WE COME UP WITH LANGUAGE SO

3540
02:16:46,575 --> 02:16:49,377
THEY CAN TRANSFER KNOWLEDGE AND

3541
02:16:49,377 --> 02:16:50,745
AUGMENT DATA AND ACCESS

3542
02:16:50,745 --> 02:16:51,480
DISCOVERIES AND ALL THOSE THINGS

3543
02:16:51,480 --> 02:16:54,583
TO CONSIDER WHEN WE THINK ABOUT

3544
02:16:54,583 --> 02:16:56,318
THESE WHEN DESIGNING AN

3545
02:16:56,318 --> 02:16:56,685
AUTONOMOUS LAB.

3546
02:16:56,685 --> 02:17:00,856
SO THINK ABOUT THIS FOR DATA IN

3547
02:17:00,856 --> 02:17:03,058
MY LAB, WHICH IS THE GOLD

3548
02:17:03,058 --> 02:17:06,695
STANDARD TO USE, THE FLOW

3549
02:17:06,695 --> 02:17:08,363
REACTORS BASICALLY MICROFLUIDIC

3550
02:17:08,363 --> 02:17:09,564
REACTOR, IT'S BEEN AROUND

3551
02:17:09,564 --> 02:17:12,400
FOREMORTHAN 30 YEARS BUT

3552
02:17:12,400 --> 02:17:13,368
BASICALLY CONTINUOUSLY FLOWING

3553
02:17:13,368 --> 02:17:14,436
THE PRECOURSORS OF YOUR

3554
02:17:14,436 --> 02:17:15,537
REACTIONS AND THE REACTIONS ARE

3555
02:17:15,537 --> 02:17:17,005
HAPPENING AT A VERY SMALL

3556
02:17:17,005 --> 02:17:18,473
CHANNEL AND THE REACTION HAPPENS

3557
02:17:18,473 --> 02:17:20,075
FOR A SPECIFIC REACTION TIME AND

3558
02:17:20,075 --> 02:17:21,676
AT A SPECIFIC TEMPERATURE AND

3559
02:17:21,676 --> 02:17:23,011
THATIA A TIME THAT YOU'RE

3560
02:17:23,011 --> 02:17:26,381
REGIONS ARE SPENDING IN A VERY

3561
02:17:26,381 --> 02:17:34,322
SMALL MICROCHANNEL THAT THEY'RE

3562
02:17:34,322 --> 02:17:34,656
FLOWING.

3563
02:17:34,656 --> 02:17:37,025
AND 1 OF THE ADVANTAGES, THAT

3564
02:17:37,025 --> 02:17:39,594
YOU HAVE, SO, IN BASICALLY A

3565
02:17:39,594 --> 02:17:41,463
SIMPLE AWORDS WE CAN NEUTRALIZE

3566
02:17:41,463 --> 02:17:43,899
REACTIONS WHEN YOU RUN THEM IN A

3567
02:17:43,899 --> 02:17:45,634
FLUID REACTOR YOU CAN ACCESS

3568
02:17:45,634 --> 02:17:47,769
KINETICS OF A REACTION, SO IF A

3569
02:17:47,769 --> 02:17:49,971
REACTION TAKES HOURS OR DAYS IN

3570
02:17:49,971 --> 02:17:52,974
A BATCH REACTOR AND HEAT

3571
02:17:52,974 --> 02:17:54,009
TRANSMITTED FLUID, IT RUN THE

3572
02:17:54,009 --> 02:17:57,445
SAME IN A MANNER OF MINUTES IN A

3573
02:17:57,445 --> 02:17:58,647
MICROFLUIDIC REACTOR.

3574
02:17:58,647 --> 02:18:00,882
SO IT RUN HUNDREDS OF REACTIONS

3575
02:18:00,882 --> 02:18:02,517
AT 3 OR 4 ORDINANCE NUMBERERS OF

3576
02:18:02,517 --> 02:18:04,352
MAGNITUDE IN A BATCH REACTOR AND

3577
02:18:04,352 --> 02:18:07,789
THEN WE CAN PRECISELY AUTOMATE

3578
02:18:07,789 --> 02:18:09,558
THE REACTION, CONTROL ALL THE

3579
02:18:09,558 --> 02:18:10,859
REACTION CONDITIONS AND

3580
02:18:10,859 --> 02:18:12,494
INTEGRATE DIFFERENT ONLINE

3581
02:18:12,494 --> 02:18:13,061
CHARACTERIZATION THROUGH

3582
02:18:13,061 --> 02:18:17,465
CHARACTERIZE THE COMPOUND THAT

3583
02:18:17,465 --> 02:18:17,899
LEAVES THE REACTOR.

3584
02:18:17,899 --> 02:18:22,237
SO THESE ARE SOME OF THE MOST

3585
02:18:22,237 --> 02:18:24,206
FLUIDIC ADVANTAGES WHEN WE WERE

3586
02:18:24,206 --> 02:18:34,482
TALKING ABOUT THE PREVIOUS SLIDE

3587
02:18:34,482 --> 02:18:40,522
FOR THE HL MA  MAPS.

3588
02:18:40,522 --> 02:18:43,158
AND THIS MAY BE HALF A

3589
02:18:43,158 --> 02:18:45,026
MICROLITER UP TO 1 OR 2 MICROLIT

3590
02:18:45,026 --> 02:18:45,227
EARS.

3591
02:18:45,227 --> 02:18:48,630
AND YOU HAVE A AUTOFORMAT WE

3592
02:18:48,630 --> 02:18:49,965
CALL A SINGLE DROPLET, MULTIFAYE

3593
02:18:49,965 --> 02:18:53,735
I FLOW IN SIMPLE WORD WE HAVE 1

3594
02:18:53,735 --> 02:18:55,570
LIQUID DROPLET OR PLUG AND WE

3595
02:18:55,570 --> 02:18:57,439
ARE OSCILLATING THIS BACK AND

3596
02:18:57,439 --> 02:18:58,807
FORTH, IT'S LIKE SHAKING A VILE

3597
02:18:58,807 --> 02:19:02,677
AND IN THIS CASE YOU CAN HAVE A

3598
02:19:02,677 --> 02:19:03,311
SMALL REACTIVE FOOTPRINT, THEY

3599
02:19:03,311 --> 02:19:05,981
CAN HAVE A DROPLET OF 2-4 AND

3600
02:19:05,981 --> 02:19:07,682
RUN REACTIONS IN THE ORDER OF

3601
02:19:07,682 --> 02:19:10,385
SECONDS, DAYS OR HOURS OR MAYBE

3602
02:19:10,385 --> 02:19:12,487
EVEN MONTHS SO IF YOU'RE DEALING

3603
02:19:12,487 --> 02:19:14,189
RELATIVELY SMOA CLEM CHAL

3604
02:19:14,189 --> 02:19:15,724
REACTION, WE TYPICALLY USE THE

3605
02:19:15,724 --> 02:19:17,459
SINGLE DROPLET FORM AT IF WE ARE

3606
02:19:17,459 --> 02:19:18,927
DOLING WITH RELATIVELY FAST

3607
02:19:18,927 --> 02:19:20,262
REACTIONS OR DO CONTINUOUS

3608
02:19:20,262 --> 02:19:21,963
MANUFACTURING WE USE THE

3609
02:19:21,963 --> 02:19:22,497
CONTINUOUS MULTIFACE FLOW.

3610
02:19:22,497 --> 02:19:26,468
SO IN MY LAB, THEY DEVELOP THIS

3611
02:19:26,468 --> 02:19:29,271
BESPOKE FLOW REACTORS AND USE

3612
02:19:29,271 --> 02:19:29,537
ROBOTICS.

3613
02:19:29,537 --> 02:19:32,073
WE USE LIQUID [INDISCERNIBLE]

3614
02:19:32,073 --> 02:19:34,809
ROBOTS, AND THE SOLUTIONS OF THE

3615
02:19:34,809 --> 02:19:38,446
REACTIONS AND THEN THE DEVELOPER

3616
02:19:38,446 --> 02:19:43,618
ON BESPOKE ONLINE

3617
02:19:43,618 --> 02:19:44,219
CHARACTERIZATION TOOLS,

3618
02:19:44,219 --> 02:19:45,220
CHROMEATOGRAPHYS AND MASS SPEC

3619
02:19:45,220 --> 02:19:47,956
AND THEN USING A AI EAMENT TO

3620
02:19:47,956 --> 02:19:49,491
KRIEF THIS TO THE EXPERIMENTAL

3621
02:19:49,491 --> 02:19:51,326
WORK FLOW, BUT ON A THE

3622
02:19:51,326 --> 02:19:52,193
HIGHLIGHT HERE IS THE

3623
02:19:52,193 --> 02:19:53,361
SELF-DRIVING LAB IS NOT

3624
02:19:53,361 --> 02:19:54,329
REPLACING THE HUMAN SCIENTIST IN

3625
02:19:54,329 --> 02:19:56,264
THE LAB, IT IS ACTING AS A

3626
02:19:56,264 --> 02:20:05,807
ROBOTIC CO PILOT FOR THE HUMAN

3627
02:20:05,807 --> 02:20:06,975
SCIENTISTS AND AUTOSTILL

3628
02:20:06,975 --> 02:20:08,610
PROVIDING THE FRAME FOR THE

3629
02:20:08,610 --> 02:20:10,545
ACCESS EMPLOY AND WHEN WOE USE

3630
02:20:10,545 --> 02:20:12,113
THIS SELF-DRIVING LAB FOR THE

3631
02:20:12,113 --> 02:20:13,815
TECHNOLOGY, MY LAB APPLIED 3

3632
02:20:13,815 --> 02:20:15,517
DIFFERENT RESEARCHIARYS, 1 IS

3633
02:20:15,517 --> 02:20:17,319
NOW SHY, AND THE THIRD RESEARCH

3634
02:20:17,319 --> 02:20:19,321
TRANSFER WE USE THE DATA MACHINE

3635
02:20:19,321 --> 02:20:20,889
LEARNING MODELS WE DEVELOP IN

3636
02:20:20,889 --> 02:20:23,625
THE AREAS TO BETTER UNDERSTAND A

3637
02:20:23,625 --> 02:20:25,560
FUNDAMENTAL MECHANISMS OF THE

3638
02:20:25,560 --> 02:20:26,161
REACTIONS AND PROCESSES

3639
02:20:26,161 --> 02:20:30,665
HAPPENING AND FOR EXAMPLE, WE

3640
02:20:30,665 --> 02:20:32,801
CAN INSTALL FORMATION OR THE

3641
02:20:32,801 --> 02:20:34,302
TRANSITION MINNIAL REACTIONS AND

3642
02:20:34,302 --> 02:20:35,437
THE CATALYSIS AREA THAT WE'RE

3643
02:20:35,437 --> 02:20:35,904
STUDYING.

3644
02:20:35,904 --> 02:20:39,474
SO FOR THIS PRESENTATION, I WILL

3645
02:20:39,474 --> 02:20:40,909
FOCUS ON THE CATALIAISONS AREA,

3646
02:20:40,909 --> 02:20:43,345
THIS IS HOW WE CAN USE ROBOTICS

3647
02:20:43,345 --> 02:20:45,747
TO ACCELERATE OUR EXPERIMENTAL

3648
02:20:45,747 --> 02:20:49,784
FORK FLOW, SO HERE AS A CUSTOM

3649
02:20:49,784 --> 02:20:53,988
BUILT ROBOTIC SOLUTION INSIDE A

3650
02:20:53,988 --> 02:20:55,190
ENVIRONMENT, CAN AUTOMATICALLY

3651
02:20:55,190 --> 02:20:58,326
REDO SOLID AND FLUIDIS PENSING,

3652
02:20:58,326 --> 02:21:02,630
ORGANIC REACTIONS FOR EXAMPLE,

3653
02:21:02,630 --> 02:21:04,766
CCAND CROSS [INDISCERNIBLE]

3654
02:21:04,766 --> 02:21:08,036
REACTIONS THAT CAN ACTIVATE

3655
02:21:08,036 --> 02:21:09,104
REAGENTS AND STARTING PRECURSOR

3656
02:21:09,104 --> 02:21:11,806
ANDS START RUNNING THE INITIAL

3657
02:21:11,806 --> 02:21:14,109
DISCREET, SCREENING OF THE

3658
02:21:14,109 --> 02:21:15,477
DIFFERENT PARAMETERS, DIFFERENT

3659
02:21:15,477 --> 02:21:16,244
[INDISCERNIBLE] AND GIVING--YOU

3660
02:21:16,244 --> 02:21:18,046
LAND SOLVENTS AND BASE.

3661
02:21:18,046 --> 02:21:20,181
AND THEN THEY CAN BASICALLY

3662
02:21:20,181 --> 02:21:21,516
SELECT OR FIND THE HIT

3663
02:21:21,516 --> 02:21:24,519
CONDITIONS AND THEN RUN THOSE IN

3664
02:21:24,519 --> 02:21:26,287
OUR FLOW REACTORS FOR FINE TUNE

3665
02:21:26,287 --> 02:21:29,090
OPTIMIZATION OF THE REACTION

3666
02:21:29,090 --> 02:21:29,391
CONDITIONS.

3667
02:21:29,391 --> 02:21:30,925
AND THEN ONCE THIS IS FINISHED

3668
02:21:30,925 --> 02:21:33,128
WE USE MOBILE AND STATIONARY

3669
02:21:33,128 --> 02:21:36,598
ROBOTIC ARMS TO TRANSFER SAMPLES

3670
02:21:36,598 --> 02:21:38,233
TO DIFFERENT STATIONS FOR DOWN

3671
02:21:38,233 --> 02:21:39,000
STREAM PROCESSING FOR EXAMPLE,

3672
02:21:39,000 --> 02:21:40,635
IF YOU DO ANY SORT OF BI ONY

3673
02:21:40,635 --> 02:21:44,005
ASSAYS THEY CAN DO THE TRANSFER

3674
02:21:44,005 --> 02:21:45,740
FOR BIOASSAY STATIONS SO WE CAN

3675
02:21:45,740 --> 02:21:49,377
HAVE THESE 24/7 OPERATION OF OUR

3676
02:21:49,377 --> 02:21:50,879
SELF-GRINDING LABS, BUT AS I

3677
02:21:50,879 --> 02:21:51,946
MENTIONED WE USE FLOW REACTORS

3678
02:21:51,946 --> 02:21:53,915
TO RUN THE SYNTHESIS SO WE CAN

3679
02:21:53,915 --> 02:21:55,750
DO HUNDREDS OR THOUSANDS OF

3680
02:21:55,750 --> 02:21:57,218
DIFFERENT REACTIONS FOR DATA

3681
02:21:57,218 --> 02:21:59,988
THAT'S JUST 2 EXAMPLE, WE USE OF

3682
02:21:59,988 --> 02:22:02,690
DROPLETS IN A VERY SMALL FLOW

3683
02:22:02,690 --> 02:22:04,325
REACTORS, AND THEN THE DIAMETER

3684
02:22:04,325 --> 02:22:08,596
OF TUBE HERE IS 750-MICRONS, AND

3685
02:22:08,596 --> 02:22:10,131
THE REACTION CAN HAPPEN

3686
02:22:10,131 --> 02:22:11,966
TYPICALLY IS AN ORDER OF HALF A

3687
02:22:11,966 --> 02:22:13,368
MICROLITER, AND AS I SAID WE CAN

3688
02:22:13,368 --> 02:22:17,071
DO UP TO A THOUSAND REACTIONS OR

3689
02:22:17,071 --> 02:22:17,906
ORGAGIC SYNTHESIS A DAY EMPLOY

3690
02:22:17,906 --> 02:22:20,175
SO WE ARE IN THE WORKING

3691
02:22:20,175 --> 02:22:21,509
SELF-DRIVING FIELDS FOR THE PAST

3692
02:22:21,509 --> 02:22:23,578
7 OR 8 YEARS AND THE EXAMPLES OF

3693
02:22:23,578 --> 02:22:26,915
THESE LABS WE DEVELOP BOTH FOR

3694
02:22:26,915 --> 02:22:27,916
INORGANIC MATERIALS AND FOR

3695
02:22:27,916 --> 02:22:28,583
ORGANIC SYNTHESIS.

3696
02:22:28,583 --> 02:22:39,127
SO I WILL TALK ABOUT CATALYSIS,

3697
02:22:40,261 --> 02:22:41,362
AND ORGANIC TRANSFORMATIONS IS

3698
02:22:41,362 --> 02:22:42,864
HOW EFFECTIVE IS THE LIGAND

3699
02:22:42,864 --> 02:22:44,899
YOU'RE USING IN OUR ORGANIC SING

3700
02:22:44,899 --> 02:22:49,471
SIS, FOR EXAMPLE, COUPLING,

3701
02:22:49,471 --> 02:22:50,472
FORMULATIONS OR ORGAN C

3702
02:22:50,472 --> 02:22:50,738
COUPLING.

3703
02:22:50,738 --> 02:22:54,309
SO WHEN YOU HAVE A LIGAND USING

3704
02:22:54,309 --> 02:22:55,577
CATALYSIS FOR EXAMPLE, YOU CAN

3705
02:22:55,577 --> 02:23:00,348
SELECT THE YIELD THE REACTION.

3706
02:23:00,348 --> 02:23:01,382
SO TYPICALLY WHAT CHEMISTS DO,

3707
02:23:01,382 --> 02:23:02,884
THEY HAVE A LIST OF CONDITIONS

3708
02:23:02,884 --> 02:23:06,521
THEY RUN SO THEY DO THE MANUEL

3709
02:23:06,521 --> 02:23:07,755
SCREENING, OR THE EXTERNAL DAT

3710
02:23:07,755 --> 02:23:12,293
THAT ARE WITHIN THE BOUNDARIES

3711
02:23:12,293 --> 02:23:14,896
OF OF THE SILICO IN YIELD, OR

3712
02:23:14,896 --> 02:23:16,531
THE BEST SOURCE OF THE LIGAND

3713
02:23:16,531 --> 02:23:18,366
YOU'RE TESTING AND THEN COMPARE

3714
02:23:18,366 --> 02:23:19,367
THE OPTIMAL POINTS THAT YOU

3715
02:23:19,367 --> 02:23:21,336
CAN'T IMPROVE THE YIELD OR

3716
02:23:21,336 --> 02:23:24,239
SELECTIVITY OF THE YIELD OR

3717
02:23:24,239 --> 02:23:25,940
LIGAND WITHOUT A SPECIFIC POINT

3718
02:23:25,940 --> 02:23:26,774
WITHOUT COMPROMISING THE OTHER 1

3719
02:23:26,774 --> 02:23:27,976
BUT WHAT YOU WANT TO HAVE IS A

3720
02:23:27,976 --> 02:23:29,711
METHOD THAT CAN GIVE YOU THIS

3721
02:23:29,711 --> 02:23:31,246
VERY TARGETED SELECTION OF

3722
02:23:31,246 --> 02:23:33,715
EXPERIMENTS THAT YOU CAN

3723
02:23:33,715 --> 02:23:34,482
MINIMIZE YOUR CHEMICAL

3724
02:23:34,482 --> 02:23:36,251
COMSUMPTION IS IN VERY SHORT

3725
02:23:36,251 --> 02:23:38,386
AMOUNT OF TIME SHOW THIS OPTIMAL

3726
02:23:38,386 --> 02:23:42,123
POINTS FOR A GIVEN LIGAND OR

3727
02:23:42,123 --> 02:23:44,726
CALLOUS, YOU SO KULAK AT THE

3728
02:23:44,726 --> 02:23:45,894
DORPHY FOR THE SYNTHESIS.

3729
02:23:45,894 --> 02:23:47,128
SO I WILL SHOW YOU IN THE

3730
02:23:47,128 --> 02:23:49,163
FIXTURE FEW SLIDES HOW WE DESIGN

3731
02:23:49,163 --> 02:23:51,966
THE SELF-DRIVING LAB THAT CAN

3732
02:23:51,966 --> 02:23:52,634
AUTONOMOUSLY TEST   PERFORMNESS

3733
02:23:52,634 --> 02:23:55,537
FOR THE LIGAND AND USE THAT

3734
02:23:55,537 --> 02:23:56,671
INFORMATION TO FURTHER ACCESS

3735
02:23:56,671 --> 02:23:58,840
DISCOVERY OF NEW LIGANDS FOR

3736
02:23:58,840 --> 02:23:59,374
ORGANIC SENTH SIS.

3737
02:23:59,374 --> 02:24:02,010
SO 1 PART OF THE SELF-DRIVING

3738
02:24:02,010 --> 02:24:03,244
LAB IS THE HARDWARE, AND I WILL

3739
02:24:03,244 --> 02:24:12,587
TALK ABOUT HARD REACTION, SO

3740
02:24:12,587 --> 02:24:14,722
FIRST PART IS HUMAN USER, THEN

3741
02:24:14,722 --> 02:24:17,559
AI GUIDED EXPERIMENTAL PLANNING

3742
02:24:17,559 --> 02:24:18,459
AND ROBOTIC RECONFIGURURATION

3743
02:24:18,459 --> 02:24:20,995
AND THEN ONLINE DATABASE, AND DO

3744
02:24:20,995 --> 02:24:22,997
SPRPTS IN ORDER TO INTELLIGENTEE

3745
02:24:22,997 --> 02:24:23,865
NAVIGATE THAT CHEMICAL THAT I

3746
02:24:23,865 --> 02:24:25,166
WAS TALKING ABOUT AT THE

3747
02:24:25,166 --> 02:24:26,234
BEGENERATEDDING OF MY SLIDES

3748
02:24:26,234 --> 02:24:28,336
MUCH SO IN TERMS OF REACTION, WE

3749
02:24:28,336 --> 02:24:31,239
USE 2 DIFFERENT TYPES OF

3750
02:24:31,239 --> 02:24:32,273
REACTORS, SINGLE DROPLET REACTOR

3751
02:24:32,273 --> 02:24:33,975
BUT THESE ARE MUTEERALLIZED

3752
02:24:33,975 --> 02:24:35,677
REACTORS THAT THEY ARE RUN

3753
02:24:35,677 --> 02:24:38,580
REACTIONS IN THIS HALF AIAN

3754
02:24:38,580 --> 02:24:40,682
MICROLITER AND THEN WE CAN

3755
02:24:40,682 --> 02:24:42,483
REDUCE THE CHEMICAL COMSUMPTION,

3756
02:24:42,483 --> 02:24:44,886
WITH THE KINETICS AT 5-10 TIMES

3757
02:24:44,886 --> 02:24:46,988
AND RUN REACTIONS UP TO 1000

3758
02:24:46,988 --> 02:24:55,229
ORGANIC SYNTHESIS A DAY.

3759
02:24:55,229 --> 02:24:58,099
SWE CAN RUN HIGH TEMPERATURE,

3760
02:24:58,099 --> 02:24:59,667
HIGH PRESSURE REACTIONS RELIABLY

3761
02:24:59,667 --> 02:25:00,668
IN THESE PLATFORMS AND WHAT

3762
02:25:00,668 --> 02:25:02,704
WE'VE DONE IN THE PAST WE CAN

3763
02:25:02,704 --> 02:25:04,172
SHOW ASK USE THESE PLATFORMS TO

3764
02:25:04,172 --> 02:25:07,008
RUN A RANGE OF DIFFERENT ORGANIC

3765
02:25:07,008 --> 02:25:10,044
SYNTHESIS FROM THIS TO COUPLING

3766
02:25:10,044 --> 02:25:11,846
AND SiRNA REACTIONS FOR THE

3767
02:25:11,846 --> 02:25:13,281
KACCTAL SIS AND CROSS SUPT G

3768
02:25:13,281 --> 02:25:14,849
REACTIONS AND THIS SHOWS THE

3769
02:25:14,849 --> 02:25:16,951
DIVERSITY OF DIFFERENT REACTIONS

3770
02:25:16,951 --> 02:25:18,386
ORGANIC 60 SIS THAT YOU CAN DO

3771
02:25:18,386 --> 02:25:20,188
THESE FLOW CHEMIST RADIOY

3772
02:25:20,188 --> 02:25:20,455
PLATFORMS.

3773
02:25:20,455 --> 02:25:23,391
SO HERE'S AN EXAMPLE OF THIS

3774
02:25:23,391 --> 02:25:25,860
FULLY AUTOMATED FRONTAL PLATFORM

3775
02:25:25,860 --> 02:25:26,794
USING ROBOTICS FOR LIQUID

3776
02:25:26,794 --> 02:25:36,871
HANDLING OF THE LEFT, AND

3777
02:25:36,871 --> 02:25:38,239
RUNNING THE ACTUAL REACTION, AND

3778
02:25:38,239 --> 02:25:42,176
DOING ANALYSIS ON THE RIGHT WITH

3779
02:25:42,176 --> 02:25:44,779
THE ONLINE ELFSD MASS SPEC TO

3780
02:25:44,779 --> 02:25:46,147
CHARACTERIZE THIS REACTION.

3781
02:25:46,147 --> 02:25:48,950
AND HERE'S HOW YOU CAN USE THIS

3782
02:25:48,950 --> 02:25:49,884
FOR THE PREPARATION AND THEN

3783
02:25:49,884 --> 02:25:51,119
THIS IS WHERE YOU RUN THE

3784
02:25:51,119 --> 02:25:52,887
REACTION IS IN A FEW

3785
02:25:52,887 --> 02:25:54,956
MICROLIGHTERS CAN YOU DO GAS OR

3786
02:25:54,956 --> 02:25:57,625
LIQUID REACTION, SINGLE PHASE OR

3787
02:25:57,625 --> 02:25:58,192
MULTIPACE REACS.

3788
02:25:58,192 --> 02:26:08,736
IT CAN ALSO COUPLE THIS REACTOR

3789
02:26:12,674 --> 02:26:17,011
WITH A DIFFERENT BAN GAPS FOR

3790
02:26:17,011 --> 02:26:17,979
THE TRANSFORMATION AND ONCE YOU

3791
02:26:17,979 --> 02:26:20,548
OPERATING GLOBALLY MYSELF

3792
02:26:20,548 --> 02:26:21,215
REACTION, USING AUTONOMOUS

3793
02:26:21,215 --> 02:26:22,784
ACTION, CAN YOU SCALE UP GOING

3794
02:26:22,784 --> 02:26:24,819
FROM SINGLE RAPID CONTINUOUS

3795
02:26:24,819 --> 02:26:26,888
FLOW AND ACKNOWLEDGES DIRECTLY

3796
02:26:26,888 --> 02:26:27,989
TRANSFERABLE, IT'S A SCALABLE

3797
02:26:27,989 --> 02:26:28,823
KNOWLEDGE THAT YOU DISCOVER AND

3798
02:26:28,823 --> 02:26:31,459
YOU DON'T NEED TO DO ANY

3799
02:26:31,459 --> 02:26:32,393
ADDITIONAL OPTIMIZATION, YOU IT

3800
02:26:32,393 --> 02:26:34,729
CAN JUST DIRECT THE TRANSFER THE

3801
02:26:34,729 --> 02:26:37,031
LIAISON IMAND CALLOUS AND THE

3802
02:26:37,031 --> 02:26:38,966
REACTION TO THE DISCOVER TO THE

3803
02:26:38,966 --> 02:26:42,670
FLOW REACTOR AND CAN PRODUCE THE

3804
02:26:42,670 --> 02:26:46,507
TARGET COMPOUND WITH THE OPTIMAL

3805
02:26:46,507 --> 02:26:47,141
CONDITIONS WE'VE DISCOVERED, SO

3806
02:26:47,141 --> 02:26:49,343
NOW THAT WAS THE HARDWARE

3807
02:26:49,343 --> 02:26:50,478
REACTION NEUTRALIZATION SITE OF

3808
02:26:50,478 --> 02:26:51,612
THE CELLS DRIVING LABS, THE

3809
02:26:51,612 --> 02:26:53,848
SECOND PART IS HOW YOU ACTUALLY

3810
02:26:53,848 --> 02:26:55,016
MAKE A DECISION WHILE

3811
02:26:55,016 --> 02:26:56,084
CONTRIBUTING TO THE EXPERIMENTAL

3812
02:26:56,084 --> 02:26:58,052
RUN, THIS IS WHERE WE DO A LOT

3813
02:26:58,052 --> 02:26:58,853
OF PROCESS OPTIMIZATION TO MAKE

3814
02:26:58,853 --> 02:27:01,723
SURE WE HAVE THE MOST OPTIMAL

3815
02:27:01,723 --> 02:27:04,559
FINE TUNE AI AGENT FOR

3816
02:27:04,559 --> 02:27:07,128
SELF-DRIVING LAB SO IN THIS CASE

3817
02:27:07,128 --> 02:27:08,963
I'M SHOWING THE FRAMEWORK WHERE

3818
02:27:08,963 --> 02:27:11,599
I WANTED TO IMPROVE, THE PAIR TO

3819
02:27:11,599 --> 02:27:13,034
OPTIMAL POINTS FOR THE FRONT OF

3820
02:27:13,034 --> 02:27:15,536
SELECTIVITY AND YIELD OF A

3821
02:27:15,536 --> 02:27:16,904
SPECIFIC CHEMICAL REACTION, IN

3822
02:27:16,904 --> 02:27:20,374
THIS CASE OUT OF WHOM YOU HAVE

3823
02:27:20,374 --> 02:27:21,142
ALL DIFFERENT FOSTER NURSED

3824
02:27:21,142 --> 02:27:22,510
FOCUSED ON LIGANDS TO PRODUCE

3825
02:27:22,510 --> 02:27:24,712
THE HIGHS WHICH ARE THE STARTING

3826
02:27:24,712 --> 02:27:26,180
CHEMICALS IF ARE A LOT OF

3827
02:27:26,180 --> 02:27:27,281
ALCOHOLS AND OTHER APIs.

3828
02:27:27,281 --> 02:27:29,550
SO IN THIS CASE, BASICALLY

3829
02:27:29,550 --> 02:27:31,419
THEY'RE USING HYPER VOLUME

3830
02:27:31,419 --> 02:27:33,154
IMPROVE AM OR EXPECT HYPER

3831
02:27:33,154 --> 02:27:34,889
VOLUME IMPROVEMENT AND YOU START

3832
02:27:34,889 --> 02:27:36,891
YOUR REACTION WITH

3833
02:27:36,891 --> 02:27:39,694
INITIALIZATION OF SOME EITHER

3834
02:27:39,694 --> 02:27:41,429
RANDOM SELECTED EXPERIMENTS OR

3835
02:27:41,429 --> 02:27:42,296
DESIGNING EXPERIMENTAL METHODS

3836
02:27:42,296 --> 02:27:44,198
TO USE THE VISUALIZATION OF YOUR

3837
02:27:44,198 --> 02:27:45,433
OPTIMIZATION AND THEN YOU CAN

3838
02:27:45,433 --> 02:27:49,270
MODEL THE FUTURE EXPERIMENTS AND

3839
02:27:49,270 --> 02:27:50,972
ROUTE THEM BASED ON EXPECTED

3840
02:27:50,972 --> 02:27:52,707
HYPER VOLUME IMPROVEMENT WHICH

3841
02:27:52,707 --> 02:27:54,342
IS BASICALLY IN SIMPLE TERMS

3842
02:27:54,342 --> 02:27:55,376
WHAT IS THE PROBABILITY OF

3843
02:27:55,376 --> 02:27:56,677
RUNNING THE EXPERIMENT IN THE

3844
02:27:56,677 --> 02:27:58,813
FUTURE TO IMPROVE THE YIELD AND

3845
02:27:58,813 --> 02:27:59,814
SELECTIVITY OF THIS REACTION IN

3846
02:27:59,814 --> 02:28:03,551
AND THEN WE SELECTED TOP RANK

3847
02:28:03,551 --> 02:28:05,086
EXPERIMENTS, TO THE HARDWARE

3848
02:28:05,086 --> 02:28:07,889
OURSELVES TO RUN THAT REACTION

3849
02:28:07,889 --> 02:28:09,023
AND UPDATE THE SURROGATE MODAND

3850
02:28:09,023 --> 02:28:10,324
HE WILL RUN THROUGH THIS LOOP

3851
02:28:10,324 --> 02:28:11,559
UNTIL THEY CANNOT FURTHER

3852
02:28:11,559 --> 02:28:14,262
IMPROVE THAT PAIR TO OPTIMAL

3853
02:28:14,262 --> 02:28:14,462
POINT.

3854
02:28:14,462 --> 02:28:16,430
NOW IN ORDER TO FINE TUNE THE AI

3855
02:28:16,430 --> 02:28:17,832
AGENT OF THE SELF-DRIVING LAB,

3856
02:28:17,832 --> 02:28:20,601
WE NEED A LOT OF MODELING HERE

3857
02:28:20,601 --> 02:28:22,503
TO MAKE SURE THE BRAIN OF THE

3858
02:28:22,503 --> 02:28:23,638
SELF-DRIVING LAB IS FINE TUNED

3859
02:28:23,638 --> 02:28:26,274
AND IT'S THE MOST OPTIMAL

3860
02:28:26,274 --> 02:28:28,676
MODELING AND THE DECISION MAKING

3861
02:28:28,676 --> 02:28:31,145
ON THE ANSWER THERE SO THAT WE

3862
02:28:31,145 --> 02:28:33,181
CAN FIND THE MAP OR PERFORMANCE

3863
02:28:33,181 --> 02:28:35,183
OF THE GIVEN LIGAND IN THE

3864
02:28:35,183 --> 02:28:36,818
SHORTEST AMOUNT OF TIME

3865
02:28:36,818 --> 02:28:37,318
POSSIBLE, MINIMIZING

3866
02:28:37,318 --> 02:28:39,954
EXPERIMENTAL COST OF LIGAND AND

3867
02:28:39,954 --> 02:28:40,721
CALLOUS DORPHY.

3868
02:28:40,721 --> 02:28:43,958
SO I'M NOT GOING TO GO THROUGH

3869
02:28:43,958 --> 02:28:45,827
ALL THE DETAILS OF RESULTS I'M

3870
02:28:45,827 --> 02:28:49,764
SHOWING HERE BUT WE LOOKED AT

3871
02:28:49,764 --> 02:28:50,498
DEFINITE SURROGATE MODELING

3872
02:28:50,498 --> 02:28:52,099
ARCHITECTURES AND WE LOOK AT

3873
02:28:52,099 --> 02:28:53,634
DIFFERENT DECISION MAKING

3874
02:28:53,634 --> 02:28:56,671
POLICIES AND WE DID MORE THAN

3875
02:28:56,671 --> 02:29:05,513
6000 SIMULATIONS TO FINE TUNE

3876
02:29:05,513 --> 02:29:06,714
THE LIGANDS, AND AFTER THAT WE

3877
02:29:06,714 --> 02:29:08,482
ARE FINE TUNING THE AGENT WITH

3878
02:29:08,482 --> 02:29:10,585
THE FULLY AUTOMATED PLATFORM

3879
02:29:10,585 --> 02:29:15,690
THAT TARTS WITH  AUTOMATICALLY

3880
02:29:15,690 --> 02:29:18,092
HANDLING YOUR SOLUTIONS AND THEN

3881
02:29:18,092 --> 02:29:19,026
WE BASICALLY   PERFORMED

3882
02:29:19,026 --> 02:29:21,395
REACTION SAYS AND HALF THE

3883
02:29:21,395 --> 02:29:22,330
MILLILITER SCALE VOLUME.  THIS

3884
02:29:22,330 --> 02:29:25,466
IS AGAIN REACTION OF MAKING ALDA

3885
02:29:25,466 --> 02:29:26,801
HIDESSA HIDES THAT IT CAN USE

3886
02:29:26,801 --> 02:29:30,938
AGAIN TO USE TO FORM AND MAKE

3887
02:29:30,938 --> 02:29:32,540
DIFFERENT AND DIFFERENT ALCOHOLS

3888
02:29:32,540 --> 02:29:33,174
AND APIs.

3889
02:29:33,174 --> 02:29:36,777
AND THEN THE PRODUCT OF THE

3890
02:29:36,777 --> 02:29:38,646
AUTOMATICALLY ANALYZED ONLINE TO

3891
02:29:38,646 --> 02:29:40,915
UPDATE THE SURROGATE MODEL OF

3892
02:29:40,915 --> 02:29:42,617
THE AI AGENTS SO IT COULD SELECT

3893
02:29:42,617 --> 02:29:44,619
THE NEXT EXPERIMENT TO

3894
02:29:44,619 --> 02:29:46,187
AUTONOMOUSLY MAP THE PERFORMANCE

3895
02:29:46,187 --> 02:29:48,489
OF THE SPECIFIC LIGAND OR

3896
02:29:48,489 --> 02:29:51,425
CALLOUS GIVEN THIS REACTION.

3897
02:29:51,425 --> 02:29:55,663
SO HERE'S A PICTURE OF THE

3898
02:29:55,663 --> 02:29:56,898
ACTUAL EXPERIMENTAL PLATFORM,

3899
02:29:56,898 --> 02:29:58,366
DIFFERENT THEREFOREY PUMPS

3900
02:29:58,366 --> 02:30:00,801
THERE, THE EMBASSY MANIFOLD FLOW

3901
02:30:00,801 --> 02:30:02,236
REACTOR AND THE ONLINE ANALYSIS

3902
02:30:02,236 --> 02:30:04,405
AND IN THIS CASE, IT'S GCMS, BUT

3903
02:30:04,405 --> 02:30:05,907
WE CAN DO THE SAME THING WITH

3904
02:30:05,907 --> 02:30:09,210
THE FETAL COMPARTMENT MASS, SPEC

3905
02:30:09,210 --> 02:30:12,446
NA, AND ONLINE TMR, AND

3906
02:30:12,446 --> 02:30:14,348
INTEGRATED WITH THE SELF-DRIVING

3907
02:30:14,348 --> 02:30:15,182
MAPS.

3908
02:30:15,182 --> 02:30:18,352
HERE'S AN EXAMPLE PERFORMANCE OF

3909
02:30:18,352 --> 02:30:21,555
THE SELF-DRIVING CATALYSIS LAB

3910
02:30:21,555 --> 02:30:23,257
SHOWING IT CAN MAP THE SPACE FOR

3911
02:30:23,257 --> 02:30:26,427
THE LIGAND FOR THIS

3912
02:30:26,427 --> 02:30:28,629
HOMODPEENIOUS ORGANIC SYNTHESIS

3913
02:30:28,629 --> 02:30:31,165
AND WITHIN 30 AUTONOMOUSLY  

3914
02:30:31,165 --> 02:30:32,166
PERFORMED PRPLTS OUR

3915
02:30:32,166 --> 02:30:33,834
SELF-RUNNING LAB COULD MAP THE

3916
02:30:33,834 --> 02:30:36,103
PERFORMANCE OF THIS LIGAND FOR

3917
02:30:36,103 --> 02:30:37,204
THE GIVEN REACTION, AND IF YOU

3918
02:30:37,204 --> 02:30:39,273
ARE GOING TO DO SAME THING, WITH

3919
02:30:39,273 --> 02:30:40,708
THE HIDE HIGHLY TRAINED CHEMIST,

3920
02:30:40,708 --> 02:30:42,910
THAT WILL TAKE YOU ABOUT 9-12

3921
02:30:42,910 --> 02:30:44,845
MONTHS AND THEN WE DID THIS FOR

3922
02:30:44,845 --> 02:30:48,582
MULTIPLE DIFFERENT LIGANDS AND

3923
02:30:48,582 --> 02:30:49,684
THEN, THE TRANSFER FOCUSED ON

3924
02:30:49,684 --> 02:30:52,253
KNOWLEDGE IS THE SELF-DRIVING

3925
02:30:52,253 --> 02:30:54,956
LAB IN HALF A MIL PLOA REACTOR

3926
02:30:54,956 --> 02:30:57,358
TO LARGER KALE BATCH REACTOR,

3927
02:30:57,358 --> 02:30:58,426
WITH OUR INDUSTRIAL PARTNER TO

3928
02:30:58,426 --> 02:31:00,094
SHOW THE KNOWLEDGE OF SCHEME AND

3929
02:31:00,094 --> 02:31:01,996
ABILITY TO COMES OUT OF THIS

3930
02:31:01,996 --> 02:31:03,164
SELF-DRIVING LAB, AND WE SHOWED

3931
02:31:03,164 --> 02:31:05,399
THAT THE -- IF YOU USE THE SAME

3932
02:31:05,399 --> 02:31:07,401
LIGAND, THAT YOU DISCOVERED IN

3933
02:31:07,401 --> 02:31:08,803
YOUR SELF-DRIVING LAB WITH THE

3934
02:31:08,803 --> 02:31:09,437
SAME REACTION CONDITIONS AND

3935
02:31:09,437 --> 02:31:11,639
RIGHT IN THE LARGEST SCALE BATCH

3936
02:31:11,639 --> 02:31:13,741
REACTOR, YOU GET THE SAME

3937
02:31:13,741 --> 02:31:16,344
SELECTIVITY AND YIELD, SO THAT

3938
02:31:16,344 --> 02:31:17,111
KNOWLEDGE AND SCALABILITY WAS

3939
02:31:17,111 --> 02:31:18,713
VERY IMPORTANT FOR US, WE SPENT

3940
02:31:18,713 --> 02:31:20,014
ABOUT A YEAR AND A HALF TO MAKE

3941
02:31:20,014 --> 02:31:21,082
SURE THAT THE KNOWLEDGE WE

3942
02:31:21,082 --> 02:31:22,950
DORPHED IS THE SMALLER SCALE

3943
02:31:22,950 --> 02:31:25,386
REACTOR IN THE SELF-DRIVING LAB

3944
02:31:25,386 --> 02:31:26,654
IS SCALABLE AND DIRECTLY

3945
02:31:26,654 --> 02:31:29,290
TRANSFORRABLE TO OUR TREAL

3946
02:31:29,290 --> 02:31:29,523
PARTNER.

3947
02:31:29,523 --> 02:31:30,758
AND THEN YOU CAN DO A LOT OF

3948
02:31:30,758 --> 02:31:32,259
THINGS WITH THE DATA YOU

3949
02:31:32,259 --> 02:31:34,495
GENERATE WITH THE SELF-DRIVING

3950
02:31:34,495 --> 02:31:39,900
LAB, HERE'S A DIGITAL TWIN OF

3951
02:31:39,900 --> 02:31:42,336
THE ACTUAL HOMOGENIUS CATALYTIC

3952
02:31:42,336 --> 02:31:43,337
REACTION WITH DIFFERENT LIGANDS

3953
02:31:43,337 --> 02:31:45,606
YOU CAN TRAIN A GP MODEL FOR

3954
02:31:45,606 --> 02:31:46,540
EVERY LIGAND THAT YOU HAVE

3955
02:31:46,540 --> 02:31:48,175
TESTED AND THEN YOU CAN USE THAT

3956
02:31:48,175 --> 02:31:50,745
DIMINISHITTAL TWIN TO VISUALIZE

3957
02:31:50,745 --> 02:31:52,713
THE REACTION SURFACE TO STUDY

3958
02:31:52,713 --> 02:31:54,815
THE EFFECT OF DEFINITE

3959
02:31:54,815 --> 02:31:55,783
PARAMETERS OF THE DIFFERENCE OF

3960
02:31:55,783 --> 02:31:58,085
THE REACTIONS, THIS IS A GREAT

3961
02:31:58,085 --> 02:31:59,220
TOOL FOR BETTER UNDERSTANDING OF

3962
02:31:59,220 --> 02:32:01,422
WHAT'S GOING ON IN THE SPECIFIC

3963
02:32:01,422 --> 02:32:02,189
REACTION YOU'RE STUDYING.

3964
02:32:02,189 --> 02:32:03,958
AND THEN GOING BACK TO THE FLOW

3965
02:32:03,958 --> 02:32:06,027
REACTOR, I MENTIONED IT CAN GO

3966
02:32:06,027 --> 02:32:09,063
FROM SINGLE DROP CONTINUOUS

3967
02:32:09,063 --> 02:32:11,298
FLOW, WE BASICALLY USE THE

3968
02:32:11,298 --> 02:32:12,366
OPTIMIZED CONDITIONS TO SHOW

3969
02:32:12,366 --> 02:32:15,770
THAT WE CAN DO THIS ON DEMAND SO

3970
02:32:15,770 --> 02:32:18,439
WE CHANGE UP SELECTIVITY OF A

3971
02:32:18,439 --> 02:32:20,541
REACTION, USING THE CONDITIONS

3972
02:32:20,541 --> 02:32:22,043
OF THE DISCOVERY FOR

3973
02:32:22,043 --> 02:32:23,744
SELF-DRIVING LAB, THEY COULD ON

3974
02:32:23,744 --> 02:32:25,379
DEMAND, ON THE FLY CHANGE THE

3975
02:32:25,379 --> 02:32:26,714
REACTION CONDITIONS WITH THE

3976
02:32:26,714 --> 02:32:29,750
SAME LIGAND TO CHANGE THE FDA,

3977
02:32:29,750 --> 02:32:31,485
OF THE HYPE THAT YOU'RE

3978
02:32:31,485 --> 02:32:32,653
PRODUCING IN THIS FLOW REACTOR

3979
02:32:32,653 --> 02:32:35,890
AND LATER ON WE WILL ALSO USE

3980
02:32:35,890 --> 02:32:39,393
THIS METHOD TO SHOW THAT WE CAN

3981
02:32:39,393 --> 02:32:43,030
USELESS DO A HIGH MUTEILATION IN

3982
02:32:43,030 --> 02:32:44,665
THE FLOW FOR ALPHA BRANCH MEANS

3983
02:32:44,665 --> 02:32:46,667
AND WE SHOW WIDE OUR AUTONOMOUS

3984
02:32:46,667 --> 02:32:49,303
LAB WE CAN DISCOVER THE CALLOUS

3985
02:32:49,303 --> 02:32:53,340
AND THE CO CALLOUS TO ACCELERATE

3986
02:32:53,340 --> 02:32:55,476
THE [INDISCERNIBLE] 70 TIMES

3987
02:32:55,476 --> 02:32:58,679
COMPARE TO THE BEST BATCH 

3988
02:32:58,679 --> 02:32:59,613
REACTIONS BEFORE.

3989
02:32:59,613 --> 02:33:00,448
SO THIS IS EXAMPLES OF WHAT YOU

3990
02:33:00,448 --> 02:33:02,783
CAN DO WITH THE LAB BUT WHAT IS

3991
02:33:02,783 --> 02:33:06,487
THE FUTURE OF AUTONOMOUS LABS SO

3992
02:33:06,487 --> 02:33:08,856
WHAT YOU'RE WORKING ON RIGHT NOW

3993
02:33:08,856 --> 02:33:14,128
AND NATIONAL LAB PARTNERS AND ON

3994
02:33:14,128 --> 02:33:16,797
A VISION AND WE CAN HAVE THIS

3995
02:33:16,797 --> 02:33:18,833
NETWORK OF CONNECTED AUTONOMOUS

3996
02:33:18,833 --> 02:33:20,401
LABS   PERFORMING SIMILAR

3997
02:33:20,401 --> 02:33:22,803
CAMPAIGNS, SHARING DATA OVER THE

3998
02:33:22,803 --> 02:33:25,239
CLOUD AND FURTHER ACCELERATING

3999
02:33:25,239 --> 02:33:26,340
DISCOVERY OF NEW MOLECULES OF

4000
02:33:26,340 --> 02:33:27,108
CALLOUS AND MATERIALS.

4001
02:33:27,108 --> 02:33:28,742
BUT IN ORDER TO DO THIS AND

4002
02:33:28,742 --> 02:33:29,510
IMPLEMENT THIS VISION, WE NEED

4003
02:33:29,510 --> 02:33:32,613
TO WORK ON THE REPRODUCIBILITY,

4004
02:33:32,613 --> 02:33:34,748
THE KNOWLEDGE TRANSFER, AND THE

4005
02:33:34,748 --> 02:33:35,816
DATA AUGMENTATION WITHIN

4006
02:33:35,816 --> 02:33:36,951
DIFFERENT ROBOTS.

4007
02:33:36,951 --> 02:33:39,353
SO HERE'S AN EXAMPLE OF THE

4008
02:33:39,353 --> 02:33:41,755
DIFFERENT LANGUAGE OF DIFFERENT

4009
02:33:41,755 --> 02:33:43,958
ROBOTS, LIKE USING BATCH

4010
02:33:43,958 --> 02:33:45,226
REACTIVE, LOW REACTIVE TYPES OF

4011
02:33:45,226 --> 02:33:46,594
DEFINITE ARTICLE TECHNIQUES SO

4012
02:33:46,594 --> 02:33:50,631
HOW DO WE AUGMENT THIS

4013
02:33:50,631 --> 02:33:51,899
HETEROGENEOUS SOURCE OF DATA AND

4014
02:33:51,899 --> 02:33:53,868
HARREGARD, YOU NEED TO WORK ON

4015
02:33:53,868 --> 02:33:56,403
THE EXTRACTION, TOO MUCH

4016
02:33:56,403 --> 02:33:57,171
ABNORMALITIES INSTRUCTION, TOO

4017
02:33:57,171 --> 02:33:58,105
LITTLE ABSTRACTION IS NOT GOOD,

4018
02:33:58,105 --> 02:34:01,242
YOU HAVE TO DO A LOT OF

4019
02:34:01,242 --> 02:34:05,146
DEFINITIONS OF DIFFERENT DETAILS

4020
02:34:05,146 --> 02:34:05,946
OF THESE AUTONOMOUS OPERATION

4021
02:34:05,946 --> 02:34:07,448
AND THAT'S SOMETHING WE ARE

4022
02:34:07,448 --> 02:34:09,450
WORKING ON AND DEFINITELY NEEDED

4023
02:34:09,450 --> 02:34:11,085
TO IMPLEMENT THIS, AT THE

4024
02:34:11,085 --> 02:34:13,888
DIVISION OF THESE AUTONOMOUS

4025
02:34:13,888 --> 02:34:14,088
LABS.

4026
02:34:14,088 --> 02:34:14,855
VERY IMPORTANT THING WE HAVE TO

4027
02:34:14,855 --> 02:34:17,158
ALWAYS THINK ABOUT IS WHEN WE

4028
02:34:17,158 --> 02:34:18,659
DESIGN A SELF-DRIVING LAB, YOU

4029
02:34:18,659 --> 02:34:20,427
HAVE TO   PERFORM A LOT OF BUNCH

4030
02:34:20,427 --> 02:34:22,129
MARKING TESTS BOTH ON THE

4031
02:34:22,129 --> 02:34:23,197
HARDWARE AND SOFTWARE SIDE OF

4032
02:34:23,197 --> 02:34:26,333
THIS LAB SO MAKE SURE THAT WE

4033
02:34:26,333 --> 02:34:28,502
CAN BASICALLY HAVE HAD APPLE TO

4034
02:34:28,502 --> 02:34:29,937
APPLE COMPARISON AND THE

4035
02:34:29,937 --> 02:34:31,472
PERFORMANCE OF THESE AUTONOMOUS

4036
02:34:31,472 --> 02:34:33,007
LABS FROM THE CHEMICAL

4037
02:34:33,007 --> 02:34:34,675
COMSUMPTION TO THE INILLEGALSIC

4038
02:34:34,675 --> 02:34:36,243
O OPTIMIZATION MODELS THAT

4039
02:34:36,243 --> 02:34:39,346
THEY'RE   PERFORMING.

4040
02:34:39,346 --> 02:34:40,781
AND THEN, I WILL END BY

4041
02:34:40,781 --> 02:34:43,184
MENTIONING THAT WE NEED TO WORK

4042
02:34:43,184 --> 02:34:43,817
ON THIS KNOWLEDGE TRANSFER

4043
02:34:43,817 --> 02:34:46,120
BETWEEN DIFFERENT TYPES OF

4044
02:34:46,120 --> 02:34:47,254
ROBOTS AND AUTONOMOUS LABS,

4045
02:34:47,254 --> 02:34:49,523
THAT'S THE ONLY WAY WE CAN

4046
02:34:49,523 --> 02:34:50,457
REALLY DIGITALIZE AND THE

4047
02:34:50,457 --> 02:34:54,628
DISCOVERY OF NEW MOLECULES,

4048
02:34:54,628 --> 02:34:56,430
CALLOUS CAN PROCESSES, IF WE CAN

4049
02:34:56,430 --> 02:34:58,632
NOW TRANSFER KNOWLEDGE AND

4050
02:34:58,632 --> 02:34:59,833
PROTOCOLS, RECIPES, AND DATA

4051
02:34:59,833 --> 02:35:01,035
BETWEEN DIFFERENT LABS OTHER

4052
02:35:01,035 --> 02:35:02,570
EVERYTHING WE DO, ONLY LIMITED

4053
02:35:02,570 --> 02:35:04,805
TO OUR OWN IN-HOUSE DATA THAT WE

4054
02:35:04,805 --> 02:35:07,675
ARE GENERATING WITH OUR OWN

4055
02:35:07,675 --> 02:35:08,108
ROBOTS.

4056
02:35:08,108 --> 02:35:10,411
SO AT THE END I HOPE I WAS ABLE

4057
02:35:10,411 --> 02:35:12,279
TO CONVINCE YOU THAT THESE

4058
02:35:12,279 --> 02:35:13,881
AUTONOMOUS LABS ARE GREAT

4059
02:35:13,881 --> 02:35:16,083
ROBOTIC CO PILOTS TO HELP US

4060
02:35:16,083 --> 02:35:19,620
CLOSE AUTONOMOUS SCALE KNAP IN

4061
02:35:19,620 --> 02:35:20,254
MOLECULAR DISCOVERY, POTENTIALLY

4062
02:35:20,254 --> 02:35:22,623
100 TO THOUSAND TIME COMFORT TO

4063
02:35:22,623 --> 02:35:24,925
THE CONVENTIONAL WAYS TYPICALLY

4064
02:35:24,925 --> 02:35:25,993
APPROACH MOLECULAR AND CALLOUS

4065
02:35:25,993 --> 02:35:26,260
DISCOVERY.

4066
02:35:26,260 --> 02:35:27,728
AT THE END I WANT TO THANK MAY

4067
02:35:27,728 --> 02:35:29,129
GROUP MEMBER WHO IS DID ALL THE

4068
02:35:29,129 --> 02:35:30,097
WORK I PRESENTED TODAY.

4069
02:35:30,097 --> 02:35:34,001
AND I WANT TO THANK OUR FUNDING

4070
02:35:34,001 --> 02:35:35,069
SOURCES, MY COLLABORATORS, WE

4071
02:35:35,069 --> 02:35:36,570
WILL THANK YOU ALL FOR LISTENING

4072
02:35:36,570 --> 02:35:43,010
AND I WILL BE HAPPY TO TAKE ANY

4073
02:35:43,010 --> 02:35:44,345
QUESTIONS.

4074
02:35:44,345 --> 02:35:44,645
>> GREAT.

4075
02:35:44,645 --> 02:35:46,413
THANK YOU SO MUCH.

4076
02:35:46,413 --> 02:35:47,748
SO GREAT, SO FASCINATING.

4077
02:35:47,748 --> 02:35:49,016
THERE ARE A COUPLE QUESTIONS IN

4078
02:35:49,016 --> 02:35:49,850
THE CHAT.

4079
02:35:49,850 --> 02:35:53,687
LET ME ASK THE FIRST 1, YEAH

4080
02:35:53,687 --> 02:35:54,488
THAT'S INTERESTING, HOW DO YOU

4081
02:35:54,488 --> 02:35:58,425
MAKE SURE WHAT THE CELL

4082
02:35:58,425 --> 02:36:00,060
[INDISCERNIBLE] LAB DOES NOT

4083
02:36:00,060 --> 02:36:02,730
TURN INTO THE SKY NET?

4084
02:36:02,730 --> 02:36:05,266
>> THAT'S A QUESTION OF

4085
02:36:05,266 --> 02:36:06,433
CYBERSECURITY OF THESE

4086
02:36:06,433 --> 02:36:07,534
AUTONOMOUS LABS AND THERE'S

4087
02:36:07,534 --> 02:36:10,704
ALWAYS A LOT OF ACTIVITIES AND

4088
02:36:10,704 --> 02:36:14,174
ACTIVE PROGRAMS WORKING ON THE

4089
02:36:14,174 --> 02:36:15,376
CYBER INFRASTRUCTURE, CYBER

4090
02:36:15,376 --> 02:36:16,577
INFRASTRUCTURE OF THESE

4091
02:36:16,577 --> 02:36:17,544
AUTONOMOUS LABS EVEN AT THE

4092
02:36:17,544 --> 02:36:20,014
UNIVERSITY LEVEL, WE HAVE A LOT

4093
02:36:20,014 --> 02:36:22,650
OF INTERNAL PROTOCOLS ON HOW TO

4094
02:36:22,650 --> 02:36:25,119
EVEN REMOTELY OPERATE THIS

4095
02:36:25,119 --> 02:36:26,720
AUTONOMOUS LABS AND THEN THERE

4096
02:36:26,720 --> 02:36:30,157
ARE A LOT OF THINGS YOU HAVE TO

4097
02:36:30,157 --> 02:36:32,593
CONSIDER WHEN YOU ARE WORKING ON

4098
02:36:32,593 --> 02:36:33,260
INDUSTRY SPONSORED PROJECTS, HOW

4099
02:36:33,260 --> 02:36:36,163
DO YOU MAKE SURE THAT YOUR DATA

4100
02:36:36,163 --> 02:36:37,998
IS SECURE, SAFE OR KIND OF

4101
02:36:37,998 --> 02:36:40,000
SERVERS, YOU USE, BUT THE

4102
02:36:40,000 --> 02:36:41,135
ANSWER, I CANNOT ANSWER LIKE IN

4103
02:36:41,135 --> 02:36:43,671
THE NEXT 2 MINUES BUT THE ANSWER

4104
02:36:43,671 --> 02:36:45,272
IS THE ROBUST CYBER

4105
02:36:45,272 --> 02:36:46,940
INFRASTRUCTURE AND A LOT OF

4106
02:36:46,940 --> 02:36:47,875
FUNDING SOURCES, FEDERAL FUNDING

4107
02:36:47,875 --> 02:36:49,510
SOURCES IN THE U.S. ARE ACTUALLY

4108
02:36:49,510 --> 02:36:54,615
COMING UP WITH NEW PROGRAMS ON

4109
02:36:54,615 --> 02:36:57,651
PROMOTING THE CYBER

4110
02:36:57,651 --> 02:36:58,352
INFRASTRUCTURE FOR

4111
02:36:58,352 --> 02:37:00,321
EXPERIMENTATION AND CLOUD LABS

4112
02:37:00,321 --> 02:37:01,155
AND AUTONOMOUS LABS.

4113
02:37:01,155 --> 02:37:01,622
>> GREAT.

4114
02:37:01,622 --> 02:37:04,358
THEN OUR QUESTION IS, THE

4115
02:37:04,358 --> 02:37:05,125
MICROFLUID APPROACH IS GREAT TO

4116
02:37:05,125 --> 02:37:06,860
APPLY IN THE CHEMISTRY REACTION,

4117
02:37:06,860 --> 02:37:09,330
HOW WOULD YOU APPLY THIS TO YOUR

4118
02:37:09,330 --> 02:37:10,464
ENZYME REACTIONS OR BIOLOGICAL

4119
02:37:10,464 --> 02:37:12,333
REACTION FOR THE DRUG DORPHY.

4120
02:37:12,333 --> 02:37:14,401
DO YOU NEED TO BUILD DIFFERENT

4121
02:37:14,401 --> 02:37:17,204
WORK FLOW TO EACH DIFFERENT

4122
02:37:17,204 --> 02:37:17,471
REACTION.

4123
02:37:17,471 --> 02:37:18,672
NIS THAT'S A GREAT QUESTION.

4124
02:37:18,672 --> 02:37:20,107
SO IF YOU DON'T START EVERY

4125
02:37:20,107 --> 02:37:22,209
SINGLE REACTION IN FLOW, WE

4126
02:37:22,209 --> 02:37:23,577
START -- I BRIEFLY MENTIONED

4127
02:37:23,577 --> 02:37:23,877
THAT.

4128
02:37:23,877 --> 02:37:26,580
WE START WITH AN INITIAL

4129
02:37:26,580 --> 02:37:29,216
[INDISCERNIBLE] KREENING SO IF

4130
02:37:29,216 --> 02:37:32,753
YOU HAVE A REACTION THAT INVOFS

4131
02:37:32,753 --> 02:37:34,188
SOLID PRECIPITATION OR SOLID ARK

4132
02:37:34,188 --> 02:37:36,590
DEGZ, WE WILL RUN IT IN FLOW.

4133
02:37:36,590 --> 02:37:39,293
BUT IF THERE'S A REACTION, WE

4134
02:37:39,293 --> 02:37:40,861
KNOW THERE'S NO PRECIPITATION

4135
02:37:40,861 --> 02:37:42,563
AND THERE'S NO SOLID ADDITION,

4136
02:37:42,563 --> 02:37:44,098
WE RUN THAT REACTION IN FLOW, SO

4137
02:37:44,098 --> 02:37:45,933
WE DON'T FORCE THE REACTIONS IN

4138
02:37:45,933 --> 02:37:48,235
FLOW, BUT WHEN THE REACTION IS

4139
02:37:48,235 --> 02:37:50,671
ONLY SOLUTION PHASE, THERE'S NO

4140
02:37:50,671 --> 02:37:52,072
PRECIPITATION CHALLENGE AND

4141
02:37:52,072 --> 02:37:53,574
THERE'S NO SOLID EDITION DURING

4142
02:37:53,574 --> 02:37:55,876
THE REACTION AND IT'S ALL

4143
02:37:55,876 --> 02:37:57,211
SOLUTION PHASED AND RUN THE

4144
02:37:57,211 --> 02:37:58,312
REACTION IN FLOW, NOW IN TERMS

4145
02:37:58,312 --> 02:38:01,682
OF THE DESIGN OF THE REACTOR,

4146
02:38:01,682 --> 02:38:03,450
THE FLOW REACTORS HAVE EVOLVED

4147
02:38:03,450 --> 02:38:08,555
SIGNIFICANTLY IN THE LAST 20

4148
02:38:08,555 --> 02:38:09,990
YEARS NOW..

4149
02:38:09,990 --> 02:38:11,792
WE ARE AT A LEVEL FOR THE FLOW

4150
02:38:11,792 --> 02:38:15,329
REACTORS THAT WE CAN USE ORGANIC

4151
02:38:15,329 --> 02:38:16,163
AND INORGANIC SYNTHESIS WITHOUT

4152
02:38:16,163 --> 02:38:19,133
THE NEED TO REDESIGN THE WORK

4153
02:38:19,133 --> 02:38:22,936
FLOW, WE CAN USE THE SAME

4154
02:38:22,936 --> 02:38:24,671
DESIGN, SAME PLATES, MOST OF

4155
02:38:24,671 --> 02:38:26,039
THOSE ARE TEFLON TUBING AND IT'S

4156
02:38:26,039 --> 02:38:27,775
THE DESIGN OF THE HEATING PLATE

4157
02:38:27,775 --> 02:38:29,343
THAT WE TYPICALLY CHANGE FOR

4158
02:38:29,343 --> 02:38:31,245
DIFFERENT REACTIONS, SO THE

4159
02:38:31,245 --> 02:38:32,980
HEATING PLATE IS REUSABLE AND

4160
02:38:32,980 --> 02:38:35,783
TUBES ARE VERY EXPENSIVE TO

4161
02:38:35,783 --> 02:38:36,016
CHANGE.

4162
02:38:36,016 --> 02:38:37,551
>> LAST QUESTION, AND IF
YOU CAN

4163
02:38:37,551 --> 02:38:38,352
ANSWER SOME QUESTIONS IN THE

4164
02:38:38,352 --> 02:38:40,387
CHAT, THAT WOULD BE GREAT, BUT

4165
02:38:40,387 --> 02:38:43,590
THE LAST QUESTION, QUICK 1, SO

4166
02:38:43,590 --> 02:38:45,826
WHEN WE SHOULD EXPECT THE CELL

4167
02:38:45,826 --> 02:38:47,461
DRIVING LABS WILL BE FUNCTIONAL

4168
02:38:47,461 --> 02:38:48,862
COMLITELY IN WHO IS YOUR

4169
02:38:48,862 --> 02:38:50,431
FORECAST, IT'S WORKING TODAY, WE

4170
02:38:50,431 --> 02:38:52,433
CAN RUN OR IT'S LIKE 5 DAYS,

4171
02:38:52,433 --> 02:38:56,303
FROM NOW, WHAT IS YOUR LIKE

4172
02:38:56,303 --> 02:38:56,603
EXPECTATION?

4173
02:38:56,603 --> 02:38:59,773
WHAT YOU WANT TO DESIGN?

4174
02:38:59,773 --> 02:39:01,608
>> THE WHOLE FIELD OF

4175
02:39:01,608 --> 02:39:02,776
SELF-DRIVING LAB IS IN THE

4176
02:39:02,776 --> 02:39:03,610
EARLY, EARLY STAGES AND THERE

4177
02:39:03,610 --> 02:39:05,546
ARE A LOT OF DIFFERENT VERSIONS

4178
02:39:05,546 --> 02:39:07,414
IN HOW TO EVOLVE THESE

4179
02:39:07,414 --> 02:39:08,615
AUTONOMOUS LABS.

4180
02:39:08,615 --> 02:39:14,688
THERE'S A VISION THAT WE JUST

4181
02:39:14,688 --> 02:39:16,924
NEED ROBOTS AND YOU CAN GO IN

4182
02:39:16,924 --> 02:39:18,358
AND USE EVERYTHING IN AN

4183
02:39:18,358 --> 02:39:19,593
EXISTING LAB TO DO HA AND

4184
02:39:19,593 --> 02:39:21,328
THERE'S THE OTHER VISION THAT

4185
02:39:21,328 --> 02:39:23,096
YOU NEED NEW TOOLS, NEW HARDWARE

4186
02:39:23,096 --> 02:39:25,899
AND NEW AUTOMATION TECHNOLOGIES

4187
02:39:25,899 --> 02:39:28,202
TO INCREASE EFFICIENCY AND

4188
02:39:28,202 --> 02:39:29,670
REDUCE CHEMICAL COMSUMPTION OF

4189
02:39:29,670 --> 02:39:30,938
YOUR REACTION, SO I WOULD SAY

4190
02:39:30,938 --> 02:39:33,340
MAYBE THERE'S A HANDFUL OF FULLY

4191
02:39:33,340 --> 02:39:35,142
ATON MOUSE LABS FULLY

4192
02:39:35,142 --> 02:39:36,143
OPERATIONAL AROUND THE WORLD AND

4193
02:39:36,143 --> 02:39:37,878
IT'S GOING TO PROBABLY TAKE

4194
02:39:37,878 --> 02:39:40,481
ANOTHER DECADE TO SEE GOING FROM

4195
02:39:40,481 --> 02:39:43,617
HANDFUL TO HUNDREDS AROUND THE

4196
02:39:43,617 --> 02:39:43,851
WORLD.

4197
02:39:43,851 --> 02:39:46,053
SO WE'RE STILL IN THE EARLY DAYS

4198
02:39:46,053 --> 02:39:48,555
OF THESE AUTONOMOUS LABS.

4199
02:39:48,555 --> 02:39:48,922
>> GOT IT.

4200
02:39:48,922 --> 02:39:51,091
THANK YOU SO MUCH FOR GREAT TALK

4201
02:39:51,091 --> 02:39:51,725
AND PRESENTATION.

4202
02:39:51,725 --> 02:39:52,926
WITH THAT SAID LET ME MOVE

4203
02:39:52,926 --> 02:39:54,361
FORWARD TO OUR LAST SPEAKER IN

4204
02:39:54,361 --> 02:39:56,196
THE SESSION.

4205
02:39:56,196 --> 02:39:57,631
DAVID BAKER, LET ME BRUCE HIM SO

4206
02:39:57,631 --> 02:39:59,366
HE'S THE DIRECTOR OF INSTITUTE

4207
02:39:59,366 --> 02:40:05,038
FOR THE PROTEIN DESIGN AND

4208
02:40:05,038 --> 02:40:06,173
HOWARD HUGHES INVESTIGATOR,

4209
02:40:06,173 --> 02:40:07,741
PROFESSOR OF BIOCHEMIST RADIOY

4210
02:40:07,741 --> 02:40:11,178
AND PROFESSOR OF GENOMIC

4211
02:40:11,178 --> 02:40:11,678
SCIENCE, BIOENGINEERING,

4212
02:40:11,678 --> 02:40:12,613
COMPUTER SCIENCE AND PHYSICS AT

4213
02:40:12,613 --> 02:40:14,715
THE UNIVERSITY OF WASHINGTON.

4214
02:40:14,715 --> 02:40:21,321
HIS RESEARCH GROUP IS FOCUSED ON

4215
02:40:21,321 --> 02:40:22,923
DESIGN OF MICROMOLECULAR PAPERS,

4216
02:40:22,923 --> 02:40:25,025
HE'S BEEN GRANTED OVER A HUNDRED

4217
02:40:25,025 --> 02:40:27,928
PATENTS AND CONFOUNDER OF 21

4218
02:40:27,928 --> 02:40:28,295
COMPANIES.

4219
02:40:28,295 --> 02:40:29,963
DAVID BAKER AND RECIPIENT OF OF

4220
02:40:29,963 --> 02:40:32,065
THE BREAK THROUGH PRIZE IN THE

4221
02:40:32,065 --> 02:40:36,303
LIFE SCIENCE AS A MEMBER OF THE

4222
02:40:36,303 --> 02:40:37,204
NATIONAL AKAH DEEMIA OF SCIENCE

4223
02:40:37,204 --> 02:40:39,406
AND ART AND SCIENCE AND AS YOU

4224
02:40:39,406 --> 02:40:42,776
MAY KNOW, HE'S RECENTLY GOTTEN A

4225
02:40:42,776 --> 02:40:45,579
NOBEL PRIZE ON BEHALF OF ALL

4226
02:40:45,579 --> 02:40:46,113
ORGANIZATIONING COMMITTEE

4227
02:40:46,113 --> 02:40:47,915
MEMBERS AND ALL OF US HERE AT

4228
02:40:47,915 --> 02:40:49,449
NCATS WE CONGRATULATE YOU ON

4229
02:40:49,449 --> 02:40:50,717
RECEIVING THE NOBEL PRIZE.

4230
02:40:50,717 --> 02:40:54,221
IT'S HUGE HONOR FOR US TO HAVE

4231
02:40:54,221 --> 02:40:58,525
YOU HERE AND PLEASE, THE  STAGE

4232
02:40:58,525 --> 02:40:59,059
IS YOURS.

4233
02:40:59,059 --> 02:41:00,727
>> I DON'T THINK WE CAN
HEAR

4234
02:41:00,727 --> 02:41:01,094
YOU.

4235
02:41:01,094 --> 02:41:07,334
>> CAN YOU SAY --

4236
02:41:07,334 --> 02:41:09,136
>> CAN YOU HEAR ME NOW?

4237
02:41:09,136 --> 02:41:10,170
>> OKAY, YEAH, WELL, THANK
YOU

4238
02:41:10,170 --> 02:41:11,705
FOR THE INTRODUCTION AND LET ME

4239
02:41:11,705 --> 02:41:15,742
JUST TRY AND SHARE MY SLIDES.

4240
02:41:15,742 --> 02:41:22,316
LET'S SEE, HERE WE GO.

4241
02:41:22,316 --> 02:41:23,483
LET'S SEE SLIDE SHOW.

4242
02:41:23,483 --> 02:41:27,421
CAN YOU SEE MY SLIDES.

4243
02:41:27,421 --> 02:41:28,155
>> YES NPERFECT.

4244
02:41:28,155 --> 02:41:30,223
THANK YOU NI WILL NOT TALK ABOUT

4245
02:41:30,223 --> 02:41:32,025
SMALL MOLECULE DESIGN BUT ABOUT

4246
02:41:32,025 --> 02:41:33,527
PROTEIN DESIGN, BUT THERE'S MAUL

4247
02:41:33,527 --> 02:41:38,465
PROTEINS SO IT'S CLOSE.

4248
02:41:38,465 --> 02:41:40,901
SO THE LAST FEW YEARS WE HAVE

4249
02:41:40,901 --> 02:41:42,536
BEEN FOCUSING ON DEVELOPING NEW

4250
02:41:42,536 --> 02:41:44,371
DEEP LEARNING BASED METHODS FOR

4251
02:41:44,371 --> 02:41:45,405
PROTEIN DESIGN, SO GIIVE A

4252
02:41:45,405 --> 02:41:49,209
PROBLEM WE WANT TO SOLVE, WE

4253
02:41:49,209 --> 02:41:50,510
START BY SPECIFY THE PROBLEM WE

4254
02:41:50,510 --> 02:41:52,946
WANT TO SOLVE AND THEN IMENERATE

4255
02:41:52,946 --> 02:41:54,114
PROTEIN BACKBONES THAT SOLVE

4256
02:41:54,114 --> 02:41:56,183
THAT PROBLEM USING RF DIFFUSION

4257
02:41:56,183 --> 02:41:58,986
THEN WE,A SIGN SEQUENCES TO

4258
02:41:58,986 --> 02:42:03,490
THOSE PROTEINS USING PROTEIN

4259
02:42:03,490 --> 02:42:04,925
MPNN, AND THEN WEE CHECK TO SEE

4260
02:42:04,925 --> 02:42:09,262
IF THEY ENCODE THOSE STRUCTURES

4261
02:42:09,262 --> 02:42:11,398
BY USING ROSETTA FOLD OR ALPHA

4262
02:42:11,398 --> 02:42:11,665
FOLD.

4263
02:42:11,665 --> 02:42:13,333
AND IF THEY DO, TO MAKE THE

4264
02:42:13,333 --> 02:42:15,769
PROTEINS WE HAVE TO MAKE SIN

4265
02:42:15,769 --> 02:42:17,638
SETTHETIC GENES AND THEN WE PUT

4266
02:42:17,638 --> 02:42:18,572
THEM INTO BACTERIA OR YEAST

4267
02:42:18,572 --> 02:42:19,339
ANDEE IF THEY WORK.

4268
02:42:19,339 --> 02:42:21,141
SO EVERYTHING I WILL TELL YOU

4269
02:42:21,141 --> 02:42:23,810
ABOUT STARTS WITH A CALCULATION

4270
02:42:23,810 --> 02:42:25,646
THAT WE MAKE A SYNTHETIC GENE,

4271
02:42:25,646 --> 02:42:26,813
MAKE THE PROTEIN AND TEST IT.

4272
02:42:26,813 --> 02:42:29,149
SO I WILL SAY JUST A FEW WORDS

4273
02:42:29,149 --> 02:42:32,719
ABOUT HOW RF DIFFUSION WORKS, SO

4274
02:42:32,719 --> 02:42:34,955
IT'S VERY SIMILAR TO METHODS

4275
02:42:34,955 --> 02:42:36,156
LIKE, DOLLY FOR GENERATING

4276
02:42:36,156 --> 02:42:38,558
IMAGES WHERE YOU TAKE MANY, MANY

4277
02:42:38,558 --> 02:42:40,494
DIFFERENT IMAGES, YOU NOISE THEM

4278
02:42:40,494 --> 02:42:41,728
TO DIFFERENT EXTENTS, YOU TRAIN

4279
02:42:41,728 --> 02:42:44,164
A NETWORK TO REMOVE THE NOISE

4280
02:42:44,164 --> 02:42:45,766
AND THEN ONCE THAT'S BEEN DONE,

4281
02:42:45,766 --> 02:42:47,901
CAN YOU START WITH RANDOMINIZE

4282
02:42:47,901 --> 02:42:49,836
AND DENOISE IT TO GENERATE

4283
02:42:49,836 --> 02:42:51,204
IMAGES, IN THE SAME WAY WE TOOK

4284
02:42:51,204 --> 02:42:53,707
ALL THE STRUCTURALLY ARES IN THE

4285
02:42:53,707 --> 02:42:55,375
PROTEIN DATA BANK, NOISED THEM

4286
02:42:55,375 --> 02:42:57,010
TO DIFFERENT EXTENTS, TRAINED

4287
02:42:57,010 --> 02:42:58,311
DURING RF DIFFUSION TO MOVE THE

4288
02:42:58,311 --> 02:43:00,247
NOISE AND THEN WE TART

4289
02:43:00,247 --> 02:43:02,049
COMPLETELY RANDOMINIZE AND

4290
02:43:02,049 --> 02:43:04,117
REMOVE THE NOISE AND GENERATE

4291
02:43:04,117 --> 02:43:05,752
PROTEINS STRUCTURES THAT LOOK

4292
02:43:05,752 --> 02:43:09,289
LIKE THEY WERE IN THE PDU BUT

4293
02:43:09,289 --> 02:43:13,794
THEY'RE ACTUALLY NOT. 

4294
02:43:13,794 --> 02:43:14,561
OOPS, LET'S SEE.

4295
02:43:14,561 --> 02:43:16,930
SO I JUST TRIED A COUPLE

4296
02:43:16,930 --> 02:43:19,266
DIFFERENT APPLICATIONS OF THIS,

4297
02:43:19,266 --> 02:43:22,235
THE FIRST IS DESIGNING BINDERS.

4298
02:43:22,235 --> 02:43:24,471
SO WHAT WE CAN DO IS TAKE THE

4299
02:43:24,471 --> 02:43:26,006
TARGET, THE INSULIN RECEPTOR AND

4300
02:43:26,006 --> 02:43:27,407
CARRY OUT THIS DIFFUSION PROCESS

4301
02:43:27,407 --> 02:43:30,343
AND WHAT IS DOES IS IT DPENERATE

4302
02:43:30,343 --> 02:43:31,411
ACE BACKBONE THAT CONFORMS TO

4303
02:43:31,411 --> 02:43:32,813
THE SURFACE OF THE TARGET AS CAN

4304
02:43:32,813 --> 02:43:35,649
YOU SEE HERE, SO AS THE NOISING

4305
02:43:35,649 --> 02:43:38,385
HAPPENS, IT'S BEEN TRAINED ON

4306
02:43:38,385 --> 02:43:39,352
PROTEIN-PROTEIN COMPLEXES SO IT

4307
02:43:39,352 --> 02:43:40,287
KNOWS WHAT THEY SHOULD LOOK LIKE

4308
02:43:40,287 --> 02:43:42,022
AND YOU CAN SEE THE STRUCTURE

4309
02:43:42,022 --> 02:43:43,690
FITS NICELY AGAINST THE TARGET,

4310
02:43:43,690 --> 02:43:46,226
SO THIS APPLYING THIS TO THE TNF

4311
02:43:46,226 --> 02:43:48,061
RECEPTOR, YOU KNOW A MAJOR DRUG

4312
02:43:48,061 --> 02:43:52,299
TARGET FOR INFLAMMATORY DISEASE,

4313
02:43:52,299 --> 02:43:54,234
AND THE GEARS, YOU SEE THE

4314
02:43:54,234 --> 02:43:55,368
DIFFUSION PROCESS, IT GENERATES

4315
02:43:55,368 --> 02:43:57,404
A STRUCTURE WHICH IS VERY, VERY

4316
02:43:57,404 --> 02:43:59,873
SHAPED COMPLIMENTARY TO THE

4317
02:43:59,873 --> 02:44:03,176
RECEPTOR, AND AFTER -- SO THIS

4318
02:44:03,176 --> 02:44:07,514
IS AFTER TAKING THE BEST

4319
02:44:07,514 --> 02:44:10,450
DESIGNS, REDOING THE -- ADDING

4320
02:44:10,450 --> 02:44:11,485
NOISE, DENOISING, TESTING

4321
02:44:11,485 --> 02:44:13,553
ANOTHER SMALL HANDFUL OF

4322
02:44:13,553 --> 02:44:17,057
DESIGNS, WE GET DOWN TO 8 PICK O

4323
02:44:17,057 --> 02:44:19,893
MOLAR AND WE HAVE NEAT PROPERS

4324
02:44:19,893 --> 02:44:23,430
IN ANIMALS, THEY GIVE COMPLETE

4325
02:44:23,430 --> 02:44:25,632
PROTECTION AGAINST LPS INDUCED

4326
02:44:25,632 --> 02:44:27,601
INFLAMMATION, THEY'RE ACTUALLY

4327
02:44:27,601 --> 02:44:29,269
MORE EFFECTIVE THAN ENBREL WHICH

4328
02:44:29,269 --> 02:44:30,637
IS PRETTY COOL BECAUSE THEY'RE

4329
02:44:30,637 --> 02:44:34,741
NOT BLOCKING THE TNF RECEPTOR 2,

4330
02:44:34,741 --> 02:44:35,709
WHICH IS ANTIINFLAMMATORY, AND

4331
02:44:35,709 --> 02:44:39,846
WE CAN TAKE THESE COMPOUNDS AND

4332
02:44:39,846 --> 02:44:41,982
THESE DESIGNS AND DO THIS

4333
02:44:41,982 --> 02:44:43,950
PARTIAL DIFFUSION METHOD TO

4334
02:44:43,950 --> 02:44:45,685
GENERATE BINDERS TO OTHER DEATH

4335
02:44:45,685 --> 02:44:47,087
RECEPTOR FAMILY MEMBERS AND WE

4336
02:44:47,087 --> 02:44:49,890
CAN ALSO CONVERT THESE INTO

4337
02:44:49,890 --> 02:44:52,092
QUITE SELECTIVE AGONISTS BY

4338
02:44:52,092 --> 02:44:53,727
BRINGING -- THAT BRING TOGETHER

4339
02:44:53,727 --> 02:44:58,732
MULTIPLE COPIES OF THE RECEPTOR.

4340
02:44:58,732 --> 02:45:02,302
WE'VE BEEN MAKING BIEBDERS FOR A

4341
02:45:02,302 --> 02:45:03,870
NUMBER OF YEARS AS I'VE SPOKEN

4342
02:45:03,870 --> 02:45:06,640
WITH MANY PEOPLE AT THE NIH

4343
02:45:06,640 --> 02:45:09,576
ABOUT THIS AGAINST THE MAJOR

4344
02:45:09,576 --> 02:45:12,045
DIFFERENT PANDEMIC FAMILIES FOR

4345
02:45:12,045 --> 02:45:12,712
EXAMPLE, INFLUENZA, CORONA

4346
02:45:12,712 --> 02:45:14,181
VIRUS, WE HAVE A COMPOUND MAKING

4347
02:45:14,181 --> 02:45:17,584
ITS WAY TO THE CLINIC THERE,

4348
02:45:17,584 --> 02:45:22,622
MERSYSTEM, WE HAVE VERY POTENT

4349
02:45:22,622 --> 02:45:24,457
NEUTRALIZER VIRUSES, NIPA AND

4350
02:45:24,457 --> 02:45:26,193
HINDRA, I HAVE TALKED ABOUT THAT

4351
02:45:26,193 --> 02:45:28,028
BEFORE AND I WILL NOT TALK ABOUT

4352
02:45:28,028 --> 02:45:29,563
IT TODAY, BUT MORE RECENTLY

4353
02:45:29,563 --> 02:45:31,832
WE'VE BEEN BINDING THEM TO

4354
02:45:31,832 --> 02:45:35,135
NCI--AND VENOM TOXINS AND

4355
02:45:35,135 --> 02:45:36,903
SUZANNEA TORRES HAS MADE THESE

4356
02:45:36,903 --> 02:45:38,505
THAT NEUTRALIZE THE COMPONENTS

4357
02:45:38,505 --> 02:45:40,040
OF SNAKE VENOM AND GIVE

4358
02:45:40,040 --> 02:45:42,475
PROTECTION IN ANIMALS WHEN

4359
02:45:42,475 --> 02:45:44,211
ADMINISTERED, EVEN WHEN

4360
02:45:44,211 --> 02:45:45,645
ADMINISTERED SUBSTANTIALLY AFTER

4361
02:45:45,645 --> 02:45:46,046
THE TOXIN.

4362
02:45:46,046 --> 02:45:47,848
SO 1 THINK THIS I DIDN'T SAY IS

4363
02:45:47,848 --> 02:45:49,216
THAT DESIGN PROTEINS ARE SUPER

4364
02:45:49,216 --> 02:45:51,084
STABLE AND HAVE REALLY LONG

4365
02:45:51,084 --> 02:45:52,652
SHELF LIFE SO FOR APPLICATIONS

4366
02:45:52,652 --> 02:45:55,255
WHERE YOU KNOW YOU NEED SAY AN

4367
02:45:55,255 --> 02:45:56,823
ANTIVENOM THAT, CAN LAST FOR A

4368
02:45:56,823 --> 02:46:00,193
LONG TIME, THESE COULD BE  VERY

4369
02:46:00,193 --> 02:46:01,127
GOOD.

4370
02:46:01,127 --> 02:46:03,763
THESE ARE BINDERS AGAINST TOXIN

4371
02:46:03,763 --> 02:46:05,532
TCSL THAT CAUSES TOXIC SHOCK AND

4372
02:46:05,532 --> 02:46:08,501
THESE AGAIN ARE EXTREMELY POTENT

4373
02:46:08,501 --> 02:46:11,605
AND VERY POTENT PROTECTION FROM

4374
02:46:11,605 --> 02:46:12,606
LETHAL TOXIN ADMINITRATION IN

4375
02:46:12,606 --> 02:46:23,016
ANIMALS SO YOU CAN SEE HERE, AND

4376
02:46:23,016 --> 02:46:28,822
THESE ARE BINDERS THAT PROTECT

4377
02:46:28,822 --> 02:46:29,422
AGAINST THE TOXIN.

4378
02:46:29,422 --> 02:46:35,462
WE CAN NOT ONLY CONDITION THIS

4379
02:46:35,462 --> 02:46:36,863
PROCESS ON THE TARGET BUT ALSO

4380
02:46:36,863 --> 02:46:38,999
WE CAN TELL WHAT TYPE OF FOLD WE

4381
02:46:38,999 --> 02:46:40,600
WANT IT TO MAKE, SO IN THIS

4382
02:46:40,600 --> 02:46:43,570
CASE, WE CAN MAKE IT -- WE CAN

4383
02:46:43,570 --> 02:46:45,071
SPECIFY AN ANTIBODY OR A NANO

4384
02:46:45,071 --> 02:46:46,940
BODY FOLD AND WHAT THE DIFFUSION

4385
02:46:46,940 --> 02:46:48,708
PROCESS GENERATES IS A NANO BODY

4386
02:46:48,708 --> 02:46:50,877
STRUCTURE AND HERE'S A CRYOEM

4387
02:46:50,877 --> 02:46:52,545
STRUCTURE OF THE DIFFUSED NANO

4388
02:46:52,545 --> 02:46:54,147
BODY THAT BOUND TO A NANO BODY.

4389
02:46:54,147 --> 02:46:55,882
I SHOULD STRESS THIS IS NOT

4390
02:46:55,882 --> 02:46:58,251
TAKING AN ALREADY EXISTING

4391
02:46:58,251 --> 02:47:00,186
ANTIBODY ASPECT GEN STRECTURE

4392
02:47:00,186 --> 02:47:02,188
AND REDESIGNING IT, THIS IS

4393
02:47:02,188 --> 02:47:03,556
REBUILDING IT FROM SCRATCH SO

4394
02:47:03,556 --> 02:47:05,292
THESE BINDING MODE AND THE LOOP

4395
02:47:05,292 --> 02:47:06,793
IS COMPLETELY KNEW AND IT'S

4396
02:47:06,793 --> 02:47:08,228
DISCUSSED BY THE DIFFUSION STOP

4397
02:47:08,228 --> 02:47:11,364
PART, YOU CANEE THERE'S A VERY

4398
02:47:11,364 --> 02:47:12,299
CLOSE SIMILARITY BETWEEN THE

4399
02:47:12,299 --> 02:47:16,102
STRUCTURE AND THE FUSE, SO WE

4400
02:47:16,102 --> 02:47:18,038
CAN PUT BINDERS TOGETHER TO MAKE

4401
02:47:18,038 --> 02:47:20,040
AGONIST, HERE'S AN AGONIST THAT

4402
02:47:20,040 --> 02:47:21,207
IS THIS GREAT COLLABORATION

4403
02:47:21,207 --> 02:47:24,711
WHICH IS VERY POTENT, IL-21

4404
02:47:24,711 --> 02:47:27,647
AGONIST AND IT TURNS OUT, WE'RE

4405
02:47:27,647 --> 02:47:29,883
FINDING THAT THESE DESIGN

4406
02:47:29,883 --> 02:47:31,651
PROTEINS ARE HAVING KIND OF MUCH

4407
02:47:31,651 --> 02:47:34,487
MORE POTENT EFFECTS IN ANIMALS

4408
02:47:34,487 --> 02:47:35,522
THAN THEIR NATURAL COUNTERPART

4409
02:47:35,522 --> 02:47:38,758
AND SO FOR EXAMPLE, WITH IL-21,

4410
02:47:38,758 --> 02:47:42,195
THERE IS A MOUSE SURVIVAL CURVE

4411
02:47:42,195 --> 02:47:45,732
IN A TUMOR MODEL, AND YOU CAN

4412
02:47:45,732 --> 02:47:47,467
SEE WHAT THE ANALOGUE WE'RE

4413
02:47:47,467 --> 02:47:51,304
GETTING MUCH MORE PROLONGED

4414
02:47:51,304 --> 02:47:53,373
PROTECTION AND THAT'S WHAT MIKE

4415
02:47:53,373 --> 02:47:56,443
IS STEPHANIE HAVE FOUND THAT THE

4416
02:47:56,443 --> 02:47:57,577
STAT SIGNAL ACTIVATION ALLOWS

4417
02:47:57,577 --> 02:47:59,546
FOR MUCH, MUCH, LONGER AND THE

4418
02:47:59,546 --> 02:48:01,147
PROTEIN BYPASSEDS TIGHTLY TO THE

4419
02:48:01,147 --> 02:48:02,782
RECEPTOR AND IT'S TABLE SO IT

4420
02:48:02,782 --> 02:48:04,818
DOESN'T GET TURNED OVER WHEREAS

4421
02:48:04,818 --> 02:48:05,819
MOST NATURAL SIGNALING MOLECULES

4422
02:48:05,819 --> 02:48:07,153
ARE MEANT TO TURNOVER QUICKLY SO

4423
02:48:07,153 --> 02:48:08,922
THERE COULD BE A NUMBER OF

4424
02:48:08,922 --> 02:48:10,156
ADVANTAGES OF THESE

4425
02:48:10,156 --> 02:48:10,523
THERAPEUTICS.

4426
02:48:10,523 --> 02:48:13,660
WE CAN MAKE THESE PROTEINS, WE

4427
02:48:13,660 --> 02:48:15,595
CAN CAGE THEM IN VARIOUS WAYS SO

4428
02:48:15,595 --> 02:48:17,063
THEY'RE NOT ACTIVE SYSTEMICALLY,

4429
02:48:17,063 --> 02:48:21,634
THIS IS SHOWING CAGING WITH THE

4430
02:48:21,634 --> 02:48:24,270
MATE RICK METAL O PROTEASE, AND

4431
02:48:24,270 --> 02:48:26,106
THESE ARE ONLY ACTIVE WHEN A

4432
02:48:26,106 --> 02:48:28,108
TARGETING SYSTEM SET PRESENT FOR

4433
02:48:28,108 --> 02:48:30,910
EXAMPLE, PDL1, SO HERE WE HAVE A

4434
02:48:30,910 --> 02:48:32,712
CAGED MIMIC THAT UNTIL AND

4435
02:48:32,712 --> 02:48:36,950
UNLESS PDL 1 IS PRESENT, IS

4436
02:48:36,950 --> 02:48:38,218
COMPLETELY INACTIVE, IF PDL 1 IS

4437
02:48:38,218 --> 02:48:40,186
OPEN ON A TARGET CELL THEN IT

4438
02:48:40,186 --> 02:48:44,858
CAN OPEN UP AND ACTIVATE ON AN

4439
02:48:44,858 --> 02:48:45,291
EFFECTOR CELL.

4440
02:48:45,291 --> 02:48:47,594
AND WE'RE FINDING FOR OTHER

4441
02:48:47,594 --> 02:48:49,729
TYPINGS OF COMPOUNDS LIKE

4442
02:48:49,729 --> 02:48:51,664
LEPTINS, WE'RE FINDING VERY

4443
02:48:51,664 --> 02:48:55,735
PROFOUND IN VIVO CENTER FOR

4444
02:48:55,735 --> 02:48:56,603
EXCELLENCE ON AGINGS.

4445
02:48:56,603 --> 02:49:00,640
AND WE CAN TARGET THESE

4446
02:49:00,640 --> 02:49:01,875
COMPOUNDS TO TUMORS BY ATTACHING

4447
02:49:01,875 --> 02:49:03,743
THEM TO BINDING PROTEINS AND

4448
02:49:03,743 --> 02:49:06,513
ALSO GET POTENT POTENTIATION OF

4449
02:49:06,513 --> 02:49:06,913
SIGNALING.

4450
02:49:06,913 --> 02:49:10,450
WE CAN USE THESE BINDERS FOR

4451
02:49:10,450 --> 02:49:11,985
TARGETED DEGRADATION, SO WE

4452
02:49:11,985 --> 02:49:14,487
FIGURED OUT HOW TO DESIGN

4453
02:49:14,487 --> 02:49:17,157
BINDERS TO PROTEIN SWITCH

4454
02:49:17,157 --> 02:49:19,159
UNDERGO ENDOCYTOSIS AND

4455
02:49:19,159 --> 02:49:20,060
TRANSSIGNIFYITOSEIS.

4456
02:49:20,060 --> 02:49:21,127
WE CONFUSE THESE WOPROTEINS

4457
02:49:21,127 --> 02:49:22,395
WHICH BINDS A TARGET AND THEN

4458
02:49:22,395 --> 02:49:25,198
WHAT HAPPENS S&P THE DESIGN

4459
02:49:25,198 --> 02:49:27,934
PROTEIN BINDS THE TARGET ON THE

4460
02:49:27,934 --> 02:49:30,070
1 END AND THE ENDOSIGNIFYITOSING

4461
02:49:30,070 --> 02:49:32,305
ON THE OTHER AND INTERNAL LIGHT

4462
02:49:32,305 --> 02:49:35,442
TAKES IT TO THE LYSOSOME TO

4463
02:49:35,442 --> 02:49:36,509
DEGRADE THE TARGET.

4464
02:49:36,509 --> 02:49:46,219
SO WE CAN TARGET AND TRARNZ

4465
02:49:46,219 --> 02:49:46,786
SIGNIFYITOSING IN DIFFERENT

4466
02:49:46,786 --> 02:49:48,855
TISSUES AND THIS IS LOOK BEING

4467
02:49:48,855 --> 02:49:52,358
AT TISSUE DEGRADATION IN THESE

4468
02:49:52,358 --> 02:49:53,093
DEFINITE TISSUES.

4469
02:49:53,093 --> 02:49:56,463
THIS, WE CAN GREATLY ENHANCE THE

4470
02:49:56,463 --> 02:49:58,131
POTENCY OF ANTAGONISTIC

4471
02:49:58,131 --> 02:49:59,566
ANTIBODIES WITH INFIESING THIS

4472
02:49:59,566 --> 02:50:00,767
SORT OF TARGETED DEGRADATION

4473
02:50:00,767 --> 02:50:03,636
SIGNAL SO THIS IS A PDL 1

4474
02:50:03,636 --> 02:50:05,171
ANTIBODY WHICH HAS SOME EFFICACY

4475
02:50:05,171 --> 02:50:08,274
ON ITS OWN BUT IF YOU LOOK AT,

4476
02:50:08,274 --> 02:50:10,477
BUT WHEN YOU FUSE IT TO, SO

4477
02:50:10,477 --> 02:50:11,845
HERE'S THE TUMOR SIZE FOR

4478
02:50:11,845 --> 02:50:14,180
EXAMPLE, WHEN YOU FUSE IT TO 1

4479
02:50:14,180 --> 02:50:16,216
OF THESE RECYCLING RECEPTORS YOU

4480
02:50:16,216 --> 02:50:17,350
GET GREAT POTENTIATION OF THE

4481
02:50:17,350 --> 02:50:17,717
EFFECT.

4482
02:50:17,717 --> 02:50:19,319
THESE ARE NOT ONLY BLOCKING THE

4483
02:50:19,319 --> 02:50:21,287
TARGET, THEY ARE DEGRADING IT.

4484
02:50:21,287 --> 02:50:22,555
AND WE'VE RECENTLY FIGURED OUT

4485
02:50:22,555 --> 02:50:24,591
HOW TO MAKE THIS WHOLE PROCESS

4486
02:50:24,591 --> 02:50:26,526
CATALYTIC SO WE CAN -- SO, 1

4487
02:50:26,526 --> 02:50:29,696
COMPOUND CAN LEAD TO THE

4488
02:50:29,696 --> 02:50:32,065
TURNOVER OF MANY, MANY DIFFERENT

4489
02:50:32,065 --> 02:50:33,900
TARGET MOLECULES, IN THIS CASE,

4490
02:50:33,900 --> 02:50:38,171
IGGs OR DISEASES THAT INVOLVE

4491
02:50:38,171 --> 02:50:38,638
OVERABUNDANCE OF IGG.

4492
02:50:38,638 --> 02:50:40,140
AND I JUST WANT TO SHOW YOU

4493
02:50:40,140 --> 02:50:42,308
QUICKLY THAT WE CAN ALSO MAKE

4494
02:50:42,308 --> 02:50:44,144
BINDERS TO WHAT'S TYPICALLY BEEN

4495
02:50:44,144 --> 02:50:46,012
VERY HARD FOR DRUG DISCOVERY

4496
02:50:46,012 --> 02:50:47,013
WHICH IS COMPLETELY DISORDER

4497
02:50:47,013 --> 02:50:48,815
PROTEINS AND WE CAN DO THAT IN 2

4498
02:50:48,815 --> 02:50:52,152
WAYS AS SHOWN HERE, WE CAN JUST

4499
02:50:52,152 --> 02:50:53,620
JOURNAL DIFFUSION PROCESS WE CAN

4500
02:50:53,620 --> 02:50:55,755
LET THE TARGET SAMPLE DIFFERENT

4501
02:50:55,755 --> 02:50:57,323
COMFIRMATIONS AS WELL AND WE CAN

4502
02:50:57,323 --> 02:50:59,526
GET -- WE GET OUT QUITE POTENT

4503
02:50:59,526 --> 02:51:00,527
AND SPECIFIC BIEBDERS FOR A

4504
02:51:00,527 --> 02:51:06,199
VARIETY OF DIFFERENT TARGETS.

4505
02:51:06,199 --> 02:51:09,135
WE CAN USE THIS APPROACH IN

4506
02:51:09,135 --> 02:51:10,403
AMYLOID DEDISEASE OR DISEASES

4507
02:51:10,403 --> 02:51:12,438
LIKE ABETTA AND TAU AND WE FIND

4508
02:51:12,438 --> 02:51:13,406
THESE SUPPRESS AG GREEN CELLS

4509
02:51:13,406 --> 02:51:14,207
IMAIGZ, THAT'S WHAT YOU'RE

4510
02:51:14,207 --> 02:51:15,909
LOOKING AT HERE IS AGGREGATION

4511
02:51:15,909 --> 02:51:19,779
OF ABETTA FOR EXAMPLE AS YOU ADD

4512
02:51:19,779 --> 02:51:27,787
INCREASINGA, MOUNTS OF BINDER.

4513
02:51:27,787 --> 02:51:30,423
AND WE CAN ALSO MAKE THESE

4514
02:51:30,423 --> 02:51:32,158
EXTENDED GROUPS AS ILLUSTRATED

4515
02:51:32,158 --> 02:51:34,127
HERE FOR THIS DISCIPLINARY

4516
02:51:34,127 --> 02:51:35,528
NOEVERIN BINDER WHERE THEY HAVE

4517
02:51:35,528 --> 02:51:37,664
POCKETS FOR EACH OF THE AMINO

4518
02:51:37,664 --> 02:51:40,433
ACIDS SOEE CAN MAKE BINDERS TO A

4519
02:51:40,433 --> 02:51:44,404
WIDE VARIETY OF DIFFERENT

4520
02:51:44,404 --> 02:51:45,872
PROTEIN ANDS PEPTIDES, THEY'RE

4521
02:51:45,872 --> 02:51:48,141
SHOWN HERE, THIS BYPASSEDS QUITE

4522
02:51:48,141 --> 02:51:49,509
TIGHTLY FOR THE CRYSTAL

4523
02:51:49,509 --> 02:51:50,677
STRUCTURE AND DESIGN MODEL FOR 1

4524
02:51:50,677 --> 02:51:52,378
OF THESE IS AGAIN THEY'RE

4525
02:51:52,378 --> 02:51:56,616
REMARKABLY SPECIFIC, SO WE CAN

4526
02:51:56,616 --> 02:51:59,619
TARGET VERY SPECIFIC PROTEINS

4527
02:51:59,619 --> 02:52:04,224
AND 1 RECENT THING, THIS IS WORK

4528
02:52:04,224 --> 02:52:08,194
BY K. JOE WU, AND THEY MADE

4529
02:52:08,194 --> 02:52:09,729
SPECIFIC BINDERRINGS FOR EACH OF

4530
02:52:09,729 --> 02:52:11,898
THE 4 ISOFORMS WHICH IS REALLY

4531
02:52:11,898 --> 02:52:14,534
HARD BECAUSE THEY ONLY DIFFER BY

4532
02:52:14,534 --> 02:52:16,402
A FEW RESIDUES AT THE

4533
02:52:16,402 --> 02:52:18,705
C-TERMINUS, SO I WOULD SAY NOW

4534
02:52:18,705 --> 02:52:19,706
TARGETED DISORDER REGIONS HAS

4535
02:52:19,706 --> 02:52:21,708
KNOWN FROM THE HARDEST PROBLEM

4536
02:52:21,708 --> 02:52:23,176
TO THE EASEST PROBLEM BECAUSE WE

4537
02:52:23,176 --> 02:52:25,111
CAN KIND OF FORCE THEM INTO

4538
02:52:25,111 --> 02:52:29,916
WHATEVER STATE WE WANT.

4539
02:52:29,916 --> 02:52:31,284
SO, YOU KNOW THAT'S WHAT I WANT

4540
02:52:31,284 --> 02:52:33,419
TO TELL YOU ABOUT TODAY AND

4541
02:52:33,419 --> 02:52:34,387
REALLY EXCITED ABOUT

4542
02:52:34,387 --> 02:52:38,458
COLLABORATIONS AND USING THESE

4543
02:52:38,458 --> 02:52:38,791
TECHNOLOGIES.

4544
02:52:38,791 --> 02:52:40,126
AND I THINK I DON'T HAVE MUCH

4545
02:52:40,126 --> 02:52:41,828
TIME SO PEOPLE WHO DID THE WORK

4546
02:52:41,828 --> 02:52:42,795
ARE LISTED HERE AND I MENTIONED

4547
02:52:42,795 --> 02:52:44,063
SOME OF THEM ALREADY AND I WILL

4548
02:52:44,063 --> 02:52:50,270
BE HAPPY TO TAKE ANY QUESTIONS.

4549
02:52:50,270 --> 02:52:51,437
>>  REALRY IMPRESSIVE

4550
02:52:51,437 --> 02:52:52,038
PRESENTATION, THANK YOU VERY

4551
02:52:52,038 --> 02:52:55,041
MUCH DAVID I THINK WE HAVE

4552
02:52:55,041 --> 02:52:56,576
MULTIPLE QUESTIONS IN THE CHAT.

4553
02:52:56,576 --> 02:52:59,312
SO, I WILL START WITH THE FIRST

4554
02:52:59,312 --> 02:52:59,445
1.

4555
02:52:59,445 --> 02:53:02,148
MOST OF YOUR EXAMPLES HAVE

4556
02:53:02,148 --> 02:53:03,549
HELIXES AND ONLY 1 EXAMPLE WITH

4557
02:53:03,549 --> 02:53:08,821
THE [INDISCERNIBLE] IN THEM, ARE

4558
02:53:08,821 --> 02:53:12,091
THOSE DE NOVO DESIGN BIASES TO

4559
02:53:12,091 --> 02:53:12,525
THE [INDISCERNIBLE]?

4560
02:53:12,525 --> 02:53:14,227
>> YEAH, IT'S AN
INTERESTING

4561
02:53:14,227 --> 02:53:16,262
QUESTION, WE JUST POSTED A

4562
02:53:16,262 --> 02:53:18,197
PREPRINT THAT'S ON DESIGN OF

4563
02:53:18,197 --> 02:53:20,667
BINDERS THAT ARE EXCLUSIVELY

4564
02:53:20,667 --> 02:53:22,035
AGAINST BETA SHEET CONTAINING

4565
02:53:22,035 --> 02:53:23,169
PROTEINS AND THE BASIC IDEA IS

4566
02:53:23,169 --> 02:53:25,438
IF YOU HAVE AN ID FOLD YOU WANT

4567
02:53:25,438 --> 02:53:27,340
TO TARGET FOR EXAMPLE, A COMMON

4568
02:53:27,340 --> 02:53:30,143
FEATURE IS THE EDGE BETA STRAND

4569
02:53:30,143 --> 02:53:32,011
AND SO NOW WE CONDITION THE

4570
02:53:32,011 --> 02:53:33,980
DIFFUSION PROCESS TO BASICALLY

4571
02:53:33,980 --> 02:53:37,984
BUILD A BETA STRAND WHICH PAIRS,

4572
02:53:37,984 --> 02:53:40,119
SO, -- AND WE'RE FINDING THAT'S

4573
02:53:40,119 --> 02:53:41,921
A PRETTY POWERFUL APPROACH FOR

4574
02:53:41,921 --> 02:53:43,523
BETA CONTAINING TARGETS SO THE

4575
02:53:43,523 --> 02:53:46,759
EDGE OF THE BETA STRAND IS

4576
02:53:46,759 --> 02:53:49,295
POLAR, BUT A COMPLIMENTARY BETA

4577
02:53:49,295 --> 02:53:50,830
STRAND IS A GOOD WAY TO DO THAT

4578
02:53:50,830 --> 02:53:52,532
SO I THINK IF YOU MAKE THE

4579
02:53:52,532 --> 02:53:54,000
PROTEIN SMALLER IN THE DIFFUSION

4580
02:53:54,000 --> 02:53:57,937
PROCESS, THERE IS A BIAS TOWARDS

4581
02:53:57,937 --> 02:53:59,906
ALPHA HELIX BUT I WOULD SAY WHEN

4582
02:53:59,906 --> 02:54:02,108
WE MAKE BINDERS NOW, I WOULD SAY

4583
02:54:02,108 --> 02:54:03,609
IT'S ABOUT 50/50.

4584
02:54:03,609 --> 02:54:04,043
NGOT IT.

4585
02:54:04,043 --> 02:54:08,314
THANK YOU SO MUCH. 

4586
02:54:08,314 --> 02:54:10,249
>> AND OUR QUESTION, HOW
ABOUT

4587
02:54:10,249 --> 02:54:11,918
VIABILITY, THE ABILITY AND AG

4588
02:54:11,918 --> 02:54:13,419
GREEN CELLS IMAIGZ AND OTHER

4589
02:54:13,419 --> 02:54:14,721
RELEVANT IN VIVO EFFECT,

4590
02:54:14,721 --> 02:54:15,388
PROTEINS TAKEN INTO

4591
02:54:15,388 --> 02:54:20,059
CONSIDERATION FOR THESE PROTEIN

4592
02:54:20,059 --> 02:54:22,628
BINDERS SO HAVE YOU THOUGHT

4593
02:54:22,628 --> 02:54:24,263
ABOUT THAT COMPLICITY IN YOUR

4594
02:54:24,263 --> 02:54:24,697
MODEL?

4595
02:54:24,697 --> 02:54:24,997
>> RIGHT.

4596
02:54:24,997 --> 02:54:26,632
A COUPLE OF THESE HAVE GONE TO

4597
02:54:26,632 --> 02:54:27,166
CLINICAL TRIALS.

4598
02:54:27,166 --> 02:54:30,670
SO WE KNOW SOMETHING ABOUT

4599
02:54:30,670 --> 02:54:32,071
IMMUNO GENERATEDISSITY, I WOULD

4600
02:54:32,071 --> 02:54:33,373
SAY IN GENERAL THESE SMALL

4601
02:54:33,373 --> 02:54:36,275
PROTEINS ARE VERY STABLE, VERY

4602
02:54:36,275 --> 02:54:37,410
SOLUBLE AND THEY'RE OBVIOUSLY

4603
02:54:37,410 --> 02:54:37,610
SMALL.

4604
02:54:37,610 --> 02:54:39,846
IF YOU PUT THEM INTO AN ANIMAL

4605
02:54:39,846 --> 02:54:42,048
THEY WILL BE GONE IN 10 MINUTES,

4606
02:54:42,048 --> 02:54:43,449
THEY GET FILTERED OUT IN THE

4607
02:54:43,449 --> 02:54:45,385
KIDNEY, EXCEPT FOR THE AMOUNT OF

4608
02:54:45,385 --> 02:54:47,053
PROTEIN THAT THE PROTEIN IS

4609
02:54:47,053 --> 02:54:48,521
BOUND TO TO THE RECEPTOR, SO

4610
02:54:48,521 --> 02:54:52,125
LIKE I SAID WITH THE IL-21 MIMIC

4611
02:54:52,125 --> 02:54:53,926
WHAT THE DUGANS FOUND WAS THAT

4612
02:54:53,926 --> 02:54:57,063
YOU COULD STILL MEASURE ACTIVITY

4613
02:54:57,063 --> 02:54:59,065
AFTER 24 HOURS SO THE PROTEIN

4614
02:54:59,065 --> 02:55:00,600
THAT BINDS THE TARGET IS STAYS

4615
02:55:00,600 --> 02:55:01,834
THERE, EVERYTHING ELSE GETS

4616
02:55:01,834 --> 02:55:05,905
WASHED OUT. 

4617
02:55:05,905 --> 02:55:08,241
AS FAR AS IMMUNO GENESISSITY, WE

4618
02:55:08,241 --> 02:55:10,510
SEE THE RESPONSES SO WE THINK

4619
02:55:10,510 --> 02:55:11,711
IT'S BECAUSE THEY'RE VERY

4620
02:55:11,711 --> 02:55:13,613
SOLUBLE AND STABLE SO THEY DON'T

4621
02:55:13,613 --> 02:55:15,047
GET PICKED UP BY THE PROCESS BUT

4622
02:55:15,047 --> 02:55:16,649
IT'S CLEAR THAT THAT WILL ALWAYS

4623
02:55:16,649 --> 02:55:18,184
HAVE TO BE CHECKED AS ANYTHING

4624
02:55:18,184 --> 02:55:19,652
GOES THROUGH THE DEVELOPMENT

4625
02:55:19,652 --> 02:55:22,121
PIPELINE, SO WE HAD AN IL2 MIMIC

4626
02:55:22,121 --> 02:55:24,757
THAT WENT THROUGH -- YOU KNOW

4627
02:55:24,757 --> 02:55:25,425
PRETTY EXTENSIVE CLINICAL TRIALS

4628
02:55:25,425 --> 02:55:28,461
AND THEY DID NOT OBSERVE MAJOR

4629
02:55:28,461 --> 02:55:29,429
NEUTRALIZING ANTIBODIES AND

4630
02:55:29,429 --> 02:55:34,500
WE'RE DEVELOPING METHODS NOW FOR

4631
02:55:34,500 --> 02:55:35,468
ELIMINATING THE T-CELL EPITOPES

4632
02:55:35,468 --> 02:55:37,270
IN THE PROCESS BUT SO FAR IT

4633
02:55:37,270 --> 02:55:39,372
HASN'T BEEN A BIG ISSUE AND

4634
02:55:39,372 --> 02:55:40,807
AGGREGATION TYPICALLY IS NOT A

4635
02:55:40,807 --> 02:55:43,943
PROBLEM FOR THESE PROTEINS.

4636
02:55:43,943 --> 02:55:44,477
NGOT IT.

4637
02:55:44,477 --> 02:55:46,779
THERE IS A SIMILAR QUESTION

4638
02:55:46,779 --> 02:55:47,980
REGARDING IMMUNO GENERATEDISIT

4639
02:55:47,980 --> 02:55:49,081
OF THESE DESIGN OF PROTEINS BUT

4640
02:55:49,081 --> 02:55:51,484
I THINK YOU TOUCHED BASE ON THAT

4641
02:55:51,484 --> 02:55:51,717
AS WELL.

4642
02:55:51,717 --> 02:55:55,154
DO YOU HAVE PLANS FOR USING YOUR

4643
02:55:55,154 --> 02:55:57,390
MODELS LIKE,A ROUND DIFFUSION IN

4644
02:55:57,390 --> 02:55:59,826
THE FIELD OF [INDISCERNIBLE]

4645
02:55:59,826 --> 02:56:00,126
BIOLOGY IN.

4646
02:56:00,126 --> 02:56:02,161
>> YEAH, WE'RE DOING QUITE
A LOT

4647
02:56:02,161 --> 02:56:02,495
OF THAT.

4648
02:56:02,495 --> 02:56:04,263
AND YOU KNOW SYNTHETIC BIOLOGY

4649
02:56:04,263 --> 02:56:08,034
MEANS A WIDE VARIETY OF THINGS,

4650
02:56:08,034 --> 02:56:10,603
BUT YEAH, SO, I JUST FOCUSED ON

4651
02:56:10,603 --> 02:56:12,104
BINDER DESIGN, AND THERAPEUTICS

4652
02:56:12,104 --> 02:56:13,973
FOR THIS TALK, BUT FOR SYNTHETIC

4653
02:56:13,973 --> 02:56:17,944
BIOLOGY, I MEAN WE'RE MAKING

4654
02:56:17,944 --> 02:56:20,046
ORGANIZATIONS THOGINAL RECEPTOR

4655
02:56:20,046 --> 02:56:22,615
LIGAND PAIRS FOR EXAMPLE, TO ARK

4656
02:56:22,615 --> 02:56:25,685
LOW FOR THE ORTHOGONAL

4657
02:56:25,685 --> 02:56:27,954
SIGNALING, WE ARE MAKING ENZYMES

4658
02:56:27,954 --> 02:56:31,123
FOR NEW SYNTHETIC APPROACHES AND

4659
02:56:31,123 --> 02:56:33,359
I THINK YEAH, WHEN I THINK ABOUT

4660
02:56:33,359 --> 02:56:34,193
SYNTHETIC BIOLOGY, I THINK

4661
02:56:34,193 --> 02:56:36,696
ABOUT, I SORT OF THINK THAT'S

4662
02:56:36,696 --> 02:56:37,730
WHAT YEAR DOG.

4663
02:56:37,730 --> 02:56:39,899
A LOT OF SYNTHETIC BIOLOGY IS

4664
02:56:39,899 --> 02:56:40,933
MIX XG MATCHING PARTS THAT ARE

4665
02:56:40,933 --> 02:56:42,702
ALREADY IN NATURE BUT THE NEXT,

4666
02:56:42,702 --> 02:56:47,807
THE FUTURE OF SYNTHETIC BIOLOGY

4667
02:56:47,807 --> 02:56:49,175
IS BUILDING EVERYTHING FROM

4668
02:56:49,175 --> 02:56:50,476
SCRATCH TO DO WHAT YOU WANT.

4669
02:56:50,476 --> 02:56:51,777
>> I SEE, THE FOLLOW UP
QUESTION

4670
02:56:51,777 --> 02:56:53,613
TO THAT IS ARE THERE ANY PLANS

4671
02:56:53,613 --> 02:56:58,217
TO USE ANY [INDISCERNIBLE]

4672
02:56:58,217 --> 02:56:59,785
PROTEIN BINDERS?

4673
02:56:59,785 --> 02:57:01,053
>> SO WE CAN INCORPORATE
NATURAL

4674
02:57:01,053 --> 02:57:04,924
AMINO ACIDS AND WE DO THAT IN

4675
02:57:04,924 --> 02:57:08,961
SOME CASES TO GET A COVALENT

4676
02:57:08,961 --> 02:57:11,430
INTERACTION WITH THE TARGET WE

4677
02:57:11,430 --> 02:57:13,833
DO THAT FOR ENZYMES BECAUSE WE

4678
02:57:13,833 --> 02:57:14,967
CAN GET UNNATURAL CHEMISTRIES IN

4679
02:57:14,967 --> 02:57:17,036
THAT WAY, YOU TAKE THE LITTLE

4680
02:57:17,036 --> 02:57:18,604
BIT OF A PRICE, BECAUSE OF THE

4681
02:57:18,604 --> 02:57:19,705
MANUFACTURING GETS HARD, THE

4682
02:57:19,705 --> 02:57:24,744
MAIN WAY WE'VE DONE THIS IS MY

4683
02:57:24,744 --> 02:57:25,311
COLLEAGUE [INDISCERNIBLE] IS

4684
02:57:25,311 --> 02:57:26,078
LEADING THE GROUND ON THAT

4685
02:57:26,078 --> 02:57:28,481
BECAUSE WE CAN USE THESE SAME

4686
02:57:28,481 --> 02:57:29,682
TECHNIQUES TO MAKE THE PEPTIDES

4687
02:57:29,682 --> 02:57:35,087
AND THERE WE CAN, SENTENCE WE'RE

4688
02:57:35,087 --> 02:57:36,222
CHEMICALLY SINCEICIZING THEM, WE

4689
02:57:36,222 --> 02:57:40,259
CAN USE A WIDE RANGE OF AMINO

4690
02:57:40,259 --> 02:57:42,028
ACIDS.

4691
02:57:42,028 --> 02:57:42,361
>> GOT IT.

4692
02:57:42,361 --> 02:57:44,130
YOU MENTIONED FRAMING THE

4693
02:57:44,130 --> 02:57:45,831
PROTEIN, PROTEIN COMPLEX DATA,

4694
02:57:45,831 --> 02:57:48,568
THIS DATA TYPE, IT'S THE BIGGEST

4695
02:57:48,568 --> 02:57:51,103
BOTTLENECK TO IMPROVE THE

4696
02:57:51,103 --> 02:57:52,038
PERFORMANCE, BASICALLY, I THINK

4697
02:57:52,038 --> 02:57:53,573
IT'S ASKING ABOUT THE

4698
02:57:53,573 --> 02:57:54,974
[INDISCERNIBLE], RIGHT?

4699
02:57:54,974 --> 02:57:56,208
>> YEAH, I THINK 1 OF THE
THINGS

4700
02:57:56,208 --> 02:57:57,810
I THINK EVERYONE KNOWS, BUT YOU

4701
02:57:57,810 --> 02:58:00,580
KNOW THE REASON WHY, THE STUFF

4702
02:58:00,580 --> 02:58:02,114
WE'RE DOING AND THE DEEP LINE

4703
02:58:02,114 --> 02:58:04,317
STUFF WORKS IS BECAUSE THE BDB

4704
02:58:04,317 --> 02:58:05,751
IS SUCH AN AMAZING RESOURCE AND

4705
02:58:05,751 --> 02:58:08,854
IT HAS THE WORK OF, YOU KNOW 10S

4706
02:58:08,854 --> 02:58:11,190
OF THOUSANDS OF PEOPLE OVER 60

4707
02:58:11,190 --> 02:58:13,092
YEARS WITH THE COST OF 10S OF

4708
02:58:13,092 --> 02:58:14,393
BILLIONS OF DOLLARS AND IT'S NOT

4709
02:58:14,393 --> 02:58:16,195
ONLY AN INCREDIBLE ARK MOUNT OF

4710
02:58:16,195 --> 02:58:18,898
DATA BUT IT'S INCREDIBLY MORE

4711
02:58:18,898 --> 02:58:20,266
CURATED SO MORE STRUCTURES WOULD

4712
02:58:20,266 --> 02:58:20,700
HELP.

4713
02:58:20,700 --> 02:58:21,834
SO I THINK ANOTHER PLACE IS IF

4714
02:58:21,834 --> 02:58:24,136
YOU WANT TO TRAIN A MODEL TO

4715
02:58:24,136 --> 02:58:25,771
PREDICTIVE AFFINITY WE DON'T

4716
02:58:25,771 --> 02:58:27,873
HAVE THOSE CORRESPONDING DATA

4717
02:58:27,873 --> 02:58:29,275
SETS FOR AFFINITY, AND SO, IT'S

4718
02:58:29,275 --> 02:58:32,378
HARD TO TRAIN MODELS TO DO THAT.

4719
02:58:32,378 --> 02:58:33,646
SO MORE STRUCTURES WOULD BE

4720
02:58:33,646 --> 02:58:33,846
GOOD.

4721
02:58:33,846 --> 02:58:36,782
ONE PLACE WHERE THAT WILL BE

4722
02:58:36,782 --> 02:58:39,552
GOOD IS WHERE DESIGNING TCRs

4723
02:58:39,552 --> 02:58:40,886
NOW TO MHCs AND THERE ARE ONLY

4724
02:58:40,886 --> 02:58:42,888
ABOUT A HUNDRED STRUCTURES SO

4725
02:58:42,888 --> 02:58:44,490
THEY'RE MORE STRUCTURES WILL BE

4726
02:58:44,490 --> 02:58:46,559
USEFUL, CERTAINLY MORE

4727
02:58:46,559 --> 02:58:47,960
STRUCTURES OF PROTEINS MORE

4728
02:58:47,960 --> 02:58:49,495
MOLECULAR COMPLEXES AND MORE

4729
02:58:49,495 --> 02:58:50,730
DIVERSE 1S AND WE SPENT QUITE A

4730
02:58:50,730 --> 02:58:52,298
BIT OF EFFORT TO GET PHARMA

4731
02:58:52,298 --> 02:58:53,366
CUTCHES TO MAKE THOSE AVAILABLE,

4732
02:58:53,366 --> 02:58:54,567
BECAUSE YOU KNOW THEY OBVIOUSLY

4733
02:58:54,567 --> 02:58:56,902
HAD A LOT OF STRUCTURES OF

4734
02:58:56,902 --> 02:58:57,837
PROTEIN SMALL MOLECULE COMPLEX

4735
02:58:57,837 --> 02:59:00,473
ANDS THAT WOULD CERTAINLY HELP,

4736
02:59:00,473 --> 02:59:02,775
YOU KNOW FOR BECAUSE WE'RE

4737
02:59:02,775 --> 02:59:04,543
EXCITED ABOUT SMALL MOLECULE

4738
02:59:04,543 --> 02:59:06,112
DESIGN USING THESE METHODS BUT

4739
02:59:06,112 --> 02:59:08,114
THAT IS DEFINITELY DATA EMILY

4740
02:59:08,114 --> 02:59:08,381
THED NOW.

4741
02:59:08,381 --> 02:59:10,383
YAW THAT WOULD BE GREAT TO SEE

4742
02:59:10,383 --> 02:59:12,418
THAT PROGRESS IN THAT FIELD AS

4743
02:59:12,418 --> 02:59:13,252
WELL.

4744
02:59:13,252 --> 02:59:15,221
IS THE OFF-TARGET

4745
02:59:15,221 --> 02:59:15,755
[INDISCERNIBLE] PROTEINS A

4746
02:59:15,755 --> 02:59:16,856
PROBLEM IN THE FIELD AND IF IT

4747
02:59:16,856 --> 02:59:20,326
IS, HOW IS IT THAT AGGRESSIVE

4748
02:59:20,326 --> 02:59:20,493
IN.

4749
02:59:20,493 --> 02:59:21,193
>> IT DOESN'T REALLY SEEM
TO BE.

4750
02:59:21,193 --> 02:59:23,062
THIS IS WHAT YOU THINK WOULD BE

4751
02:59:23,062 --> 02:59:24,463
THE HARDEST PROBLEM BECAUSE THIS

4752
02:59:24,463 --> 02:59:26,832
IS LIKE DISCIPLINARY NOEVERIN A

4753
02:59:26,832 --> 02:59:27,466
VERSUS DISCIPLINARY NOEVERIN B

4754
02:59:27,466 --> 02:59:30,069
AND THESE ARE SOME RELATED

4755
02:59:30,069 --> 02:59:31,771
PROTEINS AND PEPTIDES, AND THESE

4756
02:59:31,771 --> 02:59:33,906
ARE ALL DISORDERED SO THERE'S NO

4757
02:59:33,906 --> 02:59:35,174
STRUCTURE, WE'RE JUST DOG AN

4758
02:59:35,174 --> 02:59:36,008
EXPERIMENT NOW, IT'S PRETTY

4759
02:59:36,008 --> 02:59:36,375
COOL.

4760
02:59:36,375 --> 02:59:39,779
SO WE HAVE MADE ABOUT 250

4761
02:59:39,779 --> 02:59:41,013
BINDERS TO ABOUT 250 DIFFERENT

4762
02:59:41,013 --> 02:59:44,316
TARGETS NOW AND WE HAVE THIS

4763
02:59:44,316 --> 02:59:45,618
GREAT COLLABORATION WITH THE

4764
02:59:45,618 --> 02:59:47,186
MENS LAB BUT THEY'RE DOING PULL

4765
02:59:47,186 --> 02:59:48,554
DOWNS ON THOSE, SO WE WILL

4766
02:59:48,554 --> 02:59:50,956
REALLY LEARN WHAT THE

4767
02:59:50,956 --> 02:59:53,626
SPECIFICITY IS, BUT SO FAR, --

4768
02:59:53,626 --> 02:59:55,127
SO FAR THEY LOOK TO BE PRETTY

4769
02:59:55,127 --> 02:59:56,429
SPECIFIC, I MEAN THE OTHER THING

4770
02:59:56,429 --> 02:59:58,264
WE CAN DO IS KNOCK OUT THE

4771
02:59:58,264 --> 03:00:00,666
TARGET PROTEIN AND SEE IF

4772
03:00:00,666 --> 03:00:01,400
THERE'S ANY RESIDUAL BINDING IN

4773
03:00:01,400 --> 03:00:03,602
THE CELL SO WE WILL LEARN A LOT

4774
03:00:03,602 --> 03:00:04,670
ABOUT SPECIFICITY, BUT WE'VE

4775
03:00:04,670 --> 03:00:06,272
DONE THIS AGAINST ALL

4776
03:00:06,272 --> 03:00:08,107
EXPERIMENTS AGAINST PURIFIED

4777
03:00:08,107 --> 03:00:11,043
PROTEINS, THEY'RE USUALLY PRETTY

4778
03:00:11,043 --> 03:00:11,310
SPECIFIC.

4779
03:00:11,310 --> 03:00:12,578
>> IS ANY FREE RADICALS

4780
03:00:12,578 --> 03:00:13,512
GENERATED ON DESIGNING BINDERS

4781
03:00:13,512 --> 03:00:17,083
FOR THE RNA SPECIFICALLY, LIKE

4782
03:00:17,083 --> 03:00:19,552
RNA BINDING PROTEINS >> 
YEAH,

4783
03:00:19,552 --> 03:00:21,587
YEAH, WE ARE DESIGNING PROTEINS

4784
03:00:21,587 --> 03:00:23,122
NOW, WE HAVE BINDERS, NOT CLEAR

4785
03:00:23,122 --> 03:00:26,125
HOW SPECIFIC THEY ARE, WE ARE A

4786
03:00:26,125 --> 03:00:27,326
LOT FURTHER ALONG WITH DNA

4787
03:00:27,326 --> 03:00:28,394
BINDING PROTEINS AND THERE WE

4788
03:00:28,394 --> 03:00:30,196
CAN MAKE THE DNA BINDING

4789
03:00:30,196 --> 03:00:32,698
PROTEINS AND WE'RE INCORPORATING

4790
03:00:32,698 --> 03:00:35,935
NUCLEACE SITES INTO THEM TO SAY

4791
03:00:35,935 --> 03:00:40,372
-- WITH THE  BINDER STUFF NOW, I

4792
03:00:40,372 --> 03:00:44,143
TELL MY STUDENTS THAT I CAN'T

4793
03:00:44,143 --> 03:00:46,912
WORK ON THOSE NOW TO SOLVE A

4794
03:00:46,912 --> 03:00:52,585
PROBLEM BUT WE ARE INCORPORATING

4795
03:00:52,585 --> 03:00:54,386
CLEAVE TARGETS INTO IT.

4796
03:00:54,386 --> 03:00:56,455
>> WHAT IS YOUR VISION IN
TERMS

4797
03:00:56,455 --> 03:00:59,759
OF USING SYNTHETIC PROTEINS IN

4798
03:00:59,759 --> 03:00:59,992
HUMANS?

4799
03:00:59,992 --> 03:01:03,329
>> WELL, I THINK, YOU
KNOW, I

4800
03:01:03,329 --> 03:01:05,464
THINK THAT I'M VERY OPTIMISTIC

4801
03:01:05,464 --> 03:01:07,299
ABOUT THESE KIND OF TECHNIQUES

4802
03:01:07,299 --> 03:01:07,967
REALLY TRANSFOR THE PURPOSING

4803
03:01:07,967 --> 03:01:08,234
MEDICINE.

4804
03:01:08,234 --> 03:01:10,069
I MEAN YOU CAN GET -- YOU CAN

4805
03:01:10,069 --> 03:01:11,737
JUST GET MUCH MORE

4806
03:01:11,737 --> 03:01:12,505
SOPHISTICATION LIKE A WAS

4807
03:01:12,505 --> 03:01:14,240
SHOWING YOU COULD GET, RIGHT

4808
03:01:14,240 --> 03:01:16,509
NOW, YOU KNOW WE TREAT

4809
03:01:16,509 --> 03:01:18,344
AUTOIMMUNE DISEASE IN CANCER

4810
03:01:18,344 --> 03:01:21,046
WITH THESE KIND OF GLOBAL IMMUNE

4811
03:01:21,046 --> 03:01:23,883
SUPPRESS ANTS OR GLOBAL IMMUNE

4812
03:01:23,883 --> 03:01:25,751
ACTIVATORS, YOU CAN GET MUCH

4813
03:01:25,751 --> 03:01:27,386
MORE PRECISION OR MAKE MEDICINES

4814
03:01:27,386 --> 03:01:29,088
THAT ACT IN THE RIGHT TIME,

4815
03:01:29,088 --> 03:01:31,323
PLACE AND BODY WITH MORE

4816
03:01:31,323 --> 03:01:32,358
INTRICATE CONTROL MECHANISMS

4817
03:01:32,358 --> 03:01:33,259
WITH THESE DESIGN APPROACHES

4818
03:01:33,259 --> 03:01:37,930
THAN YOU CAN JUST BELLING OFF

4819
03:01:37,930 --> 03:01:38,230
ANTIBODIES.

4820
03:01:38,230 --> 03:01:38,898
>> I SEE.

4821
03:01:38,898 --> 03:01:44,837
DO YOU SEE YOURSELF THAT THE

4822
03:01:44,837 --> 03:01:46,605
MORE DIFFUSION MODELS,

4823
03:01:46,605 --> 03:01:47,406
CRYSTALLOGRAPHY [INDISCERNIBLE]

4824
03:01:47,406 --> 03:01:48,674
IN THE NEAR FUTURE, LIKE THE

4825
03:01:48,674 --> 03:01:50,209
QUALITY OF THAT, WHATEVER YOUR

4826
03:01:50,209 --> 03:01:51,944
GENERATE WOULD BE MATCHING THE

4827
03:01:51,944 --> 03:01:59,051
QUALITY OF THE PDB?

4828
03:01:59,051 --> 03:02:01,453
>> WELL WHEN WE SOLVE
STRUCTURED

4829
03:02:01,453 --> 03:02:03,422
FOR THESE DESIGN PROTEINS, THEY

4830
03:02:03,422 --> 03:02:07,059
ARE OFTEN ESSENTIALLY IDENTICAL

4831
03:02:07,059 --> 03:02:09,628
TO THE CRYSTAL STRUCTURES ARE

4832
03:02:09,628 --> 03:02:11,497
IDEBTICAL TO DESIGN MODELS,

4833
03:02:11,497 --> 03:02:12,364
WE'RE GERONTOLOGYSTSING, IT'S

4834
03:02:12,364 --> 03:02:14,166
NOT ALWAYS TRUE BUT WE'RE OFTEN

4835
03:02:14,166 --> 03:02:17,369
-- WE'RE GETTING REALLY, REALLY

4836
03:02:17,369 --> 03:02:18,637
HIGH ACCURACY SO ON THIS 1 THERE

4837
03:02:18,637 --> 03:02:20,372
ARE A LITTLE BIT OF DIFFERENCES

4838
03:02:20,372 --> 03:02:21,640
SO THIS 1 ISN'T RIGHT ON BUT WE

4839
03:02:21,640 --> 03:02:24,109
HAVE SOME OF THE SIDE CHAINS ARE

4840
03:02:24,109 --> 03:02:25,177
POINTING IN THE WRONG DIRECTION

4841
03:02:25,177 --> 03:02:27,112
BUT THERE ARE OTHER EXAMPLES I

4842
03:02:27,112 --> 03:02:28,013
SHOWED HERE, LET'S SEE, LIKE I

4843
03:02:28,013 --> 03:02:30,616
THINK I HAD A PICTURE OF A

4844
03:02:30,616 --> 03:02:37,690
CRYSTAL STRUCTURE, WHERE WAS IT

4845
03:02:37,690 --> 03:02:39,191
IN I DON'T KNOW WHERE ELSE I HAD

4846
03:02:39,191 --> 03:02:40,793
BUT A LOT OF THESE ARE -- I

4847
03:02:40,793 --> 03:02:42,127
GUESS I DON'T HAVE EXACT THINGS

4848
03:02:42,127 --> 03:02:44,697
BUT A LOT OF THESE ARE JUST

4849
03:02:44,697 --> 03:02:45,331
EXACTLY SPOT-ON.

4850
03:02:45,331 --> 03:02:47,166
OH YEAH, THIS IS 2 CRYSTAL

4851
03:02:47,166 --> 03:02:48,334
STRUCTURES AGAIN WHICH ARE

4852
03:02:48,334 --> 03:02:49,034
ALMOST --

4853
03:02:49,034 --> 03:02:53,272
>> I SEE, YEAH, IT'S VERY

4854
03:02:53,272 --> 03:02:54,240
IMPRESS OF RESULT.

4855
03:02:54,240 --> 03:02:55,374
IN THE FUTURE WHICH RESEARCH

4856
03:02:55,374 --> 03:02:57,509
DIRECTION DO YOU THINK WILL BE

4857
03:02:57,509 --> 03:02:59,144
YOUR TOP PRIORITY RELATED TO

4858
03:02:59,144 --> 03:03:00,212
THESE ISSUES.

4859
03:03:00,212 --> 03:03:02,514
THAT'S AGAIN THE AUDIENCE

4860
03:03:02,514 --> 03:03:04,049
REALLY.

4861
03:03:04,049 --> 03:03:05,517
>> WELL, LET'S SEE, TODAY,
I'M

4862
03:03:05,517 --> 03:03:09,521
KIND OF INTERESTED IN A BROAD

4863
03:03:09,521 --> 03:03:10,856
RANGE OF THINGS EMPLOY AS FAR AS

4864
03:03:10,856 --> 03:03:12,992
WHAT I TALKED ABOUT TODAY AND

4865
03:03:12,992 --> 03:03:13,659
MAKING PROTEIN THERAPEUTICS, I

4866
03:03:13,659 --> 03:03:15,160
GUESS LIKE I SAID, MY INTEREST

4867
03:03:15,160 --> 03:03:19,431
IS REALLY IN MAKING CONDITIONAL

4868
03:03:19,431 --> 03:03:20,466
MEDICINE, THINGS THAT ARE MUCH

4869
03:03:20,466 --> 03:03:21,500
MORE TARGETED AND SELECTED

4870
03:03:21,500 --> 03:03:23,168
EMPLOY I'M ALSO INTERESTED, YOU

4871
03:03:23,168 --> 03:03:25,604
KNOW I TOLD YOU ABOUT MAKING

4872
03:03:25,604 --> 03:03:27,039
SITE SPECIFIC PROTEASES, SO IF

4873
03:03:27,039 --> 03:03:28,073
WE COULD MAKE -- IMAGINE, YOU

4874
03:03:28,073 --> 03:03:29,275
KNOW THINK ABOUT THE COST AND

4875
03:03:29,275 --> 03:03:32,544
THE LOAD ON THE BOOED, AND

4876
03:03:32,544 --> 03:03:33,445
MAKING THERAPEUTIC ANTAGONISTIC

4877
03:03:33,445 --> 03:03:34,914
ANTIBODIES, BUT IT MAKES SMALL

4878
03:03:34,914 --> 03:03:36,682
PROTEINS THAT JUST WENT IN AND

4879
03:03:36,682 --> 03:03:38,417
DESTROYED THE TARGET FOR

4880
03:03:38,417 --> 03:03:40,219
ANTIVIRALS OR AGAINST TUMOR

4881
03:03:40,219 --> 03:03:41,553
PROTEIN CELL SURFACE PROTEINS, I

4882
03:03:41,553 --> 03:03:43,656
THINK THAT'S GOING TO BE A

4883
03:03:43,656 --> 03:03:44,156
THING.

4884
03:03:44,156 --> 03:03:45,291
I'M ERROR EXCITED ABOUT MAKING

4885
03:03:45,291 --> 03:03:46,125
MODEL CITIZEN LEAKULAR MACHINES

4886
03:03:46,125 --> 03:03:49,962
SO IN OUR CELLS WE HAVE MODEL

4887
03:03:49,962 --> 03:03:54,199
CITIZEN LEAKULAR CHAPERONS LIKE

4888
03:03:54,199 --> 03:03:55,167
HSP90, AND 70 FOR QUALITY

4889
03:03:55,167 --> 03:03:56,468
CONTROL, THERE'S NOTHING LIKE

4890
03:03:56,468 --> 03:03:58,671
THAT IN PRODUCK, BUT I THINK WE

4891
03:03:58,671 --> 03:04:00,940
CAN DESIGN MACHINES LIKE THAT

4892
03:04:00,940 --> 03:04:02,474
THAT CAN GO AROUND AND CLEAN

4893
03:04:02,474 --> 03:04:03,242
THINGS UP.

4894
03:04:03,242 --> 03:04:04,843
>> GOT IT AND I GUESS THE
LAST

4895
03:04:04,843 --> 03:04:06,712
QUESTION I WILL TAKE, DO YOU

4896
03:04:06,712 --> 03:04:07,680
INCORPORATE THE DYNAMIC NATURE

4897
03:04:07,680 --> 03:04:09,782
OF THE PROTEINS IN YOUR

4898
03:04:09,782 --> 03:04:12,184
COMPUTATIONAL DESIGN PROCESS FOR

4899
03:04:12,184 --> 03:04:14,586
INSTANCE THE DEFECTORS OF ANY

4900
03:04:14,586 --> 03:04:15,888
[INDISCERNIBLE] OF DIE

4901
03:04:15,888 --> 03:04:16,288
NAMES--NAMESSICS?

4902
03:04:16,288 --> 03:04:16,422
?

4903
03:04:16,422 --> 03:04:17,790
YEAH, WELL, IT'S HARD.

4904
03:04:17,790 --> 03:04:20,626
THERE JUST ISN'T A LOT OF DATA

4905
03:04:20,626 --> 03:04:22,528
ON DYNAMICS, YOU CAN DO MD

4906
03:04:22,528 --> 03:04:25,831
SIMULATIONS BUT IN TERMS OF

4907
03:04:25,831 --> 03:04:28,600
ACTUALLY TRAINING ON DYNAMICS

4908
03:04:28,600 --> 03:04:30,869
DATA, IT'S -- THERE ISN'T REALLY

4909
03:04:30,869 --> 03:04:33,038
A LOT OF, THE NMR DATA ISN'T

4910
03:04:33,038 --> 03:04:34,406
COLLECTED IN A WAY YOU CAN TRAIN

4911
03:04:34,406 --> 03:04:37,309
ON, SO I THINK THERE WE ARE

4912
03:04:37,309 --> 03:04:38,978
LIMITED BY TRAINING DATA.

4913
03:04:38,978 --> 03:04:41,680
I THINK THINGS LIKE ENZYMES, I

4914
03:04:41,680 --> 03:04:42,948
DIDN'T TALK ABOUT CATALIAISONS

4915
03:04:42,948 --> 03:04:44,016
DESIGN BUT UNDERSTANDING HOW

4916
03:04:44,016 --> 03:04:47,753
THINGS MOVE IS CLEARLY

4917
03:04:47,753 --> 03:04:48,020
IMPORTANT.

4918
03:04:48,020 --> 03:04:48,354
NGOT IT.

4919
03:04:48,354 --> 03:04:50,856
THANK YOU SO MUCH FOR YOUR TIME

4920
03:04:50,856 --> 03:04:51,090
TODAY.

4921
03:04:51,090 --> 03:04:52,424
IT'S A REALLY GREAT PRESENTATION

4922
03:04:52,424 --> 03:04:53,993
AND ANSWER TO THE QUESTION,

4923
03:04:53,993 --> 03:04:54,994
REALLY APPRECIATE IT.

4924
03:04:54,994 --> 03:04:57,062
>> OKAY, ALL RIGHT,
THANKS!

4925
03:04:57,062 --> 03:04:59,765
>> THANK YOU VERY MUCH.

4926
03:04:59,765 --> 03:05:02,267
SO I GUESS IF THAT'S THAT, WE

4927
03:05:02,267 --> 03:05:04,203
ARE GOING TO CLOSE THIS SESSION

4928
03:05:04,203 --> 03:05:09,241
NOW AND WE WILL HAVE THE SLIDE

4929
03:05:09,241 --> 03:05:12,845
30 MINUTES BREAK UNTIL 3:00 P.M.

4930
03:05:12,845 --> 03:05:14,680
AND WHEN WE ARE REALLY EXCITED

4931
03:05:14,680 --> 03:05:18,317
TO GO TO OUR PANEL DISCUSSION SO

4932
03:05:18,317 --> 03:05:18,784
PLEASE STAY TUNED.

4933
03:05:18,784 --> 03:05:21,420
COMING BACK AND BE READY TO ASK

4934
03:05:21,420 --> 03:05:24,723
QUESTIONS AND DISCUSSIONS. 

4935
03:05:24,723 --> 03:05:29,328
BACK TO YOU.

4936
03:05:29,328 --> 03:05:31,263
>> THANK YOU ALEXEY, THAT
WAS

4937
03:05:31,263 --> 03:05:34,166
SUCH AN AWESOME SESSION, I HAVE

4938
03:05:34,166 --> 03:05:37,403
THE AGENDA UP LIKE ALEXEY SAID,

4939
03:05:37,403 --> 03:05:39,004
WE WILL BE BACK AT 3:00 SO

4940
03:05:39,004 --> 03:05:40,906
EVERYONE GETS A BIT OF A BREAK

4941
03:05:40,906 --> 03:05:42,408
BUT MAKE SURE TO COME BACK AT

4942
03:05:42,408 --> 03:05:43,909
THRESHOLD BECAUSE IT WILL BE A

4943
03:05:43,909 --> 03:05:44,710
SUPER INTERESTING SESSION, WE

4944
03:05:44,710 --> 03:05:45,944
WILL DISCUSS A LOT OF THE GAPS

4945
03:05:45,944 --> 03:05:47,046
IN THE FIELD AND WE'RE GOING TO

4946
03:05:47,046 --> 03:05:50,149
TAKE A LOT OF QUESTIONS AS WELL.

4947
03:05:50,149 --> 03:05:52,151
ESPECIALLY THE 1S WE COULDN'T

4948
03:05:52,151 --> 03:05:53,919
ANSWER IN TODAY, AND YESTERDAY.

4949
03:05:53,919 --> 03:06:00,661
ALL RIGHT, BACK AT 3:00.

4950
03:06:00,661 --> 03:06:01,862
>> HI, EVERYONE, WELCOME
BACK,

4951
03:06:01,862 --> 03:06:02,897
WE'RE SUPER EXCITED TO HAVE YOU

4952
03:06:02,897 --> 03:06:05,866
BACK WITH US TODAY FOR A FINAL

4953
03:06:05,866 --> 03:06:07,802
SESSION HERE ON BRIDGING THE

4954
03:06:07,802 --> 03:06:08,636
GAPS IN TRANSLATION.

4955
03:06:08,636 --> 03:06:10,171
SO FOR THIS I WOULD LIKE TO

4956
03:06:10,171 --> 03:06:11,772
INTRODUCE THE CHAIR ISSUES

4957
03:06:11,772 --> 03:06:13,874
MODERATOR AND THE SPEAKER IN

4958
03:06:13,874 --> 03:06:19,747
THIS SESSION DR. ROMMIE AMARO,

4959
03:06:19,747 --> 03:06:26,554
ROMMIE, HOLDS THE DISTINGUISHED

4960
03:06:26,554 --> 03:06:28,422
POSITION OF DIRECTOR OF

4961
03:06:28,422 --> 03:06:29,490
UNIVERSITY OF CALIFORNIA AT SAN

4962
03:06:29,490 --> 03:06:31,058
DISCIPLINARY OAGY, HE AS A

4963
03:06:31,058 --> 03:06:32,393
PRESIDENTIAL EARLY CAREER,A

4964
03:06:32,393 --> 03:06:34,762
WARDS OF SCIENCE AND ENGINEERS

4965
03:06:34,762 --> 03:06:37,131
AND OTHERS AND HER SCIENTIFIC

4966
03:06:37,131 --> 03:06:40,234
INTERESTS LIE AT THE COMPUTER

4967
03:06:40,234 --> 03:06:44,438
AIDED DRUG DISCOVERY AND

4968
03:06:44,438 --> 03:06:46,407
BIOPHYSICAL SIMULATION.

4969
03:06:46,407 --> 03:06:47,942
ROMMIE, THE FLOOR IS YOURS.

4970
03:06:47,942 --> 03:06:50,444
WE ARE EXCITED TO HEAR THE

4971
03:06:50,444 --> 03:06:50,678
SESSION.

4972
03:06:50,678 --> 03:06:51,112
NTHANK YOU.

4973
03:06:51,112 --> 03:07:01,522
LET ME SHARE MY SLIDES.

4974
03:07:03,624 --> 03:07:07,128
OKAY, HOPEFULLY YOU ARE STARTING

4975
03:07:07,128 --> 03:07:07,461
TO SEE THOSE.

4976
03:07:07,461 --> 03:07:09,897
CAN YOU SEE EVERYTHING?

4977
03:07:09,897 --> 03:07:12,800
>> THANK YOU.

4978
03:07:12,800 --> 03:07:14,702
>> NO PROBLEM.

4979
03:07:14,702 --> 03:07:15,069
HELLO, EVERYONE.

4980
03:07:15,069 --> 03:07:16,537
I'M HAPPY TO BE HERE, I THINK SO

4981
03:07:16,537 --> 03:07:18,005
FAR IT'S BEEN A GREAT 2 DAYS AND

4982
03:07:18,005 --> 03:07:19,740
WE'RE BRINGING UP THE KACCT BOOS

4983
03:07:19,740 --> 03:07:22,009
NOW, AND WHERE WE WANTED TO

4984
03:07:22,009 --> 03:07:24,111
REALLY FOCUS IS IN BRIDGING THE

4985
03:07:24,111 --> 03:07:28,883
GAPS IN TRANSLATION, AND WE HAVE

4986
03:07:28,883 --> 03:07:29,750
OUR SESSION TODAY, I'M GOING TO

4987
03:07:29,750 --> 03:07:33,020
GIVE A LITTLE BIT OF AN INTRO,

4988
03:07:33,020 --> 03:07:34,688
EVERYBODY WILL GET 20 MINUES

4989
03:07:34,688 --> 03:07:36,524
INCLUDING MYSELF AND THEN PAT

4990
03:07:36,524 --> 03:07:38,292
WALTERS FROM RELAY THERAPEUTICS

4991
03:07:38,292 --> 03:07:40,895
IS GOING TO GIVE A 20 MINUTE

4992
03:07:40,895 --> 03:07:43,197
PERSPECTIVE ON DANGERS OF

4993
03:07:43,197 --> 03:07:45,032
IMPORTANT DATA, RUSS ALTMAN WILL

4994
03:07:45,032 --> 03:07:46,800
GIVE A PERSPECTIVE ON LANGUAGE

4995
03:07:46,800 --> 03:07:48,836
MODELS, AND THEIR EVALUATION OF

4996
03:07:48,836 --> 03:07:50,304
REGULATORY SUBMISSIONS, THEN WE

4997
03:07:50,304 --> 03:07:52,206
HAVE JOEL FROM GSK WHO WILL TALK

4998
03:07:52,206 --> 03:07:54,208
ABOUT TEAM ENVIRONMENT FOR

4999
03:07:54,208 --> 03:07:55,209
BETTER TRANSLATION AND THEN, WE

5000
03:07:55,209 --> 03:07:59,446
ARE GOING TO GO INTO -- WE WILL

5001
03:07:59,446 --> 03:08:01,015
GO DIRECTLY INTO A PANEL

5002
03:08:01,015 --> 03:08:03,717
DISCUSSION WITH ALL OF US HERE

5003
03:08:03,717 --> 03:08:06,053
INCLUDING MARTY HEAD FROM AMJEN

5004
03:08:06,053 --> 03:08:13,360
WHO WAS 1 OF THE OPENING

5005
03:08:13,360 --> 03:08:14,862
SPEAKERS YESTERDAY.

5006
03:08:14,862 --> 03:08:16,463
SO IT SHOULD BE GOOD.

5007
03:08:16,463 --> 03:08:18,165
SO WE WILL LET FOLKS GIVE A

5008
03:08:18,165 --> 03:08:21,268
TWEBT MINUTE TALK AND INSTEAD OF

5009
03:08:21,268 --> 03:08:23,003
HAVING Q&A, WE'RE GOING TO
SAVE

5010
03:08:23,003 --> 03:08:26,440
ALL THE Q&A DIRECTLY FOR
THAT

5011
03:08:26,440 --> 03:08:27,775
LAST 40 MINUTE SESSION AND I

5012
03:08:27,775 --> 03:08:31,579
ALSO WANT TO SAY TO INVITE FOLKS

5013
03:08:31,579 --> 03:08:33,581
AND REMIND AUDIENCE MEMBERS TO

5014
03:08:33,581 --> 03:08:34,515
PLEASE SUBMIT QUESTIONS FOR THAT

5015
03:08:34,515 --> 03:08:37,084
BECAUSE I THINK WHAT WE WANT TO

5016
03:08:37,084 --> 03:08:39,019
DO IS HAVE A GOOD DISCUSSION

5017
03:08:39,019 --> 03:08:40,554
ABOUT TOPICS THAT ARE OF

5018
03:08:40,554 --> 03:08:43,057
INTEREST TO FOLKS WHO ARE

5019
03:08:43,057 --> 03:08:44,692
ATTENDING AND MAYBE YOU KNOW GET

5020
03:08:44,692 --> 03:08:45,993
INSIGHTS THAT AREN'T NECESSARILY

5021
03:08:45,993 --> 03:08:47,962
PART OF LIKE A TYPICAL

5022
03:08:47,962 --> 03:08:48,295
PRESENTATION.

5023
03:08:48,295 --> 03:08:52,166
OKAY, AND THEN I HAVE TO CONFESS

5024
03:08:52,166 --> 03:08:53,400
THAT WHEN I WAS LISTENING TO

5025
03:08:53,400 --> 03:08:55,069
EVERYTHING, I WAS GETTING SO

5026
03:08:55,069 --> 03:08:56,237
EXCITED FOR THE SCIENCE BECAUSE

5027
03:08:56,237 --> 03:08:59,306
I CAN'T HELP IT.

5028
03:08:59,306 --> 03:09:01,075
INSTEAD OF JUST -- BECAUSE WE

5029
03:09:01,075 --> 03:09:04,445
HAVE 40 MINUTES AT THE END WHERE

5030
03:09:04,445 --> 03:09:05,746
WE CANIAMER AND DISCUSS

5031
03:09:05,746 --> 03:09:06,614
DIFFERENT CHALLENGES SO I WANT

5032
03:09:06,614 --> 03:09:07,848
TO GIVE A LITTLE BIT OF A VIEW

5033
03:09:07,848 --> 03:09:09,516
ABOUT WHAT WE ARE WORKING ON IN

5034
03:09:09,516 --> 03:09:13,354
THIS SPACE AND I THINK IT, YOU

5035
03:09:13,354 --> 03:09:15,256
KNOW IT'S, IT'S HARD TO FOLLOW

5036
03:09:15,256 --> 03:09:16,957
DAVID BAKER, IT'S NOT THE FIRST

5037
03:09:16,957 --> 03:09:18,892
TIME I'VE BEEN IN THIS POSITION,

5038
03:09:18,892 --> 03:09:21,862
DARN IT AND NOW WITH HIS NOBEL

5039
03:09:21,862 --> 03:09:24,198
PRIZE, CONGRATULATIONS DAVID

5040
03:09:24,198 --> 03:09:24,465
AGAIN.

5041
03:09:24,465 --> 03:09:25,633
BUT HE DID MENTION AT THE END

5042
03:09:25,633 --> 03:09:27,835
AND IT'S COME UP, YOU KNOW WHAT

5043
03:09:27,835 --> 03:09:29,136
ABOUT DYNAMICS AND YOU KNOW

5044
03:09:29,136 --> 03:09:30,170
DYNAMICS IS SOMETHING THAT

5045
03:09:30,170 --> 03:09:31,772
MYSELF AND GROUPS HAS BEEN SUPER

5046
03:09:31,772 --> 03:09:34,408
INTERESTED IN AND WE'VE HEARD A

5047
03:09:34,408 --> 03:09:36,510
BIT ABOUT IT AND IT'S ESPECIALLY

5048
03:09:36,510 --> 03:09:40,247
WHAT WE'RE INTERESTED TO DO IS

5049
03:09:40,247 --> 03:09:44,518
USE DATA CENTRIC PHYSICS BASED

5050
03:09:44,518 --> 03:09:46,654
MOLECULAR DEMONSTRATIONS AS A CO

5051
03:09:46,654 --> 03:09:47,288
CALLED COMPUTATIONAL MICROSCOPE

5052
03:09:47,288 --> 03:09:48,756
AND SO THIS MEANS WE CAN

5053
03:09:48,756 --> 03:09:50,557
INTEGRATE ALL TYPES OF DIFFERENT

5054
03:09:50,557 --> 03:09:52,026
EXPERIMENTAL DATA SETS COMING

5055
03:09:52,026 --> 03:09:53,127
FROM DIFFERENT MODALITIES, THEY

5056
03:09:53,127 --> 03:09:54,695
COULD BE DIFFERENT STRUCTURAL

5057
03:09:54,695 --> 03:09:58,365
DATA SETS TAKEN AT RESOLUTIONS,

5058
03:09:58,365 --> 03:10:01,201
LIKE CRYOEM, ET, LIPID OMICS,

5059
03:10:01,201 --> 03:10:02,703
FLI COLSIS COMICS, WE CAN

5060
03:10:02,703 --> 03:10:04,171
COMBINE ALL THESE DIFFERENT DATA

5061
03:10:04,171 --> 03:10:06,173
SES TAKEN IN DIFFERENT GROUPS

5062
03:10:06,173 --> 03:10:08,609
FROM DIFFERENT REPOSITORIES AND

5063
03:10:08,609 --> 03:10:11,612
THEN BUILD THESE SORT OF

5064
03:10:11,612 --> 03:10:13,514
SYSTEMS, THESE ATOMIC LEVEL

5065
03:10:13,514 --> 03:10:14,348
REPRESENTATIONS OF BIOLOGICAL

5066
03:10:14,348 --> 03:10:15,849
SYSTEMS THAT REALLY ARE SORT OF

5067
03:10:15,849 --> 03:10:18,285
ALLOWS FOR THE SORT OF COHESIVE

5068
03:10:18,285 --> 03:10:20,621
UNDERSTANDING OF DRUG TARGETS

5069
03:10:20,621 --> 03:10:24,825
AND AND THEN ALL WE'RE DOING IS

5070
03:10:24,825 --> 03:10:27,394
APPROX MIGHTING IT DOWN TO THE

5071
03:10:27,394 --> 03:10:29,196
ATOMS, WE HAVE THE FUNCTION HERE

5072
03:10:29,196 --> 03:10:31,598
THAT DESCRIBES EACH OF THE

5073
03:10:31,598 --> 03:10:32,366
INTERACS WITH EVERYTHING ELSE IN

5074
03:10:32,366 --> 03:10:34,068
THE SYSTEM AND I THINK JOHN

5075
03:10:34,068 --> 03:10:35,002
MIGHT HAVE TALKED A LITTLE BIT

5076
03:10:35,002 --> 03:10:38,339
BUT I'M NOT SURE IF HE TALKED A

5077
03:10:38,339 --> 03:10:39,506
LITTLE BIT ABOUT DYSFUNCTIONAL

5078
03:10:39,506 --> 03:10:40,708
FORM BUT THERE'S A LOT PEOPLE

5079
03:10:40,708 --> 03:10:43,010
ARE DOING TO TRY TO UPDATE THIS

5080
03:10:43,010 --> 03:10:44,044
FUNCTIONAL FORM WHICH FOLKS HAD

5081
03:10:44,044 --> 03:10:45,946
BEEN USING FROM THE 70S BUT IN

5082
03:10:45,946 --> 03:10:48,215
ANY CASE, THEN WE DO, WE

5083
03:10:48,215 --> 03:10:50,951
GENERATE OFTEN TONS OF DATA,

5084
03:10:50,951 --> 03:10:54,288
COULD BE TERRA BYTES OF DATA, OF

5085
03:10:54,288 --> 03:10:57,491
JUST SORT OF STRUCTURAL DYNAMIC

5086
03:10:57,491 --> 03:10:59,860
DATA SETS, THROUGH THE

5087
03:10:59,860 --> 03:11:01,261
INTEGRATION OF NEWTON'S EQUATION

5088
03:11:01,261 --> 03:11:02,696
OF TIME WITH THESE OTOMIC

5089
03:11:02,696 --> 03:11:04,098
COORDINATES AND SO THE ANALYSIS

5090
03:11:04,098 --> 03:11:05,165
OF ANOTHER PLACE WHERE THERE'S A

5091
03:11:05,165 --> 03:11:12,940
LOT OF JUST SORT OF OF COURSE

5092
03:11:12,940 --> 03:11:15,042
CODING HAS TAKEN PLACE BUT WHERE

5093
03:11:15,042 --> 03:11:17,144
MACHINE LEARNING HAS COME INTO

5094
03:11:17,144 --> 03:11:18,445
IT AND INTERPRET THE EXPERIMENTS

5095
03:11:18,445 --> 03:11:19,580
WE COULD ADVANCE OUR

5096
03:11:19,580 --> 03:11:20,781
UNDERSTANDING OF BIOLOGY AND

5097
03:11:20,781 --> 03:11:23,016
INCLUDING THE DEVELOPMENT OF NEW

5098
03:11:23,016 --> 03:11:25,219
SMALL MOLECULE DRUGS THAT FOR

5099
03:11:25,219 --> 03:11:26,587
EXAMPLE COMBINED CRYPTIC POCKETS

5100
03:11:26,587 --> 03:11:28,889
THAT AREN'T PRESENT IN THE

5101
03:11:28,889 --> 03:11:30,357
CRYSTAL GRAPHIC OR CRYOEM

5102
03:11:30,357 --> 03:11:31,592
STRUCTURES BECAUSE WE KNOW ALL

5103
03:11:31,592 --> 03:11:33,227
THESE TARGETS ARE DYNAMIC AND

5104
03:11:33,227 --> 03:11:35,696
OFTEN THEY DO OPEN UP NEW

5105
03:11:35,696 --> 03:11:36,530
POCKETS FOR SMALL MOLECULE

5106
03:11:36,530 --> 03:11:40,033
DISCOVERY, WHAT I WANTED TO TELL

5107
03:11:40,033 --> 03:11:42,336
YOU ABOUT TODAY ARE OUR EFFORTS

5108
03:11:42,336 --> 03:11:46,440
MORE ON PROTEIN DESIGN, AND AS

5109
03:11:46,440 --> 03:11:49,076
DAVID MENTIONED THERE'S BEEN A

5110
03:11:49,076 --> 03:11:50,010
HUGE SMALL PROTEIN BINDERS AND I

5111
03:11:50,010 --> 03:11:51,111
WILL ACTUALLY TAKE IT A LITTLE

5112
03:11:51,111 --> 03:11:56,316
BIT FURTHER AND TALK ABOUT HOW

5113
03:11:56,316 --> 03:11:57,785
SIMULATION CAN ALLOW US TO SEE

5114
03:11:57,785 --> 03:12:01,889
THE UNSEEN FOR TARGETS FOR

5115
03:12:01,889 --> 03:12:02,289
IMMUNOGEN DESIGN.

5116
03:12:02,289 --> 03:12:03,424
I DON'T WANT TO SPEND TOO MUCH

5117
03:12:03,424 --> 03:12:04,825
TIME ON JUST THE METHOD BUT

5118
03:12:04,825 --> 03:12:08,896
REALLY NOW SORT OF A MULTISCALE

5119
03:12:08,896 --> 03:12:09,463
COMPUTATIONAL MICROSCOPE

5120
03:12:09,463 --> 03:12:11,698
ACTUALLY THAT LETS FOLKS PROBE

5121
03:12:11,698 --> 03:12:13,200
DRUG TARGETS AND THEIR

5122
03:12:13,200 --> 03:12:15,035
ENVIRONMENTS FROM THE QUANTUM TO

5123
03:12:15,035 --> 03:12:15,602
THE MACROSCALE REALM.

5124
03:12:15,602 --> 03:12:17,004
WHAT I WANT TO TALK ABOUT TODAY

5125
03:12:17,004 --> 03:12:22,443
IS THE EFFORTS IN THE NANO

5126
03:12:22,443 --> 03:12:22,676
SCALE.

5127
03:12:22,676 --> 03:12:23,911
SO THE EXAMPLE I WANT TO TELL

5128
03:12:23,911 --> 03:12:25,979
YOU ABOUT QUICKLY IS WITH THE

5129
03:12:25,979 --> 03:12:29,716
SARS-COV-2 VIRUS, WE ALL KNOW

5130
03:12:29,716 --> 03:12:32,085
ABOUT THIS, IT'S A LIPID

5131
03:12:32,085 --> 03:12:33,420
PROFESSIONAL VIRUS, IT HAS

5132
03:12:33,420 --> 03:12:34,655
STRUCTURAL PROTEINS LIKE THE

5133
03:12:34,655 --> 03:12:36,323
EPROTEIN AND THERE'S A FEW TILED

5134
03:12:36,323 --> 03:12:38,325
INSIDE THE MEMBRANE, THERE'S

5135
03:12:38,325 --> 03:12:39,259
M-PROTEIN, HUNDREDS OF THOSE ON

5136
03:12:39,259 --> 03:12:41,195
THE INSIDE OF THE VIRAL MEMBRANE

5137
03:12:41,195 --> 03:12:44,097
THAT PROVIDE STRUCTURE AND THIS

5138
03:12:44,097 --> 03:12:46,233
IS AN EXAMPLE OF 1 OF THE

5139
03:12:46,233 --> 03:12:47,634
STRUCTURAL MODELS WE BUILT FOR

5140
03:12:47,634 --> 03:12:49,336
THE WHOLE VIRUS, WE AND MANY

5141
03:12:49,336 --> 03:12:51,338
FOLKS ARE INTERESTED IN THE

5142
03:12:51,338 --> 03:12:52,840
TBLIEK O PROTEINS INCLUDING THE

5143
03:12:52,840 --> 03:12:54,007
SPIKE PROTEIN BECAUSE IT SITS ON

5144
03:12:54,007 --> 03:12:56,210
OUTSIDE OF THE VIRUS AND PLAYS A

5145
03:12:56,210 --> 03:12:57,744
KEY ROLE IN INFECTION BUT IT

5146
03:12:57,744 --> 03:13:03,684
ALSO IS A KEY ELEMENT OF

5147
03:13:03,684 --> 03:13:06,420
IMMUNITY BECAUSE THAT'S WHERE A

5148
03:13:06,420 --> 03:13:09,223
LOT OF OUR ADAPTIVE IMMUNITY

5149
03:13:09,223 --> 03:13:10,357
RESPONSES ARE DIRECTED.

5150
03:13:10,357 --> 03:13:14,661
SO WE BUILT A COMPREHENSIVE

5151
03:13:14,661 --> 03:13:16,730
MODEL THAT INCLUDED SORT OF AN

5152
03:13:16,730 --> 03:13:18,832
INTEGRATIVE MODEL I SHOULD SAY

5153
03:13:18,832 --> 03:13:20,534
THAT INCLUDED BOTH CRYOEM

5154
03:13:20,534 --> 03:13:21,969
RESOLVED PARTS OF THIS AS WELL

5155
03:13:21,969 --> 03:13:23,637
AS USED PREDICTIONS FOR OTHER

5156
03:13:23,637 --> 03:13:26,573
BITS THAT WERE NOT ABLE TO BE

5157
03:13:26,573 --> 03:13:28,075
EXPERIMENTALLY RESOLVED THAT

5158
03:13:28,075 --> 03:13:29,276
INCLUDES THE GLYCANS THAT ARE

5159
03:13:29,276 --> 03:13:30,944
PRESENT IN THESE SYSTEMS WHICH

5160
03:13:30,944 --> 03:13:34,882
ARE ALSO HIGHLY MOBILE, THERE'S

5161
03:13:34,882 --> 03:13:36,450
ALSO THE TRANSMEMBRANE PART SO

5162
03:13:36,450 --> 03:13:38,352
REALLY THE ENTIRE END-TO-END

5163
03:13:38,352 --> 03:13:39,686
SEQUENCE IS REPRESENTED AT THE

5164
03:13:39,686 --> 03:13:42,923
ATOMIC LEVEL AND THEN WE CAN

5165
03:13:42,923 --> 03:13:45,559
LIKE I SAID SAYING BEFORE,

5166
03:13:45,559 --> 03:13:46,126
ANIMATE THE DYNAMICS.

5167
03:13:46,126 --> 03:13:47,995
THIS IS HOW WE GET THE PROTEINS,

5168
03:13:47,995 --> 03:13:50,330
THIS IS REALLY FORTUNATELY OR

5169
03:13:50,330 --> 03:13:51,765
UNFORTUNATELY COMPUTING AND

5170
03:13:51,765 --> 03:13:52,866
SIMULATION IS REALLY THE ONLY

5171
03:13:52,866 --> 03:13:57,471
WAY TO GET AT THESE HIGH

5172
03:13:57,471 --> 03:13:59,106
RESOLUTION DYNAMICAL DATA SETS,

5173
03:13:59,106 --> 03:14:00,874
THERE IS TIME RESOLUTION CRYOEM

5174
03:14:00,874 --> 03:14:02,342
NOW BUT YOU WILL ONLY RESOLVE

5175
03:14:02,342 --> 03:14:03,243
SOME STATES AND THAT'S STILL

5176
03:14:03,243 --> 03:14:06,446
SORT OF A WORK IN PROGRESS,

5177
03:14:06,446 --> 03:14:08,715
SIMULATIONS ALLOW US TO VIEW THE

5178
03:14:08,715 --> 03:14:12,219
DYNAMIC SYSTEMS WITH RELATIVELY

5179
03:14:12,219 --> 03:14:14,388
HIGH FIDELITY AND SO WE HAVE

5180
03:14:14,388 --> 03:14:15,689
BEEN INTERESTED, SO AGAIN GOING

5181
03:14:15,689 --> 03:14:20,160
BACK TO THE SPIKE PROTEIN, YOU

5182
03:14:20,160 --> 03:14:22,062
KNOW MOST OF WHAT WE HAVE

5183
03:14:22,062 --> 03:14:23,463
DESIGNED IS ACTUALLY TARGETING

5184
03:14:23,463 --> 03:14:26,533
THE ENTIRE SPIKE PROTEIN WHICH

5185
03:14:26,533 --> 03:14:28,802
INCLUDES THAT S1 DOMAIN SO THE

5186
03:14:28,802 --> 03:14:31,238
SPIKE PROTEIN HAS THE S-1 DOMAIN

5187
03:14:31,238 --> 03:14:32,406
WHICH POPPED OFF THERE AND THEN

5188
03:14:32,406 --> 03:14:33,707
THAT PEELS OFF DURING THE

5189
03:14:33,707 --> 03:14:35,876
PROCESS OF INFECTION AND THEN IS

5190
03:14:35,876 --> 03:14:37,744
LEFT WITH THE S2 DOMAIN SHOWN

5191
03:14:37,744 --> 03:14:39,346
HERE IN THE MIDDLE SO THIS IS

5192
03:14:39,346 --> 03:14:40,747
REALLY SORT OF THE INSIDE CORE

5193
03:14:40,747 --> 03:14:45,452
OF THE SPIKE PROTEIN AND SO WE

5194
03:14:45,452 --> 03:14:48,488
YOINED A TEAM OF RESEARCHERS,

5195
03:14:48,488 --> 03:14:53,026
JASON AND OTHERS AND TRIED TO

5196
03:14:53,026 --> 03:14:55,128
TARGET S2 THIS INSIDE CORE

5197
03:14:55,128 --> 03:14:57,331
POTENTIALLY FOR PAN CORONA VIRUS

5198
03:14:57,331 --> 03:14:58,298
VACCINE DEVELOPMENT BECAUSE WE

5199
03:14:58,298 --> 03:15:01,034
ALL KNOW THAT VERY UNFORTUNATELY

5200
03:15:01,034 --> 03:15:02,803
S1 MUTATES LIKE CRAZY AND SO THE

5201
03:15:02,803 --> 03:15:05,205
MOST CONSERVED PARTS OF THIS

5202
03:15:05,205 --> 03:15:07,708
BEAST ARE ACTUALLY ON THE INSIDE

5203
03:15:07,708 --> 03:15:10,043
CORE OF THE PROTEIN SO S2 IS

5204
03:15:10,043 --> 03:15:11,645
HIGHLY CONSERVED ACROSS THE

5205
03:15:11,645 --> 03:15:12,746
FAMILY SO IT'S BEEN THOUGHT IT

5206
03:15:12,746 --> 03:15:16,083
WOULD BE A POTENTIALLY GOOD

5207
03:15:16,083 --> 03:15:17,451
TARGET.

5208
03:15:17,451 --> 03:15:20,053
IT ALSO HAS IMMUNOGENS AT THE

5209
03:15:20,053 --> 03:15:21,655
TOP THERE, AND SOME I'M SHOWING

5210
03:15:21,655 --> 03:15:23,423
HERE IN RED AND ALSO IN YELLOW

5211
03:15:23,423 --> 03:15:26,893
AND HAVE YOU 1 IN THE STALK THAT

5212
03:15:26,893 --> 03:15:28,962
ARE THERE SO IT'S ALSO NEWT O

5213
03:15:28,962 --> 03:15:32,466
GENIC, 1 CHALLENGE HAS BEEN

5214
03:15:32,466 --> 03:15:33,333
STABILITY OF THIS.

5215
03:15:33,333 --> 03:15:37,738
SO YOU PRESS IT IF IT'S AN MRNA

5216
03:15:37,738 --> 03:15:39,473
CONSTRUCT, SO IT BASICALLY FALLS

5217
03:15:39,473 --> 03:15:41,775
APART SO IT HAS THE PREFUSION S2

5218
03:15:41,775 --> 03:15:43,610
WHERE YOU HAVE THE EPITOPE

5219
03:15:43,610 --> 03:15:46,413
SYSTEM UNSTABLE AND THIS IS THE

5220
03:15:46,413 --> 03:15:48,048
CASE EVEN FOR THESE REALLY SORT

5221
03:15:48,048 --> 03:15:51,418
OF HIGHLY DESIGNED AND OPTIMIZED

5222
03:15:51,418 --> 03:15:52,886
STABLE HEXAPRO MUTANTS OF S2,

5223
03:15:52,886 --> 03:15:54,921
THEY ALSO JUST SORT OF FLOP

5224
03:15:54,921 --> 03:15:56,757
APART, SO REALLY FURTHER

5225
03:15:56,757 --> 03:15:59,459
STABILIZATION IS REQUIRED IN

5226
03:15:59,459 --> 03:16:01,028
ORDER TO USE THIS AND AN ANTIGEN

5227
03:16:01,028 --> 03:16:02,663
AND SO THIS IS WHERE WE WANT TO

5228
03:16:02,663 --> 03:16:04,164
BRING IN MODEL CITIZEN LEAKULAR

5229
03:16:04,164 --> 03:16:08,602
SIMULATION AND NOT ONLY DID WE

5230
03:16:08,602 --> 03:16:10,804
JUST DO IN THIS CASE WE WANT TO

5231
03:16:10,804 --> 03:16:13,173
USE A SORT OF MORE ADVANCED OR

5232
03:16:13,173 --> 03:16:15,042
ENHANCED SAMPLING TECHNIQUE

5233
03:16:15,042 --> 03:16:16,243
CALLED WEIGHTED ONSEMBLE

5234
03:16:16,243 --> 03:16:17,544
MOLECULAR DYNAMICS WHICH IS A

5235
03:16:17,544 --> 03:16:19,746
WAY OF OVERCOMING 1 OF THE

5236
03:16:19,746 --> 03:16:21,848
CENTRAL LIMITATIONS OF THE

5237
03:16:21,848 --> 03:16:24,651
METHOD OF MD SO YOU KNOW 1 OF

5238
03:16:24,651 --> 03:16:26,286
THE ISSUES OR CHALLENGES WITH

5239
03:16:26,286 --> 03:16:28,889
THESE MANAGES IS THAT THEY ARE

5240
03:16:28,889 --> 03:16:30,190
PHYSICS BASED AND THEY'RE SLOW,

5241
03:16:30,190 --> 03:16:32,259
SO COMPUTING THOSE IS RATHER

5242
03:16:32,259 --> 03:16:33,660
SLOW AND COMPUTATIONALLY

5243
03:16:33,660 --> 03:16:33,927
INTENSIVE.

5244
03:16:33,927 --> 03:16:34,928
IDEALLY WE WOULD HAVE ENOUGH

5245
03:16:34,928 --> 03:16:37,931
DATA THAT WE COULD BUILD FASTER

5246
03:16:37,931 --> 03:16:38,732
SURROGATE MODELS WITH ML BUT

5247
03:16:38,732 --> 03:16:40,500
WE'RE NOT THERE YET AND SO WE'RE

5248
03:16:40,500 --> 03:16:41,768
STILL -- WE AND MANY OTHERS IN

5249
03:16:41,768 --> 03:16:43,670
THE FIELD ARE IN THE PROCESS OF

5250
03:16:43,670 --> 03:16:44,671
GENERATING HIGH QUALITY DATA

5251
03:16:44,671 --> 03:16:47,407
SETS THAT CAN BE UTILIZED LIKE A

5252
03:16:47,407 --> 03:16:49,242
MACHINE LEARNING AND AI

5253
03:16:49,242 --> 03:16:50,644
APPROACHES BUT WE'RE NOT THERE

5254
03:16:50,644 --> 03:16:51,178
QUITE YET.

5255
03:16:51,178 --> 03:16:54,448
SO IN ANY CASE, WE TURN TO

5256
03:16:54,448 --> 03:16:56,950
WEIGHTED ONSEMBLE MD WHICH IS AN

5257
03:16:56,950 --> 03:16:58,185
ENHANCED SAMPLING.

5258
03:16:58,185 --> 03:17:00,954
IT'S CALLED PAST SAMPLING,

5259
03:17:00,954 --> 03:17:02,823
APPROACH THAT BASICALLY HELPS TO

5260
03:17:02,823 --> 03:17:05,726
GET US OUT OF THESE BASINS, I

5261
03:17:05,726 --> 03:17:08,128
LOST MY MOUSE BUT THAT'S OKAY,

5262
03:17:08,128 --> 03:17:10,697
IT HELPS GET US OUT OF THESE

5263
03:17:10,697 --> 03:17:12,199
WELLS THAT WE'RE USUALLY STUCK

5264
03:17:12,199 --> 03:17:15,602
IN WITH DYNAMICS BY SORT OF

5265
03:17:15,602 --> 03:17:16,303
SPAWNING MANY TRAJECTORIES AND

5266
03:17:16,303 --> 03:17:18,271
FOLLOWING THE 1S THAT ARE SORT

5267
03:17:18,271 --> 03:17:20,107
OF SUCCESSFULLY PROGRESSING

5268
03:17:20,107 --> 03:17:21,875
ACROSS THE PREDEFINED REACTION

5269
03:17:21,875 --> 03:17:22,876
COORDINATE IN WHICH CASE WHICH

5270
03:17:22,876 --> 03:17:25,912
WAS SORT OF THE OPENING OF THE

5271
03:17:25,912 --> 03:17:29,316
APEX, SO, HERE IS THE S2 SUBUNIT

5272
03:17:29,316 --> 03:17:31,885
AND WHAT I'M SHOWING, WHAT I'M

5273
03:17:31,885 --> 03:17:33,887
ABOUT TO SHOW IT'S A TRIEMER,

5274
03:17:33,887 --> 03:17:38,992
THE SPIKE PROTEIN IS A TRIMMER,

5275
03:17:38,992 --> 03:17:43,897
IT'S 1 OF MANY OF THESE TBLIEK O

5276
03:17:43,897 --> 03:17:45,332
PROFUSION PROTEINS ARE TRI MERS,

5277
03:17:45,332 --> 03:17:47,067
SO WE USED THIS TO BASICALLY

5278
03:17:47,067 --> 03:17:49,035
EXPLORE SORT OF THE OPENING

5279
03:17:49,035 --> 03:17:49,269
PATHWAY.

5280
03:17:49,269 --> 03:17:50,704
SO OUR THOUGHT WAS IN OUR

5281
03:17:50,704 --> 03:17:53,273
STRATEGY WAS, IF WE COULD LEARN

5282
03:17:53,273 --> 03:17:57,110
SOMETHING ABOUT THE MECHANISM OF

5283
03:17:57,110 --> 03:17:58,912
OPENING, OF THIS POTENTIAL

5284
03:17:58,912 --> 03:17:59,980
IMMUNOGEN, COULD WE THEN USE

5285
03:17:59,980 --> 03:18:01,148
INFORMATION ABOUT THE MECHANISM

5286
03:18:01,148 --> 03:18:04,818
AGAINST IT IN ORDER TO BASICALLY

5287
03:18:04,818 --> 03:18:07,020
TOP THAT PROGRESSION FROM

5288
03:18:07,020 --> 03:18:07,554
HAPPENING.

5289
03:18:07,554 --> 03:18:09,122
AND SO, HERE'S JUST ANOTHER VIEW

5290
03:18:09,122 --> 03:18:10,791
OF THIS MOLECULE, SORT OF

5291
03:18:10,791 --> 03:18:11,691
PROGRESSING ALONG LAWEDDER DALE

5292
03:18:11,691 --> 03:18:13,994
WE DEFINE AS OUR REACTION

5293
03:18:13,994 --> 03:18:16,296
COORDINATE, SO WE TART UP HERE

5294
03:18:16,296 --> 03:18:18,899
IN THE UPPER LEFT QUADRANT, SORT

5295
03:18:18,899 --> 03:18:20,667
OF OUR CLOSED CONFIRMATION AND

5296
03:18:20,667 --> 03:18:22,068
WE'RE PROGRESSING TOWARDS THE

5297
03:18:22,068 --> 03:18:22,435
OPENING STATE.

5298
03:18:22,435 --> 03:18:24,571
AND YOU KNOW WHEN YOU RUN

5299
03:18:24,571 --> 03:18:26,640
SIMULATIONS YOU ACTUALLY GET

5300
03:18:26,640 --> 03:18:28,008
MANY ENSEMBLES OF PATHWAYS AND A

5301
03:18:28,008 --> 03:18:29,376
LOT OF WORK HAS TO BE DONE TO

5302
03:18:29,376 --> 03:18:31,578
TRY TO TEASE OUT, YOU KNOW WHICH

5303
03:18:31,578 --> 03:18:35,215
IS THE DOMINANT PATHWAY, BUT IN

5304
03:18:35,215 --> 03:18:36,883
ANY CASE, WE CAN VIEW THAT, AND

5305
03:18:36,883 --> 03:18:39,619
SO BY LOOKING AT THAT OPENING

5306
03:18:39,619 --> 03:18:45,325
PATHWAY, WHAT WE SAW WAS WHEN

5307
03:18:45,325 --> 03:18:46,960
WREE LOOKED CLOSELY AT THE

5308
03:18:46,960 --> 03:18:48,261
MECHANISM OF OPENING WE CAN SEE

5309
03:18:48,261 --> 03:18:50,764
THAT IT'S SORT OF UNZIP INDEED

5310
03:18:50,764 --> 03:18:51,631
AN ASIMETRICAL MANNER AND WHEN

5311
03:18:51,631 --> 03:18:53,300
WE DID A CLOSER ANALYSIS, WHAT

5312
03:18:53,300 --> 03:18:55,101
WE FOUND WAS THAT IF YOU LOOK AT

5313
03:18:55,101 --> 03:18:59,539
THE CENTRAL HELIXES, STRATIFIED

5314
03:18:59,539 --> 03:19:02,175
AS A LADDER AS I'M SHOWING HERE

5315
03:19:02,175 --> 03:19:03,610
WITH THE HELIXES IN PURPLE AND

5316
03:19:03,610 --> 03:19:04,811
THE RESIDUES IN YELLOW AND

5317
03:19:04,811 --> 03:19:06,146
ORANGE, WHAT YOU CAN SEE IS THAT

5318
03:19:06,146 --> 03:19:07,113
THERE ARE PARTICULAR POSITIONS

5319
03:19:07,113 --> 03:19:17,624
AND PARTICULAR THAT TOP 1 V991

5320
03:19:22,062 --> 03:19:28,401
AND THEY WERE QUICKLY FALLING

5321
03:19:28,401 --> 03:19:28,635
APART.

5322
03:19:28,635 --> 03:19:31,404
SO, WHAT WE DID THEN WAS WE

5323
03:19:31,404 --> 03:19:33,006
ACTUALLY LEARNED THE MECHANISM

5324
03:19:33,006 --> 03:19:34,975
AND TRUST TRIED PUTTING

5325
03:19:34,975 --> 03:19:36,076
TRYPTOPHAN MUTATIONS INTO THOSE

5326
03:19:36,076 --> 03:19:37,277
LOCATIONS IN BOTH THAT TOP RUNG

5327
03:19:37,277 --> 03:19:40,146
AS WELL AS THE FOURTH RUNG.

5328
03:19:40,146 --> 03:19:41,648
AND WHEN WE DID THAT, WHAT WE

5329
03:19:41,648 --> 03:19:51,124
FOUND WAS A HIGHLY STABILIZED

5330
03:19:51,124 --> 03:19:52,292
NETWORK OF INTERACTIONS BETWEEN

5331
03:19:52,292 --> 03:19:53,360
THE DIFFERENT -- BETWEEN NOT

5332
03:19:53,360 --> 03:19:55,962
ONLY EACH OF THE SORT OF CENTRAL

5333
03:19:55,962 --> 03:19:58,632
HELIXES BUT ALSO THEIR SORT OF

5334
03:19:58,632 --> 03:20:00,066
NETWORK OF INTERACTIONS, BETWEEN

5335
03:20:00,066 --> 03:20:02,802
THE RESDUES THAT ARE SORT OF

5336
03:20:02,802 --> 03:20:03,970
UNDERNEATH IT AS WELL AS GOING

5337
03:20:03,970 --> 03:20:05,038
DOWN TO THE FOURTH LEVEL AND

5338
03:20:05,038 --> 03:20:06,973
ABOVE AND EVEN AN OUTSIDE

5339
03:20:06,973 --> 03:20:15,749
NETWORK OF STABILIZED RESIDUES

5340
03:20:15,749 --> 03:20:15,982
FURTHER.

5341
03:20:15,982 --> 03:20:18,118
IT BECAME -- SO WE MADE THESE 2

5342
03:20:18,118 --> 03:20:24,324
SUGGESTIONS AND THEN SIMULATED

5343
03:20:24,324 --> 03:20:27,260
THOSE AND WHEN WE SAW IT STOPPED

5344
03:20:27,260 --> 03:20:30,931
THE OPENING, WELL IT DIDN'T STOP

5345
03:20:30,931 --> 03:20:33,066
THE OPENING BUT IT SLOWED THE

5346
03:20:33,066 --> 03:20:34,467
OPEN BEING RELATIVE TO WHAT WE

5347
03:20:34,467 --> 03:20:36,937
HAD BEEN WOSH OTHER WITH WE THEN

5348
03:20:36,937 --> 03:20:39,973
HANDED IT TO JASON AND HIS TEAM

5349
03:20:39,973 --> 03:20:41,975
AND THEY WERE ABLE TO EXPRESS

5350
03:20:41,975 --> 03:20:43,276
THE DOUBLE MUTANT, WHAT THEY SAW

5351
03:20:43,276 --> 03:20:45,011
WAS THAT IT WAS ACTUALLY -- I'M

5352
03:20:45,011 --> 03:20:48,682
SORRY I LOST MY POINTER, THEY

5353
03:20:48,682 --> 03:20:51,117
SAW THAT ACTUALLY IF YOU LOOK AT

5354
03:20:51,117 --> 03:20:52,052
DIFFERENTIAL SCANNING AT

5355
03:20:52,052 --> 03:20:53,486
TEMPERATURE PLOTS THESE 2

5356
03:20:53,486 --> 03:20:56,222
MUTATIONS CREATED A HIGHLY

5357
03:20:56,222 --> 03:20:56,690
TEMPERATURE STABILIZED

5358
03:20:56,690 --> 03:20:58,058
CONSTRUCT, IN FACT THE MOST

5359
03:20:58,058 --> 03:21:00,627
TABLE THAT WE KNOW OF MOW AND

5360
03:21:00,627 --> 03:21:02,062
THE OTHER COOL THING WAS BECAUSE

5361
03:21:02,062 --> 03:21:07,267
OF THE FLEXIBILITY OF THIS

5362
03:21:07,267 --> 03:21:08,134
CONSTRUCT INITIALLY THEY WERE

5363
03:21:08,134 --> 03:21:09,869
NEVER ABLE TO GET HIGH

5364
03:21:09,869 --> 03:21:10,770
RESOLUTION STRUCTURES OF JUST

5365
03:21:10,770 --> 03:21:13,206
THE S2 BECAUSE IT WAS JUST TOO

5366
03:21:13,206 --> 03:21:14,441
DYNAMIC HOWEVER WITH THESE 2

5367
03:21:14,441 --> 03:21:15,875
MUTATIONS THEY WERE ABLE TO GET

5368
03:21:15,875 --> 03:21:16,943
THE HIGH RESOLUTION STRUCTURE

5369
03:21:16,943 --> 03:21:23,883
WHICH IS REALLY COOL AND IT

5370
03:21:23,883 --> 03:21:24,651
VALENTINEDICATEED ALL THE

5371
03:21:24,651 --> 03:21:26,987
FINDINGS WE HAD ABOUT THE

5372
03:21:26,987 --> 03:21:30,757
MUTATIONS AND THE STRUCTURAL

5373
03:21:30,757 --> 03:21:32,058
INTEGRITY IN THE PREFUSION

5374
03:21:32,058 --> 03:21:34,260
STATE, WE THEN PASS THIS TO

5375
03:21:34,260 --> 03:21:36,262
[INDISCERNIBLE] AND HIS TEAM AND

5376
03:21:36,262 --> 03:21:38,898
THEY DID -- THEY IMMUNIZED MICE

5377
03:21:38,898 --> 03:21:42,769
AND DID THE SERA EXPERIMENTS AND

5378
03:21:42,769 --> 03:21:46,006
THEY FOUND THAT IT WAS JUST AS

5379
03:21:46,006 --> 03:21:48,675
GOOD AGAINST THESEY RECOMBIN ANT

5380
03:21:48,675 --> 03:21:50,410
VIRUSES AGAINST ORIGINAL WUHAN

5381
03:21:50,410 --> 03:21:53,380
STRAIN BUT HAD IMPROVED

5382
03:21:53,380 --> 03:21:54,314
PERFORMANCE VERSUS OMICRON, AND

5383
03:21:54,314 --> 03:21:57,951
ACTUALLY WITH THESE BINDING

5384
03:21:57,951 --> 03:21:58,685
AFFINITY INTERACTIONS THEY --

5385
03:21:58,685 --> 03:22:00,754
THAT ARE SHOWN ON THE RIGHT HAND

5386
03:22:00,754 --> 03:22:02,689
SIDE THAT WE ACTUALLY SAW THESE

5387
03:22:02,689 --> 03:22:05,025
EPITOPES WERE STILL BEING

5388
03:22:05,025 --> 03:22:10,864
RECOGNIZED AND BOUND TO.

5389
03:22:10,864 --> 03:22:13,033
SO THERE'S A LOT OF FOLKS TO

5390
03:22:13,033 --> 03:22:13,633
ACKNOWLEDGE THERE.

5391
03:22:13,633 --> 03:22:15,235
I STILL HAVE A FEW MINUTES.

5392
03:22:15,235 --> 03:22:16,369
I JUST WANT TO SAY I FEEL LIKE

5393
03:22:16,369 --> 03:22:20,040
THIS IS SORT OF A LITTLE BIT OF

5394
03:22:20,040 --> 03:22:21,508
A DIFFERENT TACT THAN WHAT WE'VE

5395
03:22:21,508 --> 03:22:23,576
HEARD, I WILL SAY 1 OF THE

5396
03:22:23,576 --> 03:22:25,645
ADVANTAGES OF RUNNING THIS

5397
03:22:25,645 --> 03:22:26,579
TECHNIQUE, IS THAT INSTEAD OF

5398
03:22:26,579 --> 03:22:28,048
RUNNING A TON OF LAB

5399
03:22:28,048 --> 03:22:29,049
EXPERIMENTS, I THINK IT'S A

5400
03:22:29,049 --> 03:22:30,116
DIFFERENT VIEW WHERE YOU CAN

5401
03:22:30,116 --> 03:22:32,118
TAKE A DEEPER DIVE INTO

5402
03:22:32,118 --> 03:22:33,386
UNDERSTANDING A TARGET AND THE

5403
03:22:33,386 --> 03:22:35,722
MECHANISM OF ACTION, DYNAMICS

5404
03:22:35,722 --> 03:22:39,225
AND AND THEN IN FACT BE HIGHLY

5405
03:22:39,225 --> 03:22:40,527
STRATEGIC PLAN TEEJIC ABOUT

5406
03:22:40,527 --> 03:22:41,361
WHICH ACTIONS YOU'RE MAKING SO

5407
03:22:41,361 --> 03:22:43,196
THIS WAS A LARGE EFFICIENCY,

5408
03:22:43,196 --> 03:22:44,597
THEY CREATED, THE DOUBLE MUTANT

5409
03:22:44,597 --> 03:22:46,032
AND THEN EACH OF THE SINGLE

5410
03:22:46,032 --> 03:22:48,868
MUTANTS AND THEN WE HAD THESE

5411
03:22:48,868 --> 03:22:50,303
RESULTS WHICH WERE SORT OF, YOU

5412
03:22:50,303 --> 03:22:51,438
KNOW WHERE WE GOT TO THE

5413
03:22:51,438 --> 03:22:53,373
MILESTONE THAT WE WERE SORT OF

5414
03:22:53,373 --> 03:22:53,807
LOOKING FOR.

5415
03:22:53,807 --> 03:22:55,708
THAT SAID OF COURSE, THERE ARE

5416
03:22:55,708 --> 03:22:56,709
AND AS DAVID ALSO MENTIONED

5417
03:22:56,709 --> 03:23:00,313
THERE'S A LOT OF CHALLENGES IN

5418
03:23:00,313 --> 03:23:04,984
THIS FIELD AND FOR THE FIELD IN

5419
03:23:04,984 --> 03:23:09,456
GENERAL FOR DRUG AND IMMUNOGEN

5420
03:23:09,456 --> 03:23:10,356
DESIGN AND TALK ABOUT THESE IN

5421
03:23:10,356 --> 03:23:11,958
OF THE LAST SESSION AS WE'VE

5422
03:23:11,958 --> 03:23:13,059
BEEN CUSSING I THINK A LITTLE

5423
03:23:13,059 --> 03:23:14,260
BIT ALONG THE WAY BUT CERTAINLY

5424
03:23:14,260 --> 03:23:17,330
YOU KNOW WITH THE PDP, WHICH

5425
03:23:17,330 --> 03:23:19,999
ENABLED ALPHA FOLD AND ALL OF

5426
03:23:19,999 --> 03:23:21,935
THE GREAT SORT OF FRUITS OF THAT

5427
03:23:21,935 --> 03:23:23,536
LABOR WHICH TOOK MANY YEARS AND

5428
03:23:23,536 --> 03:23:25,405
THE DISCIPLINE OF A WHOLE FIELD

5429
03:23:25,405 --> 03:23:28,708
OR REAL SORT OF COMMITMENT, TO

5430
03:23:28,708 --> 03:23:29,843
SHARING DATA AND BEING IN THE

5431
03:23:29,843 --> 03:23:32,078
FORMAT OF IT AND SO FORTH, YOU

5432
03:23:32,078 --> 03:23:34,881
KNOW WE NOW SEE HOW THAT NEEDS

5433
03:23:34,881 --> 03:23:36,249
TO REVERBERATE ACROSS MANY

5434
03:23:36,249 --> 03:23:38,685
FIELDS INCLUDING PROTEIN

5435
03:23:38,685 --> 03:23:40,086
DYNAMICS, INCLUDING LIGAND SMALL

5436
03:23:40,086 --> 03:23:42,922
MOLECULE STRUCTURE DESIGN, AND

5437
03:23:42,922 --> 03:23:45,125
SO, IN MY CLOSING THING, I SEE

5438
03:23:45,125 --> 03:23:46,659
PAT, YOU'RE READY TO GO, I HAVE

5439
03:23:46,659 --> 03:23:48,261
MAYBE 2 MINUES I UOF THE WHY

5440
03:23:48,261 --> 03:23:50,597
WANT TO TOUCH A COUPLE OF

5441
03:23:50,597 --> 03:23:52,732
CHALLENGES HERE JUST TO SORT OF

5442
03:23:52,732 --> 03:23:54,000
SEE IT INTO THE DISCUSSION THAT

5443
03:23:54,000 --> 03:23:55,368
WE WILL HAVE IN A MINUTE AND I

5444
03:23:55,368 --> 03:23:58,004
WILL FLIP IT TO YOU BUT YOU KNOW

5445
03:23:58,004 --> 03:24:00,773
AS 1 OF THE THINGS THAT WE ARE

5446
03:24:00,773 --> 03:24:01,608
PARTICULARLY INTERESTED IN,

5447
03:24:01,608 --> 03:24:03,543
AGAIN WHICH I GUESS MAYBE KIND

5448
03:24:03,543 --> 03:24:05,812
OF A LITTLE BIT UNIQUE BUT IT'S

5449
03:24:05,812 --> 03:24:07,347
SORT OF ENLARGING SORT OF THE

5450
03:24:07,347 --> 03:24:08,481
SCALE ASK COMPLEXITY OF SYSTEMS

5451
03:24:08,481 --> 03:24:10,416
THAT WERE ABLE TO SIMULATE AND

5452
03:24:10,416 --> 03:24:12,185
THERE'S, YOU KNOW WITH THIS --

5453
03:24:12,185 --> 03:24:14,454
WE HAVE SUCH AMAZING DATA SETS,

5454
03:24:14,454 --> 03:24:16,356
FOR SO MANY DIFFERENT TARGETS

5455
03:24:16,356 --> 03:24:18,992
AND COMING AT THEM FROM SO MANY

5456
03:24:18,992 --> 03:24:21,127
DIFFERENT CORRECTIONS NOW, YOU

5457
03:24:21,127 --> 03:24:22,529
KNOW REALLY BEING ABLE TO, YOU

5458
03:24:22,529 --> 03:24:24,931
KNOW I SEE A LOT OF CHALLENGES,

5459
03:24:24,931 --> 03:24:26,733
RELATED TO THAT AND ALSO WITH

5460
03:24:26,733 --> 03:24:28,268
DATA, AND I THINK DATA'S KIND OF

5461
03:24:28,268 --> 03:24:31,070
LIKE IN SOME WAYS A FUNNY THING,

5462
03:24:31,070 --> 03:24:32,572
BUT THEN IT'S LIKE, MAYBE IT'S

5463
03:24:32,572 --> 03:24:35,208
NOT THE RIGHT TYPE OF DATA AND

5464
03:24:35,208 --> 03:24:39,112
ESPECIALLY IN OUR FIELD, WE SIGH

5465
03:24:39,112 --> 03:24:42,081
-- SEE A LOT OF BIOLOGY

5466
03:24:42,081 --> 03:24:43,683
LABORATORIES GENERATING DATA BUT

5467
03:24:43,683 --> 03:24:44,817
IT'S NOT NECESSARILY UNIFIED OR

5468
03:24:44,817 --> 03:24:46,052
BEING SHARE INDEED A WAY THAT

5469
03:24:46,052 --> 03:24:48,621
CAN BE USED AND I KNOW I'M NOT

5470
03:24:48,621 --> 03:24:51,324
THE OHM PERSON USING AI IN ML

5471
03:24:51,324 --> 03:24:52,058
GOING FORWARD.

5472
03:24:52,058 --> 03:24:53,693
OF COURSE THERE'S HARDWARE

5473
03:24:53,693 --> 03:24:54,360
LIMITATIONS EVEN THOUGH THERE

5474
03:24:54,360 --> 03:24:56,963
HAVE BEEN GREAT ADVANCES IN GPU

5475
03:24:56,963 --> 03:24:58,965
AND THERE ARE LIMITATIONS FOR

5476
03:24:58,965 --> 03:24:59,933
THESE ALGORITHMS THEY THINK WILL

5477
03:24:59,933 --> 03:25:01,267
NEED TO BE ADDRESSED.

5478
03:25:01,267 --> 03:25:03,469
AND THEN THERE'S A LOT OF SOCIAL

5479
03:25:03,469 --> 03:25:04,604
ASSPECS AND I'M SURE WE GET TO

5480
03:25:04,604 --> 03:25:07,473
TOUCH ON THIS IN THE LAST LITTLE

5481
03:25:07,473 --> 03:25:09,242
PANEL DISCUSSION AND YOU KNOW

5482
03:25:09,242 --> 03:25:11,311
THERE'S DIFFERENT SOCIAL ASPECTS

5483
03:25:11,311 --> 03:25:12,145
DEPENDING ON ACADEMIA ON

5484
03:25:12,145 --> 03:25:15,114
INDUSTRY, BUT I THINK THE FIELD

5485
03:25:15,114 --> 03:25:17,116
AS A WHOLE DRUG DISCOVERY AS A

5486
03:25:17,116 --> 03:25:18,284
WHOLE, THAT WHOLE ECOSYSTEM WILL

5487
03:25:18,284 --> 03:25:23,923
DO BETTER BY SORT OF TRYING TO

5488
03:25:23,923 --> 03:25:26,159
ADDRESS SOME OF THESE, AND OF

5489
03:25:26,159 --> 03:25:27,560
COURSE, PAT WILL TALK A LOT

5490
03:25:27,560 --> 03:25:30,263
ABOUT, MAYBE WE GET TO HEAR 1 OF

5491
03:25:30,263 --> 03:25:31,698
PAT'S RANTS 1 OF MY PERSONAL

5492
03:25:31,698 --> 03:25:34,467
FAVORITE THINGS TO DO, AND I

5493
03:25:34,467 --> 03:25:36,736
LOST MY CURSOR, SO I WANT TO

5494
03:25:36,736 --> 03:25:38,204
PASS IT TO YOU BUT YOU KNOW

5495
03:25:38,204 --> 03:25:39,772
WHAT, I WILL TRY TO STOP SHARING

5496
03:25:39,772 --> 03:25:40,206
MY SCREEN.

5497
03:25:40,206 --> 03:25:44,777
CAN YOU TRY TO SHARE YOUR

5498
03:25:44,777 --> 03:25:45,011
SCREEN?

5499
03:25:45,011 --> 03:25:45,712
OH THERE IT IS.

5500
03:25:45,712 --> 03:25:46,646
I CAN STOP SHARE.

5501
03:25:46,646 --> 03:25:51,017
WITH THAT I WILL PASS THE BATON

5502
03:25:51,017 --> 03:25:53,853
TO PAT WALTERS WHO IS AT RELAY

5503
03:25:53,853 --> 03:25:55,588
THERAPEUTICS AND HAS BEEN KIND

5504
03:25:55,588 --> 03:25:58,491
OF AN ORACLE IN THE FIELD FOR

5505
03:25:58,491 --> 03:26:00,526
BENCHMARKING OF DATA AND KEEPING

5506
03:26:00,526 --> 03:26:01,160
EVERYBODY HONEST.

5507
03:26:01,160 --> 03:26:03,663
SO PAT, I LOOK FORWARD TO YOUR

5508
03:26:03,663 --> 03:26:11,504
YOUR TALK, TO YOUR RANT. 

5509
03:26:11,504 --> 03:26:21,748
TAKE IT AWAY.

5510
03:26:26,819 --> 03:26:27,687
WE SEE YOUR SCREEN.

5511
03:26:27,687 --> 03:26:29,722
DID YOU LOSE YOUR MOUSE, TOO?

5512
03:26:29,722 --> 03:26:31,391
>> I DID BUT I FOUND IT.

5513
03:26:31,391 --> 03:26:38,464
>> YOU'RE GOOD, WE HEAR
YOU. 

5514
03:26:38,464 --> 03:26:38,865
THANK YOU EMPLOY.

5515
03:26:38,865 --> 03:26:41,200
>> ALL RIGHT, THANK YOU
ROMMIE,

5516
03:26:41,200 --> 03:26:43,536
AND THANK YOU EVERYBODY, YOU

5517
03:26:43,536 --> 03:26:44,671
WON'T BE DISAPPOINTED I'M GOING

5518
03:26:44,671 --> 03:26:48,908
ON A RANT. 

5519
03:26:48,908 --> 03:26:50,710
THIS FIELD IS MOVING AI IS DRUG

5520
03:26:50,710 --> 03:26:52,011
DORPHY IS MOVING AT A HUNDRED

5521
03:26:52,011 --> 03:26:53,746
MILES AN HOUR AND THERE ARE TONS

5522
03:26:53,746 --> 03:26:56,516
OF PAPERS BEING PUBLISHED, IT'S

5523
03:26:56,516 --> 03:27:00,019
A REALLY EXCITING FIELD TO WORK

5524
03:27:00,019 --> 03:27:02,055
IN, I'M REX SIGHTED PEOPLE ARE

5525
03:27:02,055 --> 03:27:03,323
RELEASING SOURCE CODE WITH THEIR

5526
03:27:03,323 --> 03:27:04,857
PAPERS AND THINGS YOU CAN TRY

5527
03:27:04,857 --> 03:27:06,392
OUT BUT I THINK THERE ARE SOME

5528
03:27:06,392 --> 03:27:07,794
THINGS WE NEED TO CONSIDER MORE

5529
03:27:07,794 --> 03:27:10,330
CAREFULLY SO I WANT TO START OUT

5530
03:27:10,330 --> 03:27:12,398
WITH JUST A FEW FRAMING

5531
03:27:12,398 --> 03:27:12,699
ASSUMPTIONS.

5532
03:27:12,699 --> 03:27:14,834
THERE ARE A LOT OF PAPERS BEING

5533
03:27:14,834 --> 03:27:15,902
PUBLISHED THESE DAYS THAT

5534
03:27:15,902 --> 03:27:17,270
PRESENT NEW MACHINE LEARNING

5535
03:27:17,270 --> 03:27:21,741
APPROACHES FOR DRUG DISCOVERY.

5536
03:27:21,741 --> 03:27:24,310
MOST OF THESE PAPERS PRESENT

5537
03:27:24,310 --> 03:27:24,944
BENCHMARK COMPARISONS SO

5538
03:27:24,944 --> 03:27:26,212
EVERYBODY WANTS TO SHOW THAT

5539
03:27:26,212 --> 03:27:28,948
THEIR METHOD IS BETTER THAN THE

5540
03:27:28,948 --> 03:27:29,482
CURRENT STATE-OF-THE-ART,

5541
03:27:29,482 --> 03:27:32,518
UNFORTUNATELY, MOST OF THESE

5542
03:27:32,518 --> 03:27:33,519
BENCHMARK COMPARISONS ARE  

5543
03:27:33,519 --> 03:27:37,857
PERFORMED ON DATA SETS THAT ARE

5544
03:27:37,857 --> 03:27:38,925
REALLY BAD.

5545
03:27:38,925 --> 03:27:41,461
AND IN ESSENCE A LOT OF PEOPLE

5546
03:27:41,461 --> 03:27:42,695
ARE PUBLISHING PAPERS, FEELING

5547
03:27:42,695 --> 03:27:44,964
THEY CAN LEARN RANDOM LABELS.

5548
03:27:44,964 --> 03:27:46,232
AND HOPEFULLY I WILL CONVINCE

5549
03:27:46,232 --> 03:27:46,699
YOU OF THAT.

5550
03:27:46,699 --> 03:27:50,269
THE SECOND THING I WANT TO

5551
03:27:50,269 --> 03:27:51,671
CONVINCE YOU OF IS THAT WE NEED

5552
03:27:51,671 --> 03:27:53,339
TO BE MORE RIGOROUS ABOUT

5553
03:27:53,339 --> 03:27:56,042
STATISTICS AND THE WAY IN WHICH

5554
03:27:56,042 --> 03:27:57,577
WE COMPARE MODELS AND THEN I'LL

5555
03:27:57,577 --> 03:27:59,545
WRAP UP BY TALKING ABOUT A

5556
03:27:59,545 --> 03:28:04,417
COUPLE OF EFFORTS THAT WE HAVE

5557
03:28:04,417 --> 03:28:05,785
GOING WHICH WILL HOPEFULLY FIX

5558
03:28:05,785 --> 03:28:06,786
SOME OF THESE SITUATIONS.

5559
03:28:06,786 --> 03:28:08,688
SO WHAT WE WANT TO DO IS GET TO

5560
03:28:08,688 --> 03:28:10,590
THE POINT WHERE WE CAN COME UP

5561
03:28:10,590 --> 03:28:12,191
WITH A SET OF GUIDELINES.

5562
03:28:12,191 --> 03:28:13,726
I THINK WE NEED GUIDELINES TO

5563
03:28:13,726 --> 03:28:15,828
HELP AUTHORS TO PUT TOGETHER THE

5564
03:28:15,828 --> 03:28:16,529
BEST PAPER.

5565
03:28:16,529 --> 03:28:18,831
I THINK EDITORS SOMETIMES DON'T

5566
03:28:18,831 --> 03:28:22,235
HAVE THE BACKGROUND IN MACHINE

5567
03:28:22,235 --> 03:28:23,369
LEARNING OR STATISTICS TO BE

5568
03:28:23,369 --> 03:28:25,838
ABLE TO ASSESS THE WORK, NOR DO

5569
03:28:25,838 --> 03:28:27,607
THE REVIEWERS, SO I THINK WE

5570
03:28:27,607 --> 03:28:29,275
NEED GUIDELINES TO HELP

5571
03:28:29,275 --> 03:28:31,344
EVERYBODY AALONG AND PUT

5572
03:28:31,344 --> 03:28:32,412
EVERYBODY ON THE SAME PAGE, SO I

5573
03:28:32,412 --> 03:28:35,381
WANT TO TALK ABOUT 2 THINGS

5574
03:28:35,381 --> 03:28:35,681
TODAY.

5575
03:28:35,681 --> 03:28:37,383
ONE IS DATA SETS.

5576
03:28:37,383 --> 03:28:39,318
AND SOME OF THE PROBLEMS WITH

5577
03:28:39,318 --> 03:28:40,987
THE EXISTING DATA SETS AND THEN

5578
03:28:40,987 --> 03:28:43,689
I WANT TO TALK JUST A LITTLE BIT

5579
03:28:43,689 --> 03:28:44,724
ABOUT STATISTICS AND COMPARING

5580
03:28:44,724 --> 03:28:47,193
METHODS AND WHAT I THINK ARE THE

5581
03:28:47,193 --> 03:28:50,496
MORE APPROPRIATE WAYS TO DO

5582
03:28:50,496 --> 03:28:50,897
THAT.

5583
03:28:50,897 --> 03:28:52,632
AND TO START OUT TALKING ABOUT

5584
03:28:52,632 --> 03:28:54,434
DATA SETS, THE DATA SET THAT'S

5585
03:28:54,434 --> 03:28:57,303
USED MORE OFTEN THAN NOT WHEN

5586
03:28:57,303 --> 03:28:59,672
PEOPLE ARE PUBLISHING PAPERS ON

5587
03:28:59,672 --> 03:29:02,341
MACHINE LEARNING IS A DATA SET

5588
03:29:02,341 --> 03:29:04,177
CALLED MOLECULE NET, SO THIS

5589
03:29:04,177 --> 03:29:07,079
DATA SET WAS ORIGINALLY

5590
03:29:07,079 --> 03:29:10,516
PUBLISHED IN 2018 BY THE

5591
03:29:10,516 --> 03:29:12,151
[INDISCERNIBLE] GROUP AT

5592
03:29:12,151 --> 03:29:12,718
STANFORD.

5593
03:29:12,718 --> 03:29:14,020
IT CONSISTS OF SEVERAL DIFFERENT

5594
03:29:14,020 --> 03:29:15,688
CATEGORIES OF DATA SETS, THERE'S

5595
03:29:15,688 --> 03:29:19,025
A SET OF QUANTUM MECHANICS

5596
03:29:19,025 --> 03:29:20,626
PARAMETERS THAT WERE CALCULATED

5597
03:29:20,626 --> 03:29:25,998
BASED ON SMALL MOLECULES, THERE

5598
03:29:25,998 --> 03:29:29,168
ARE PHYSICAL PROPERTIES,

5599
03:29:29,168 --> 03:29:31,270
SOLBILITY, BINDING DATA FOR A

5600
03:29:31,270 --> 03:29:33,406
RANGE OF PROTEINS AND THEN

5601
03:29:33,406 --> 03:29:34,240
THERE'S PHYSIOLOGICALLY BASED

5602
03:29:34,240 --> 03:29:40,580
DATA AND THESE ARE REASONABLY

5603
03:29:40,580 --> 03:29:42,181
SIZED DATA SETS THEY RANGE FROM

5604
03:29:42,181 --> 03:29:44,450
HUNDREDS OF THOUSANDS DOWN TO A

5605
03:29:44,450 --> 03:29:44,750
FEW HUNDRED.

5606
03:29:44,750 --> 03:29:47,019
SO THIS IS GOOD, THERE ARE DATA

5607
03:29:47,019 --> 03:29:49,522
SETS IN THERE FOR CLASSIFICATION

5608
03:29:49,522 --> 03:29:52,425
AS WELL AS REGRESSION

5609
03:29:52,425 --> 03:29:53,259
UNFORTUNATELY, AS I HOPE TO

5610
03:29:53,259 --> 03:29:55,161
CONVINCE YOU IN THE NEXT FEW

5611
03:29:55,161 --> 03:29:56,762
MINUTES, THERE ARE SO MANY

5612
03:29:56,762 --> 03:29:58,698
PROBLEMS WITH THIS DATA SET AND

5613
03:29:58,698 --> 03:30:02,768
IT REALLY SHOULD NEVER BE USED

5614
03:30:02,768 --> 03:30:03,135
AGAIN.

5615
03:30:03,135 --> 03:30:04,337
SO LET ME JUST START OUT BY

5616
03:30:04,337 --> 03:30:06,706
POINTING OUT SOME OF THE REALLY

5617
03:30:06,706 --> 03:30:08,774
SIMPLE TECHNICAL DIFFICULTIES IN

5618
03:30:08,774 --> 03:30:09,175
THIS DATA SET.

5619
03:30:09,175 --> 03:30:10,977
THERE ARE A LOT OF MOLECULES

5620
03:30:10,977 --> 03:30:13,479
WHERE THE CHEMICAL STRUCTURES

5621
03:30:13,479 --> 03:30:17,316
ARE JUST INCORRECT, SO THEY HAVE

5622
03:30:17,316 --> 03:30:18,618
THINGS THAT VIOLATE RULES OF

5623
03:30:18,618 --> 03:30:21,287
VALENCE, SO YOU'LL HAVE

5624
03:30:21,287 --> 03:30:22,455
TETRAVALENT NITROGENS THAT DON'T

5625
03:30:22,455 --> 03:30:24,090
HAVE CHARGE, THERE'S A BUNCH OF

5626
03:30:24,090 --> 03:30:25,324
OTHER SIMILAR PROBLEMS AND YES,

5627
03:30:25,324 --> 03:30:28,027
CAN YOU GO THROUGH AND DO HACKS

5628
03:30:28,027 --> 03:30:29,929
TO KIND OF FIX ALL THIS BUT IT

5629
03:30:29,929 --> 03:30:34,066
WE'RE GOING TO USE THESE DATA

5630
03:30:34,066 --> 03:30:35,468
SETS AS STANDARD BENCH MASHS AT

5631
03:30:35,468 --> 03:30:37,803
THE VERY LEAST THEY SHOULD BE

5632
03:30:37,803 --> 03:30:38,337
CORRECT.

5633
03:30:38,337 --> 03:30:40,139
IN ADDITION TO BEING CORRECT,

5634
03:30:40,139 --> 03:30:42,275
THEY NEED TO BE CONSISTENT, SO

5635
03:30:42,275 --> 03:30:46,779
YOU KNOW HERE'S JUST 2 EXAMPLES

5636
03:30:46,779 --> 03:30:48,748
OF ANTIBACTERIALS THAT ARE IN

5637
03:30:48,748 --> 03:30:52,285
THE MOLECULE NET DATA SET AND

5638
03:30:52,285 --> 03:30:59,659
YOU'LL SEE THAT ON THE LEFT-HAND

5639
03:30:59,659 --> 03:31:03,996
SIDE WE HAVE A RIGHT SIDE, AND

5640
03:31:03,996 --> 03:31:06,799
IT ISN'T ON THE MOLECULE, THIS

5641
03:31:06,799 --> 03:31:10,169
HAS NOTHING TO DO WITH PKA, THIS

5642
03:31:10,169 --> 03:31:12,104
HAS TO DO WITH HOW THEY DRAW THE

5643
03:31:12,104 --> 03:31:15,942
MOLECULE IN THE DATA SET, IF

5644
03:31:15,942 --> 03:31:18,578
WE'RE EVALUATING LEARNING MODELS

5645
03:31:18,578 --> 03:31:19,879
NOT SOME STANDARDIZATION

5646
03:31:19,879 --> 03:31:22,982
PROTOCOL THAT'S PUTTING THINGS

5647
03:31:22,982 --> 03:31:25,051
INTO A CONSISTENT PROTOMERRIC

5648
03:31:25,051 --> 03:31:25,284
FORM.

5649
03:31:25,284 --> 03:31:28,888
THERE'S A LOT OF CASES IN THE

5650
03:31:28,888 --> 03:31:31,857
MOLECULE DATA SET WHERE STEREO

5651
03:31:31,857 --> 03:31:32,491
CHEMISTRY IS MISSING.

5652
03:31:32,491 --> 03:31:35,528
SO HERE YOU CAN SEE, 3 TERIO

5653
03:31:35,528 --> 03:31:40,733
ISOMERS OF THE SAME MOLECULE THE

5654
03:31:40,733 --> 03:31:42,001
FIRST 2 OF THE OF THE CHEMIST

5655
03:31:42,001 --> 03:31:47,640
RADIOY IS FULLY SPECIFIED ON THE

5656
03:31:47,640 --> 03:31:49,275
FAR RIGHT, THE TERIO CHEMISTRY

5657
03:31:49,275 --> 03:31:51,143
IS MISSING FOR 1 OF THOSE STEREO

5658
03:31:51,143 --> 03:31:51,377
CENTERS.

5659
03:31:51,377 --> 03:31:53,045
WHAT DID THAT MEAN IN DOES THAT

5660
03:31:53,045 --> 03:31:56,082
MEAN THAT CENTER IS RECEIVING OR

5661
03:31:56,082 --> 03:31:57,984
SOMEBODY JUST FORGOT TO PUT THAT

5662
03:31:57,984 --> 03:31:58,684
STEREO CENTER IN.

5663
03:31:58,684 --> 03:32:07,393
SO AGAIN, DATAET ISS NEED TO BE

5664
03:32:07,393 --> 03:32:09,061
TECHNICALLY CORRECT.

5665
03:32:09,061 --> 03:32:10,830
IT'S ALSO IMPORTANT TO HAVE DATA

5666
03:32:10,830 --> 03:32:12,632
SETS WHERE THE DYNAMIC RAINCHL

5667
03:32:12,632 --> 03:32:16,736
IN THE DATA IS APPROPRIATE TO

5668
03:32:16,736 --> 03:32:18,337
WHAT'S TYPICALLY ENCOUNTERED IN

5669
03:32:18,337 --> 03:32:19,071
DRUG DISCOVERY.

5670
03:32:19,071 --> 03:32:20,406
SO IN THIS PLOT, ON THE LEFT

5671
03:32:20,406 --> 03:32:23,209
SIDE OF THE SLIDE, I'M SHOWING

5672
03:32:23,209 --> 03:32:26,178
ANOTHER DATA SET ISSUES THE

5673
03:32:26,178 --> 03:32:29,348
SOLUBILITY DATA SET, WITH THE

5674
03:32:29,348 --> 03:32:29,915
MEASURED AQUESTIONNAIRESUS

5675
03:32:29,915 --> 03:32:30,883
SOLUBLES AND THAT'S IN THE

5676
03:32:30,883 --> 03:32:32,184
MOLECULE AND THAT DATA SET, 1

5677
03:32:32,184 --> 03:32:35,354
THING TO NOTICE WHEN YOU LOOKOT

5678
03:32:35,354 --> 03:32:37,390
X-AXIS WE HAVE THE EXPERIMENTAL

5679
03:32:37,390 --> 03:32:38,724
SOLUBILITY, ON THE Y-AXIS WE

5680
03:32:38,724 --> 03:32:39,759
HAVE A PREDICTION, THE THING TO

5681
03:32:39,759 --> 03:32:43,896
NOTICE IS THAT THE DAT SET SPANS

5682
03:32:43,896 --> 03:32:45,731
14 LOGS SO WHEN YOU HAVE A DATA

5683
03:32:45,731 --> 03:32:47,700
SET WITH A RIDICULOUS DYNAMIC

5684
03:32:47,700 --> 03:32:48,934
RAINCHL LIKE THAT, IT'S VERY

5685
03:32:48,934 --> 03:32:51,804
EASY TO DO REGRESSION.

5686
03:32:51,804 --> 03:32:52,772
BUT, CORRELATION THAT YOU GET

5687
03:32:52,772 --> 03:32:57,643
FROM THAT DATA SET IS NOT GOING

5688
03:32:57,643 --> 03:32:59,211
TO REPRESENT WHAT YOU SEE IN A

5689
03:32:59,211 --> 03:32:59,912
TYPICAL DATA SET.

5690
03:32:59,912 --> 03:33:01,414
SO HERE WHAT WE HAVE ON THE PLOT

5691
03:33:01,414 --> 03:33:04,350
ON THE RIGHT IS THE MOLECULE NET

5692
03:33:04,350 --> 03:33:06,619
SOLUBILITY ON THE DATA SET IN

5693
03:33:06,619 --> 03:33:09,021
BLUE, IN ORANGE I HAVE A MORE

5694
03:33:09,021 --> 03:33:10,022
REPRESENTATIVE GATTA DETAILS

5695
03:33:10,022 --> 03:33:13,259
THAT WAS TAKEN FROM A

5696
03:33:13,259 --> 03:33:14,360
PHARMACEUTICAL COMPANY AND YOU

5697
03:33:14,360 --> 03:33:16,562
CAN SEE THAT THAT DATA SET ONLY

5698
03:33:16,562 --> 03:33:17,897
SPANS 3 LOGS, SO WE REALLY NEED

5699
03:33:17,897 --> 03:33:20,433
TO MAKE SURE THAT THE DYNAMIC

5700
03:33:20,433 --> 03:33:23,736
RANGE IN THE DATA THAT WE'RE

5701
03:33:23,736 --> 03:33:25,337
USING TO BENCHMARK REPRESENTS

5702
03:33:25,337 --> 03:33:27,940
WHAT WE'RE GOING TO SEE IN

5703
03:33:27,940 --> 03:33:28,207
PRACTICE.

5704
03:33:28,207 --> 03:33:31,010
THERE ARE ALSO JUST IN

5705
03:33:31,010 --> 03:33:32,878
RIDICULOUS DATA CURATION ERRORS

5706
03:33:32,878 --> 03:33:34,947
IN THERE, SO ANOTHER DATA SET,

5707
03:33:34,947 --> 03:33:39,185
IN MOLECULE NET IS A DATA SET OF

5708
03:33:39,185 --> 03:33:40,286
BLOOD BRAIN BARRIER PENETRATION

5709
03:33:40,286 --> 03:33:43,255
DATA SO THEY HAVE A SET OF

5710
03:33:43,255 --> 03:33:43,823
CHEMICAL STRUCTURES, THEY'RE

5711
03:33:43,823 --> 03:33:45,091
LABELED 1, IF THE MOLECULE HAS

5712
03:33:45,091 --> 03:33:47,393
BEEN REPORTED TO BE BRAIN

5713
03:33:47,393 --> 03:33:49,662
PENETRANT, 0 IF IT WAS REPORTED

5714
03:33:49,662 --> 03:33:51,897
TO BE NOT BRAIN PENETRANT.

5715
03:33:51,897 --> 03:33:54,366
THAT SEEMS FINE, THIS DATA SET

5716
03:33:54,366 --> 03:33:56,402
IS PROBABLY BEEN PUBLISHED A

5717
03:33:56,402 --> 03:33:56,936
THOUSAND TIMES, LITERALLY.

5718
03:33:56,936 --> 03:34:00,806
IF YOU LOOK AT THIS DATA SET ARE

5719
03:34:00,806 --> 03:34:02,074
THERE ARE 11 MOLECULES THAT ARE

5720
03:34:02,074 --> 03:34:04,343
IN THE DATA SET TWICE.

5721
03:34:04,343 --> 03:34:06,846
ONCE AS BRAIN PENETRANT AND ONCE

5722
03:34:06,846 --> 03:34:08,814
AS NOT BRAIN PENETRANT.

5723
03:34:08,814 --> 03:34:09,949
THIS IS SIMPLY BECAUSE WHEN

5724
03:34:09,949 --> 03:34:11,917
SOMEBODY PUT THIS DATA SET

5725
03:34:11,917 --> 03:34:13,519
TOGETHER, THEY DIDN'T

5726
03:34:13,519 --> 03:34:14,987
STANDARDIZE THE SMILE STRINGS

5727
03:34:14,987 --> 03:34:17,389
AND DIDN'T GO THROUGH AND

5728
03:34:17,389 --> 03:34:18,791
CAREFULLY CURATE THE DATA AND

5729
03:34:18,791 --> 03:34:27,433
LOOK FOR ERRORS LIKE THIS.

5730
03:34:27,433 --> 03:34:29,401
SO WHEN YOU SEE SOMEBODY

5731
03:34:29,401 --> 03:34:31,070
PREDICTING A GOOD PERFORMANCE ON

5732
03:34:31,070 --> 03:34:32,938
A DATA SET LIKE THIS, YOU REALLY

5733
03:34:32,938 --> 03:34:34,240
ARE TO QUESTION IT AND THEN

5734
03:34:34,240 --> 03:34:36,575
THERE ARE OTHER DATA SETS IN

5735
03:34:36,575 --> 03:34:39,011
THERE WHERE PEOPLE ARE JUST

5736
03:34:39,011 --> 03:34:40,579
MISUSING THEM AND PREDICTING

5737
03:34:40,579 --> 03:34:41,247
THINGS THAT THEY ACTUALLY

5738
03:34:41,247 --> 03:34:43,582
SHOULDN'T BE ABLE TO PREDICT

5739
03:34:43,582 --> 03:34:45,885
FROM A SMALL STRING OR A 1

5740
03:34:45,885 --> 03:34:46,685
DIMENSIONAL CHEMICAL STRUCTURE.

5741
03:34:46,685 --> 03:34:51,624
ONE OF THE PRIME EXAMPLES OF

5742
03:34:51,624 --> 03:34:55,561
THIS IS THESE QUANTUM MECHANICS

5743
03:34:55,561 --> 03:34:56,562
DATA SETS CALLED QM27, 8 AND 9.

5744
03:34:56,562 --> 03:35:00,199
SO THIS A SET OF CALCULATED

5745
03:35:00,199 --> 03:35:01,901
CHEMICAL PROPERTIES FOR

5746
03:35:01,901 --> 03:35:05,237
MOLECULES WITH 7, 8 AND 9 ATOMS,

5747
03:35:05,237 --> 03:35:08,440
THESE PROPERTIES ARE

5748
03:35:08,440 --> 03:35:09,308
COMFIRMATIONAL --

5749
03:35:09,308 --> 03:35:09,875
CONFIRMATIONALLY DEPENDENT.

5750
03:35:09,875 --> 03:35:12,211
ONE SHOULD NOT BE ABLE TO

5751
03:35:12,211 --> 03:35:14,914
CALCULATE THESE THINGS FROM A

5752
03:35:14,914 --> 03:35:15,581
SMILE STRING.

5753
03:35:15,581 --> 03:35:18,818
SO PEOPLE ARE PUBLISHING PAPERS

5754
03:35:18,818 --> 03:35:21,453
WHERE THEY'RE MAKING PREDICTIONS

5755
03:35:21,453 --> 03:35:22,621
THAT ARE COMPLETELY

5756
03:35:22,621 --> 03:35:23,055
INAPPROPRIATE.

5757
03:35:23,055 --> 03:35:25,491
PERHAPS THE WORST CASE THOUGH IS

5758
03:35:25,491 --> 03:35:27,159
THIS DATA SET INSIDE MOLECULE

5759
03:35:27,159 --> 03:35:30,663
NET WHICH IS CALLED CIDER AND

5760
03:35:30,663 --> 03:35:36,101
IT'S TAKEN FROM A DATABASE OF

5761
03:35:36,101 --> 03:35:36,635
SIDE EFFECTS.

5762
03:35:36,635 --> 03:35:38,337
AND AGAIN THESE ARE THE SORTS OF

5763
03:35:38,337 --> 03:35:40,372
THINGS THAT 1 SIMPLY SHOULD NOT

5764
03:35:40,372 --> 03:35:42,141
BE ABLE TO CALCULATE FROM A

5765
03:35:42,141 --> 03:35:44,743
CHEMICAL STRUCTURE. 

5766
03:35:44,743 --> 03:35:45,711
SO I'VE HIGHLIGHTED SOME OF THE

5767
03:35:45,711 --> 03:35:49,114
EPPED POINTS THAT ARE IN THIS

5768
03:35:49,114 --> 03:35:49,815
DATA SET.

5769
03:35:49,815 --> 03:35:50,616
SOCIAL CIRCUMSTANCES, HOW AM I

5770
03:35:50,616 --> 03:35:54,153
SUPPOSED TO BE ABLE TO CALCULATE

5771
03:35:54,153 --> 03:35:55,454
SOCIAL CIRCUMSTANCES FROM A

5772
03:35:55,454 --> 03:35:56,589
CHEMICAL STRUCTURE.

5773
03:35:56,589 --> 03:35:57,756
SURGICAL AND MEDICAL PROCEDURES,

5774
03:35:57,756 --> 03:36:00,059
I MEAN THIS IS JUST NONSENSICAL,

5775
03:36:00,059 --> 03:36:02,995
IT MAKES NO SENSE, I'M NOT SURE

5776
03:36:02,995 --> 03:36:05,130
WHY PEOPLE ARE PUBLISHING THIS,

5777
03:36:05,130 --> 03:36:08,467
PEOPLE SEEM TO JUST BE PLIENDLY

5778
03:36:08,467 --> 03:36:10,035
MAKING PREDICTIONS WITHOUT

5779
03:36:10,035 --> 03:36:12,838
ACTUALLY LOOKING AT WHAT'S

5780
03:36:12,838 --> 03:36:15,107
INSIDE THESE DATA SETS.

5781
03:36:15,107 --> 03:36:16,041
YOU KNOW, FINALLY AND IT WAS

5782
03:36:16,041 --> 03:36:17,910
FUNNY BECAUSE IT WAS A PAPER

5783
03:36:17,910 --> 03:36:19,945
THAT CAME OUT IN NATURE, LAST

5784
03:36:19,945 --> 03:36:20,779
WEEK, WHERE THEY WERE TALKING

5785
03:36:20,779 --> 03:36:22,615
ABOUT THE FACT THAT WE NEED MORE

5786
03:36:22,615 --> 03:36:27,653
DATA FOR MACHINE LEARNING BEFORE

5787
03:36:27,653 --> 03:36:29,421
DRUG DISCOVERY LEARNING AND 1

5788
03:36:29,421 --> 03:36:32,191
EXAMPLE THEY CITED THEY WERE SO

5789
03:36:32,191 --> 03:36:35,127
SHOWING HOW YOU COULD BUILD A

5790
03:36:35,127 --> 03:36:36,829
MACHINE LEARNING MODEL FROM THIS

5791
03:36:36,829 --> 03:36:38,898
PARTICULAR HIV DATA SET, FROM

5792
03:36:38,898 --> 03:36:39,331
THE MOLECULE AGAIN.

5793
03:36:39,331 --> 03:36:40,566
IF YOU LOOK THROUGH THIS DATA

5794
03:36:40,566 --> 03:36:42,868
SET, I WOULD ESTIMATE THAT AT

5795
03:36:42,868 --> 03:36:45,471
LEAST 80% OF THE MOLECULES THAT

5796
03:36:45,471 --> 03:36:47,439
ARE SHOWN AS BEING ACTIVE ARE

5797
03:36:47,439 --> 03:36:49,041
THINGS LIKE THE 1S THAT ARE

5798
03:36:49,041 --> 03:36:50,276
SHOWN IN THIS SLIDE.

5799
03:36:50,276 --> 03:36:56,382
THESE ARE ALL ASSAY ARTIFACTS.

5800
03:36:56,382 --> 03:36:59,018
THIS IS NOT POSITIVE IN AN

5801
03:36:59,018 --> 03:37:01,120
ASSAY, SO WE NEED TO BE MORE

5802
03:37:01,120 --> 03:37:01,987
CAREFUL ABOUT CURATING THESE

5803
03:37:01,987 --> 03:37:04,657
DATA SETS AND MAKING SURE WE'RE

5804
03:37:04,657 --> 03:37:06,225
PREDICTING SOMETHING THAT'S

5805
03:37:06,225 --> 03:37:06,525
MEANINGFUL.

5806
03:37:06,525 --> 03:37:07,393
THERE ARE GOOD DATA SETS OUT

5807
03:37:07,393 --> 03:37:07,826
THERE.

5808
03:37:07,826 --> 03:37:11,897
THIS IS A DATA SET THAT WAS

5809
03:37:11,897 --> 03:37:14,566
PUBLISHED LAST YEAR BY A GROUP

5810
03:37:14,566 --> 03:37:22,374
FROM BIO BIOGEN, IT WAS ALL

5811
03:37:22,374 --> 03:37:25,644
REASONABLE, IT WAS ALL DONE BY

5812
03:37:25,644 --> 03:37:31,116
THE SAME CALCULATIONS WITH TIME

5813
03:37:31,116 --> 03:37:32,084
LAB, GOT CONDITIONS, AT THE END

5814
03:37:32,084 --> 03:37:34,653
OF THE TALK I WILL TALK ABOUT

5815
03:37:34,653 --> 03:37:36,188
WHAT WE'RE DOING TRY TO TRY TO

5816
03:37:36,188 --> 03:37:38,257
DO A BETTER JOB AND PAINTING THE

5817
03:37:38,257 --> 03:37:40,059
FIELD TOWARD BETTER DATA SETS.

5818
03:37:40,059 --> 03:37:41,860
SO WHAT DO WE WANT TO DO, WE

5819
03:37:41,860 --> 03:37:44,296
WANT TO FOCUS OW BENCHMARKING

5820
03:37:44,296 --> 03:37:48,534
EFFORTS ON ROBUST CLEARLY

5821
03:37:48,534 --> 03:37:49,668
DEFINED END POINTS, WE WANT TO

5822
03:37:49,668 --> 03:37:51,203
UNDERSTAND THE DATA AND BE ABLE

5823
03:37:51,203 --> 03:37:57,977
TO REMOVE BAD ACTORS FROM THE

5824
03:37:57,977 --> 03:37:58,477
DATA SET.

5825
03:37:58,477 --> 03:38:00,479
WE WANT TO MAKE SURE THE DATA IS

5826
03:38:00,479 --> 03:38:00,713
CORRECT.

5827
03:38:00,713 --> 03:38:02,648
I DON'T LIKE LEADER BOARD AND

5828
03:38:02,648 --> 03:38:03,949
AMERICANS, AND TURNING THIS INTO

5829
03:38:03,949 --> 03:38:05,384
A CONTEXT, I WOULD PREFER TO

5830
03:38:05,384 --> 03:38:07,052
HAVE PEOPLE IN THE FIELD WORKING

5831
03:38:07,052 --> 03:38:08,053
TOGETHER TO FIGURE OUT HOW TO

5832
03:38:08,053 --> 03:38:13,359
COME UP WITH BETTER METHODS. 

5833
03:38:13,359 --> 03:38:16,395
SO THAT'S RANT NUMBER 1, THERE'S

5834
03:38:16,395 --> 03:38:18,330
MOVE NOW IN RANT NUMBER 2 AND

5835
03:38:18,330 --> 03:38:19,198
TALK ABOUT STATISTICS.

5836
03:38:19,198 --> 03:38:23,902
SO TYPICALLY WHEN WE'RE

5837
03:38:23,902 --> 03:38:25,471
EVALUATING A MACHINE LEARNING

5838
03:38:25,471 --> 03:38:27,906
MODEL, WE TAKE THE DATA, AND

5839
03:38:27,906 --> 03:38:30,109
SPLIT IT INTO A DATA SET, TEST

5840
03:38:30,109 --> 03:38:31,543
SET, WE CAN USE A RANGE OF

5841
03:38:31,543 --> 03:38:37,383
METHODS TO DO THAT, WE WILL THEN

5842
03:38:37,383 --> 03:38:39,385
DO, 5, 10, 25 DIFFERENT SPLITS

5843
03:38:39,385 --> 03:38:41,053
OF THE DATA SET.

5844
03:38:41,053 --> 03:38:42,521
WE WILL CALCULATE STATISTICS,

5845
03:38:42,521 --> 03:38:45,424
SO, IF WE'RE DOING A REGRESSION

5846
03:38:45,424 --> 03:38:46,859
MODEL WE MIGHT BE CALCULATING

5847
03:38:46,859 --> 03:38:50,963
SOMETHING LIKE THE MEAN AVERAGE

5848
03:38:50,963 --> 03:38:53,966
ERROR OR THE R-QUARED OVER

5849
03:38:53,966 --> 03:38:55,701
REGRESSION MODEL, WE CAN USE

5850
03:38:55,701 --> 03:38:57,202
DIFFERENT METRICS FOR

5851
03:38:57,202 --> 03:38:58,604
CLASSIFICATION MODELS BUT WHAT

5852
03:38:58,604 --> 03:39:01,140
TYPICALLY GETS REPORTED THEN IN

5853
03:39:01,140 --> 03:39:02,775
THE LITERATURE, IS SIMPLY AN

5854
03:39:02,775 --> 03:39:05,344
AVERAGE OVER THAT SET OF CROSS

5855
03:39:05,344 --> 03:39:06,011
VALIDATION FOLDS.

5856
03:39:06,011 --> 03:39:09,448
SO WHAT PEOPLE WILL DO IS PUT UP

5857
03:39:09,448 --> 03:39:10,716
A TABLE, THEY SHE ALL THE

5858
03:39:10,716 --> 03:39:12,584
METHODS IN ROWS AND THEN THEY

5859
03:39:12,584 --> 03:39:13,919
HAVE THE COLUMNS WITH THEIR

5860
03:39:13,919 --> 03:39:15,587
STATISTICS AND THEY COME UP WITH

5861
03:39:15,587 --> 03:39:18,657
WHAT I REFER TO AS THE DREADED

5862
03:39:18,657 --> 03:39:19,725
BOLD TABLE EMPLOY SO PEOPLE WILL

5863
03:39:19,725 --> 03:39:22,261
PUT A TABLE UP, THEY WILL PUT

5864
03:39:22,261 --> 03:39:26,665
THE METHOD THAT HAS, YOU KNOW A

5865
03:39:26,665 --> 03:39:28,500
HIGHER VALUE IS BET BETTER IN

5866
03:39:28,500 --> 03:39:30,702
THIS PARTICULAR CASE WE'RE

5867
03:39:30,702 --> 03:39:32,171
LOOKING ATLY DIFFERENT

5868
03:39:32,171 --> 03:39:33,372
CLASSIFICATION METHODS,LET

5869
03:39:33,372 --> 03:39:34,640
HIGHER VALUE IS BETTER, THEY

5870
03:39:34,640 --> 03:39:36,041
TAKE THE METHOD WITH THE HIGHEST

5871
03:39:36,041 --> 03:39:39,044
VALUE, PUT IT IN BOLD, IT'S

5872
03:39:39,044 --> 03:39:40,379
USUALLY THEIR METHOD WHOO HOO

5873
03:39:40,379 --> 03:39:40,712
THEY WON.

5874
03:39:40,712 --> 03:39:43,215
BUT LOOK AT THIS TABLE, LOOKA

5875
03:39:43,215 --> 03:39:48,854
THE ROCK AUC, 1 METHOD HAD .901,

5876
03:39:48,854 --> 03:39:49,655
THE OTHER IS .904.

5877
03:39:49,655 --> 03:39:52,257
IS THAT REALLY DIFFERENT?

5878
03:39:52,257 --> 03:39:53,992
AND THEN PEOPLE WILL EITHER PUT

5879
03:39:53,992 --> 03:39:56,028
UP THIS BOLD TABLE OR THEY'LL

5880
03:39:56,028 --> 03:39:57,930
PUT SOME BAR GRAPH LIKE THIS TO

5881
03:39:57,930 --> 03:40:06,705
SHOW THE METHOD COMPARISON.

5882
03:40:06,705 --> 03:40:08,173
YOU KNOW OR THEY WILL TRY TO

5883
03:40:08,173 --> 03:40:09,975
WASH AWAY THEIR SINS BUT THEY'RE

5884
03:40:09,975 --> 03:40:11,210
STILL PUTTING THIS BOLD TABLE

5885
03:40:11,210 --> 03:40:12,945
TOGETHER, YOU KNOW THE STANDARD

5886
03:40:12,945 --> 03:40:15,047
DEVIATION IS A MEASURE OF

5887
03:40:15,047 --> 03:40:16,949
VARIANCE, IT'S NOT A STATISTICAL

5888
03:40:16,949 --> 03:40:18,917
TEST, THIS MOT A VALID WAY TO

5889
03:40:18,917 --> 03:40:20,586
COMPARE METHODS.

5890
03:40:20,586 --> 03:40:22,821
YOU KNOW, PEOPLE WILL DO WHAT I

5891
03:40:22,821 --> 03:40:24,256
REFER TO AS EYE DYNAMITE PLOT

5892
03:40:24,256 --> 03:40:27,793
WHERE YOU THEN PLOT THE STANDARD

5893
03:40:27,793 --> 03:40:29,795
DEVIATION, AGAIN THIS IS NOT

5894
03:40:29,795 --> 03:40:30,095
APPROPRIATE.

5895
03:40:30,095 --> 03:40:32,231
WE HAVE TO THINK ABOUT THE FACT

5896
03:40:32,231 --> 03:40:35,434
THAT WE HAVE MULTIPLE FOLDS OF

5897
03:40:35,434 --> 03:40:36,935
CROSS VALIDATION SO WHAT YOU'RE

5898
03:40:36,935 --> 03:40:38,704
REALLY DOING IS YOU'RE COMPARING

5899
03:40:38,704 --> 03:40:40,038
DISTRIBUTIONS, SO WHAT I HAVE

5900
03:40:40,038 --> 03:40:44,443
HERE IS A BOX PLOT SHOWING 2

5901
03:40:44,443 --> 03:40:47,212
DEFINITE METHODS, I'VE DONE 25

5902
03:40:47,212 --> 03:40:53,252
FOLDS OF CROSS VALIDATION, YOU

5903
03:40:53,252 --> 03:40:54,820
KNOW THE GREEN ON THE RIGHT IS

5904
03:40:54,820 --> 03:40:56,054
BETTER, THE RED LINES ON THE

5905
03:40:56,054 --> 03:40:57,356
LEFT IS WHERE THE METHOD IS

5906
03:40:57,356 --> 03:40:59,591
BETTER SO WE GET A MIXTURE, IT'S

5907
03:40:59,591 --> 03:41:03,962
NOT ALWAYS 1 METHOD, BETTER 

5908
03:41:03,962 --> 03:41:04,363
THANTH OTHER.

5909
03:41:04,363 --> 03:41:08,634
BUT WHAT ARE WE DOING HERE?

5910
03:41:08,634 --> 03:41:09,568
WE'RE COMPARING DISTRIBUTIONS.

5911
03:41:09,568 --> 03:41:13,071
THERE'S NOR THAN A HUNDRED YEARS

5912
03:41:13,071 --> 03:41:17,876
OF STATISTICS GOING BACK TO

5913
03:41:17,876 --> 03:41:19,711
GUINNESS BEERS IN THE 1910S.

5914
03:41:19,711 --> 03:41:21,313
THERE ARE GOOD STATISTICAL

5915
03:41:21,313 --> 03:41:22,714
METHODS FOR COMPARING BRIEWKSS

5916
03:41:22,714 --> 03:41:28,020
IT'S JUST THAT OUR FIELD ISN'T

5917
03:41:28,020 --> 03:41:28,320
USING THESE.

5918
03:41:28,320 --> 03:41:29,821
SO WHAT WE NEED TO DO WHEN

5919
03:41:29,821 --> 03:41:31,557
COMPARING THESE IS SHOW A

5920
03:41:31,557 --> 03:41:34,293
P-VALUE, SHOW THERE IS

5921
03:41:34,293 --> 03:41:35,894
STATISTICALLY SIGNIFICANT

5922
03:41:35,894 --> 03:41:37,930
DIFFERENCE BETWEEN THE METHOD.

5923
03:41:37,930 --> 03:41:39,665
SHOW, CALCULATE SOMETHING LIKE

5924
03:41:39,665 --> 03:41:41,533
COHENS D AND SHOW AN EFFECT

5925
03:41:41,533 --> 03:41:41,733
SIZE.

5926
03:41:41,733 --> 03:41:44,970
WHAT I WOULD LOVE TO SEE OR I'M

5927
03:41:44,970 --> 03:41:46,605
SORRY, JUST TAKE 1 MORE STEP

5928
03:41:46,605 --> 03:41:46,939
BACK.

5929
03:41:46,939 --> 03:41:49,808
THE OTHER THING THAT GETS A BIT

5930
03:41:49,808 --> 03:41:51,643
MORE COMPLICATED WHEN YOU ARE

5931
03:41:51,643 --> 03:41:53,745
COMPARING MORE THAN 2 METHODS,

5932
03:41:53,745 --> 03:41:56,215
YOU HAVE A HIGHER PROBABILITY OF

5933
03:41:56,215 --> 03:42:00,452
JUST 1 METHOD BEING BETTER THAN

5934
03:42:00,452 --> 03:42:01,720
THE OTHER BY RANDOM CHANCE BUT

5935
03:42:01,720 --> 03:42:03,288
AGAIN, YOU KNOW THERE ARE

5936
03:42:03,288 --> 03:42:04,122
STATISTICAL METHODS FOR

5937
03:42:04,122 --> 03:42:06,425
HUNDREDLING THIS, THERE ARE

5938
03:42:06,425 --> 03:42:10,929
METHODS LIKE ANOVA, THERE ARE

5939
03:42:10,929 --> 03:42:11,997
METHODS LIKE [INDISCERNIBLE]

5940
03:42:11,997 --> 03:42:13,098
DIFFERENCE, THERE ARE METHODS

5941
03:42:13,098 --> 03:42:15,467
LIKE FREED BAN'S TEST WHICH WAS

5942
03:42:15,467 --> 03:42:16,168
DEVELOPED IN 1937.

5943
03:42:16,168 --> 03:42:21,473
SO 15 WE HAVE A LONG HISTORY OF

5944
03:42:21,473 --> 03:42:22,507
STATISTICAL METHODS FOR DOING

5945
03:42:22,507 --> 03:42:24,176
THESE SORTS OF THINGS SO WHAT I

5946
03:42:24,176 --> 03:42:27,045
WOULD LIKE TO SEE IN PAPERS IS

5947
03:42:27,045 --> 03:42:27,946
PEOPLE SHOWING STATISTICS AND

5948
03:42:27,946 --> 03:42:31,550
SHOWING THAT THEIR METHODS ARE

5949
03:42:31,550 --> 03:42:34,052
ACTUALLY BETTER IN A SIGNIFICANT

5950
03:42:34,052 --> 03:42:34,286
FASHION.

5951
03:42:34,286 --> 03:42:36,288
SO, YOU KNOW I WANT TO -- I'VE

5952
03:42:36,288 --> 03:42:38,590
WRITTEN A LOT ABOUT THIS IN THE

5953
03:42:38,590 --> 03:42:43,629
PAST BUT I WANT TO CONCLUDE WITH

5954
03:42:43,629 --> 03:42:45,330
SOMETHING POSITIVE BECAUSE SO

5955
03:42:45,330 --> 03:42:47,599
FAR I'VE BEEN THIS GIVING--YOU,

5956
03:42:47,599 --> 03:42:49,334
OLD ANYWAY A SAYING YOU KIDS

5957
03:42:49,334 --> 03:42:50,836
IMET OUT OF MY YARD SO I THINK

5958
03:42:50,836 --> 03:42:52,404
WE HAVE COUPLE OF EFFORTS

5959
03:42:52,404 --> 03:42:54,439
UNDERWAY THAT I THINK WILL

5960
03:42:54,439 --> 03:42:57,342
HOPEFULLY ADDRESS THE PROBLEM.

5961
03:42:57,342 --> 03:43:00,679
SO I'VE BEEN WORKING WITH A

5962
03:43:00,679 --> 03:43:03,315
GROUP OF PEOPLE FROM ACROSS

5963
03:43:03,315 --> 03:43:05,550
BIOTECH AND LARGE PHARMA AND

5964
03:43:05,550 --> 03:43:09,621
WE'VE BEEN PUTTING TOGETHER A

5965
03:43:09,621 --> 03:43:11,590
SET OF GUIDELINES FOR MODEL

5966
03:43:11,590 --> 03:43:14,726
EVALUATION AND THERE HAVE BEEN

5967
03:43:14,726 --> 03:43:17,629
REALLY 2 DANCHIBLE OUTCOMES FROM

5968
03:43:17,629 --> 03:43:21,033
THIS, AND THE FIRST IS POLARIS,

5969
03:43:21,033 --> 03:43:23,935
SO THERE'S A WEBSITE CALLED

5970
03:43:23,935 --> 03:43:25,637
POLARIS HUB.IO WHERE WE ARE

5971
03:43:25,637 --> 03:43:29,041
HOSTING DATA SETS BUT NOT ONLY

5972
03:43:29,041 --> 03:43:31,576
ARE WE HOSTING DATA SETS, WE'RE

5973
03:43:31,576 --> 03:43:36,181
PROVIDING RECOMMENDATIONS ON

5974
03:43:36,181 --> 03:43:37,949
THOSE DATA SETS, AND WITHIN

5975
03:43:37,949 --> 03:43:40,052
POLARIS RIGHT NOW THERE ARE 2

5976
03:43:40,052 --> 03:43:41,119
APPROVED DATA SETS SO HOPEFULLY

5977
03:43:41,119 --> 03:43:47,993
THERE WILL BE A LOT MORE COMING

5978
03:43:47,993 --> 03:43:50,696
VERY SOON EMPLOY THE OTHER THING

5979
03:43:50,696 --> 03:43:54,366
THAT WE HAVE IS A MANUSCRIPT

5980
03:43:54,366 --> 03:43:58,136
LAYING OUTSETS OF STATISTICAL

5981
03:43:58,136 --> 03:43:59,037
METHODDINGS FOR COMPARING

5982
03:43:59,037 --> 03:44:03,008
MACHINE LEARNING MODELS IN DRUG

5983
03:44:03,008 --> 03:44:03,942
DISCOVERY ACCOMPANYING THIS

5984
03:44:03,942 --> 03:44:08,080
MACHINEUE SCRIPT IS A SET OF

5985
03:44:08,080 --> 03:44:09,581
JUPIT OR NOTEBOOKS OF CODE

5986
03:44:09,581 --> 03:44:10,682
WALKING YOU THROUGH AND SHOWING

5987
03:44:10,682 --> 03:44:13,752
YOU HOW TO DO ALL OF THIS.

5988
03:44:13,752 --> 03:44:16,021
THE MANUSCRIPT IS NOW, GOING

5989
03:44:16,021 --> 03:44:18,457
THROUGH THE APPROVAL PROCESS, AT

5990
03:44:18,457 --> 03:44:21,059
ALL THE COMPANIES OF ALL THE

5991
03:44:21,059 --> 03:44:22,894
AUTHORS, I AM EXTREMELY HOPEFUL

5992
03:44:22,894 --> 03:44:26,298
THAT BY THE END OF THIS MONTH,

5993
03:44:26,298 --> 03:44:27,933
THIS WILL BE AVAILABLE ON

5994
03:44:27,933 --> 03:44:30,135
ARCHIVE, WHAT WE WOULD LIKE TO

5995
03:44:30,135 --> 03:44:33,338
DO IS PUT THIS OUT THERE FOR A

5996
03:44:33,338 --> 03:44:36,808
FEW MONTHS, GIVE THE COMMUNITY

5997
03:44:36,808 --> 03:44:38,844
AN OPPORTUNITY TO COMMENT ON IT,

5998
03:44:38,844 --> 03:44:41,747
AND THEN, YOU KNOW, OUR HOPE IS

5999
03:44:41,747 --> 03:44:43,281
TO THEN PUBLISH THIS AND HAVE A

6000
03:44:43,281 --> 03:44:47,686
SET OF GUIDE LINES THAT CAN BE

6001
03:44:47,686 --> 03:44:50,589
ADOPTED BY JOURNALS.

6002
03:44:50,589 --> 03:44:52,891
SO, YOU KNOW ULTIMATELY WHAT CAN

6003
03:44:52,891 --> 03:44:53,225
WE DO BETTER?

6004
03:44:53,225 --> 03:45:00,866
WE NEED TO USE BETTER BENCHMARK

6005
03:45:00,866 --> 03:45:01,199
SETS.

6006
03:45:01,199 --> 03:45:02,834
I THINK POLARIS CAN BE A HUGE

6007
03:45:02,834 --> 03:45:03,235
HELP THERE.

6008
03:45:03,235 --> 03:45:06,605
WE NEED TO HAVE GUIDELINES FOR

6009
03:45:06,605 --> 03:45:07,339
APPROPRIATE STATISTICAL

6010
03:45:07,339 --> 03:45:08,340
COMPARISONS.

6011
03:45:08,340 --> 03:45:09,941
IF YOU'RE USING MOLECULE NET OR

6012
03:45:09,941 --> 03:45:11,676
THE THERAPEUTIC DATA COMMONS TO

6013
03:45:11,676 --> 03:45:12,978
COMPARE METHODS, I WOULD

6014
03:45:12,978 --> 03:45:14,813
RECOMMEND THAT YOU STOP DOING

6015
03:45:14,813 --> 03:45:19,618
THAT AND GO GET A BETTER DATA

6016
03:45:19,618 --> 03:45:19,885
SET.

6017
03:45:19,885 --> 03:45:30,362
AND WITH THAT, I AM GOING TO

6018
03:45:33,331 --> 03:45:33,799
STOP.

6019
03:45:33,799 --> 03:45:36,101
>> THAT WAS AWESOME PAT,
NOT TOO

6020
03:45:36,101 --> 03:45:36,334
CRANKY.

6021
03:45:36,334 --> 03:45:38,336
I MEAN YOU BROUGHT IT UP AT THE

6022
03:45:38,336 --> 03:45:38,503
END.

6023
03:45:38,503 --> 03:45:39,671
GIVES AUSA LOT TO TALK BCH I

6024
03:45:39,671 --> 03:45:40,972
WILL POINT OUT HERE FOR THE

6025
03:45:40,972 --> 03:45:41,773
SPEAKERS, THERE ARE SOME

6026
03:45:41,773 --> 03:45:43,074
QUESTIONS YOU MIGHT WANT TO

6027
03:45:43,074 --> 03:45:44,876
ANSWER IN THE CHAT BY NOT JUST

6028
03:45:44,876 --> 03:45:46,578
LIVE BUT BY TYPING, SO WITHOUT

6029
03:45:46,578 --> 03:45:47,412
FURTHER ADO BECAUSE I WANT TO

6030
03:45:47,412 --> 03:45:50,048
MAKE SURE YOU HAVE ALL OF YOUR

6031
03:45:50,048 --> 03:45:51,750
MINUES, RUSS, SO THANK YOU PAT,

6032
03:45:51,750 --> 03:45:54,986
I WANT TO INTRODUCE PROFESSOR

6033
03:45:54,986 --> 03:45:57,122
RUSS ALTMAN FROM STANFORD

6034
03:45:57,122 --> 03:45:57,456
UNIVERSITY.

6035
03:45:57,456 --> 03:45:58,690
HE IS PROBABLY IN MANY DIDN'TS

6036
03:45:58,690 --> 03:46:00,392
BUT I KNOW HIM MOST IN THE

6037
03:46:00,392 --> 03:46:01,493
DEPARTMENT OF BIOENGINEERING AND

6038
03:46:01,493 --> 03:46:03,228
HE'S DONE TONS AND TONS OF

6039
03:46:03,228 --> 03:46:04,663
AMAZING THINGS AND A LOT ON

6040
03:46:04,663 --> 03:46:08,934
LOOKING AT DRUG ACTION ACROSS

6041
03:46:08,934 --> 03:46:09,734
SCALES.

6042
03:46:09,734 --> 03:46:11,403
AND I'LL JUST HAND IT OVER TO

6043
03:46:11,403 --> 03:46:11,570
YOU.

6044
03:46:11,570 --> 03:46:16,341
>> CAN YOU HEAR ME OKAY?

6045
03:46:16,341 --> 03:46:17,542
>> YEAH, YOU SOUND GREAT
NTHANKS

6046
03:46:17,542 --> 03:46:18,944
TO THE ORGANIZERS FOR INVITING

6047
03:46:18,944 --> 03:46:22,080
ME, WE'RE DOING A LOT OF STUFF

6048
03:46:22,080 --> 03:46:23,682
ON THE TECHNICAL SIDE LOOKING AT

6049
03:46:23,682 --> 03:46:24,683
POCKETS AND BINDING AND I'M NOT

6050
03:46:24,683 --> 03:46:25,884
GOING TO TALK ABOUT THAT TODAY,

6051
03:46:25,884 --> 03:46:27,419
I WILL TALK ABOUT A CRAZY THING

6052
03:46:27,419 --> 03:46:31,790
THAT WE'RE ALWAYS DOING, WHICH

6053
03:46:31,790 --> 03:46:33,291
IS LARGE LANGUAGE MODEL

6054
03:46:33,291 --> 03:46:34,593
SOPHISTICATED TRPT REGULATORY

6055
03:46:34,593 --> 03:46:36,061
SUBMISSION, EVALUATION, BY

6056
03:46:36,061 --> 03:46:37,896
REGULATORS, LIKE AT THE FDA, SO

6057
03:46:37,896 --> 03:46:39,631
I THINK WE ALL KNOW, OR YOU

6058
03:46:39,631 --> 03:46:41,233
WON'T BE SURPRISED TO LEARN THAT

6059
03:46:41,233 --> 03:46:42,534
MANY OF THE SUBMISSIONS THAT ARE

6060
03:46:42,534 --> 03:46:46,571
NOW COMING IN TO THE FDA, ARE

6061
03:46:46,571 --> 03:46:47,939
PARTIALLY WRITTEN BY LARGE

6062
03:46:47,939 --> 03:46:49,007
LANGUAGE MODELS, RIGHT IN PEOPLE

6063
03:46:49,007 --> 03:46:51,376
ARE USING THEM, IT'S NOT -- IT'S

6064
03:46:51,376 --> 03:46:52,911
TOTALLY ALLOWED AS LONG AS SOME

6065
03:46:52,911 --> 03:46:54,679
HUMAN TAKES RESPONSIBILITY FOR

6066
03:46:54,679 --> 03:46:55,914
THE TEXT THAT'S GENERATED AND IT

6067
03:46:55,914 --> 03:47:02,220
CAN ACTUALLY BE A PRODUCTIVITY

6068
03:47:02,220 --> 03:47:02,621
IMPROVEMENT.

6069
03:47:02,621 --> 03:47:04,356
WELL, I'M INVOLVED IN AN FDA

6070
03:47:04,356 --> 03:47:05,457
SUPPORTED TRUSTEES OF

6071
03:47:05,457 --> 03:47:07,792
EXCELLENCE, AND WE GOT A REQUEST

6072
03:47:07,792 --> 03:47:09,794
FROM THE FDA ABOUT BUILDING A

6073
03:47:09,794 --> 03:47:11,096
LARGE LANGUAGE MODEL, YOU KNOW

6074
03:47:11,096 --> 03:47:12,230
NOT TO MAKE DECISIONS ABOUT YOU

6075
03:47:12,230 --> 03:47:14,666
TO BE KIND OF A CO PILOT FOR

6076
03:47:14,666 --> 03:47:16,501
REGULATORS AS THEY LOOK AT NEW

6077
03:47:16,501 --> 03:47:18,136
SUBMISSIONS, AND SO THIS IS

6078
03:47:18,136 --> 03:47:19,771
PRETTY EXCITING, IT DOES RAISE

6079
03:47:19,771 --> 03:47:21,339
THE POSSIBILITY THAT AI WILL BE

6080
03:47:21,339 --> 03:47:23,041
WRITING THE SUBMISSIONS AND THAT

6081
03:47:23,041 --> 03:47:24,910
AI WILL BE EVALUATING THE

6082
03:47:24,910 --> 03:47:26,244
SUBMISSIONS, YOU KNOW WELCOME TO

6083
03:47:26,244 --> 03:47:27,078
THE NEW WORLD.

6084
03:47:27,078 --> 03:47:29,381
I HOPE THE AI DOESN'T KILL TOO

6085
03:47:29,381 --> 03:47:29,748
MANY HUMANS.

6086
03:47:29,748 --> 03:47:31,249
SO I WANT TO TELL YOU ABOUT THAT

6087
03:47:31,249 --> 03:47:32,484
PROJECT WHICH IS REALLY VERY

6088
03:47:32,484 --> 03:47:34,486
MUCH IN PROGRESS AND I'M NOT

6089
03:47:34,486 --> 03:47:37,155
GOING TO BLOW YOU AWAY FOR A

6090
03:47:37,155 --> 03:47:37,856
SYSTEM THAT'S READY FOR YEEL

6091
03:47:37,856 --> 03:47:39,558
TIME USE BUT I WANT TO TELL YOU

6092
03:47:39,558 --> 03:47:42,561
HOW IT'S WORKING AND THIS IS

6093
03:47:42,561 --> 03:47:45,063
WORK OF A GREAT Ph.D. STUDENT

6094
03:47:45,063 --> 03:47:46,932
IN MY LAB, BETTY [INDISCERNIBLE]

6095
03:47:46,932 --> 03:47:49,601
WORKING WITH A COMPUTER

6096
03:47:49,601 --> 03:47:50,969
SCIENTISTS AT STANFORD AND

6097
03:47:50,969 --> 03:47:52,137
COLLABORATORS AT THE FDA.

6098
03:47:52,137 --> 03:47:53,572
SO THE SPECIFIC USE CASE WE'RE

6099
03:47:53,572 --> 03:47:56,241
BELLING FOR IS WITH THE OFFICE

6100
03:47:56,241 --> 03:47:57,642
OF GENERIC DRUGS WHERE WE'RE

6101
03:47:57,642 --> 03:48:00,512
LOOKING AT NEW GENERIC DRUG

6102
03:48:00,512 --> 03:48:01,680
APPLICATIONS WHERE, YOU KNOW

6103
03:48:01,680 --> 03:48:03,548
IT'S A TYPICAL CASE, THE DRUG

6104
03:48:03,548 --> 03:48:07,052
HAS -- THE ORIGINAL DRUG HAS

6105
03:48:07,052 --> 03:48:07,819
GONE OFF PATENT.

6106
03:48:07,819 --> 03:48:09,521
A NEW COMPANY WANTS TO SELL A

6107
03:48:09,521 --> 03:48:11,523
COPY AND THEY NEED TO PROVE THAT

6108
03:48:11,523 --> 03:48:12,490
IT'S EQUIVALENT.

6109
03:48:12,490 --> 03:48:15,160
SO THIS IS A VERY FOCUSED CASE,

6110
03:48:15,160 --> 03:48:16,561
BUT WE'RE VERY AWARE THAT IF WE

6111
03:48:16,561 --> 03:48:18,897
CAN DO THIS, WE MIGHT BE ABLE TO

6112
03:48:18,897 --> 03:48:20,899
GENTLEMENNIZE TO OTHER, TO

6113
03:48:20,899 --> 03:48:21,700
SUPPORTING OTHER REGULATORY

6114
03:48:21,700 --> 03:48:24,502
DECISIONS, SO OF COURSE YOU WANT

6115
03:48:24,502 --> 03:48:28,540
THE PK, FORMA CO DYNAMICS AND

6116
03:48:28,540 --> 03:48:29,474
ABSORPTION DISTRIBUTION

6117
03:48:29,474 --> 03:48:35,614
METABOLISM AND EXKRIEWGZ TO BE

6118
03:48:35,614 --> 03:48:35,947
EQUIVALENT.

6119
03:48:35,947 --> 03:48:37,782
SO SO YOU CAN MAKE A DECISION ON

6120
03:48:37,782 --> 03:48:39,751
WHETHER THIS NEW DRUG WILL BE

6121
03:48:39,751 --> 03:48:40,218
APPROVED.

6122
03:48:40,218 --> 03:48:41,119
AND THEY SEE AN OPPORTUNITY AND

6123
03:48:41,119 --> 03:48:44,089
WE SEE AN OPPORTUNITY TO BUILD A

6124
03:48:44,089 --> 03:48:49,160
CHAT BOT AS A CO PILOT AS I SAID

6125
03:48:49,160 --> 03:48:50,795
BEFORE FOR REVIEWERS AT THE

6126
03:48:50,795 --> 03:48:52,964
OFFICE OF GENERIC DRUGS.

6127
03:48:52,964 --> 03:48:54,199
JUST ANSWERING QUESTIONS, THESE

6128
03:48:54,199 --> 03:48:56,601
APPLICATIONS CAN BE HUNDREDS OF

6129
03:48:56,601 --> 03:48:58,737
PAGES AND YOU MIGHT WANT TO SAY

6130
03:48:58,737 --> 03:49:01,439
DID THEY STUDY IMPACT ON

6131
03:49:01,439 --> 03:49:02,374
PREGNANCY, HOPEFULLY THEY DIDN'T

6132
03:49:02,374 --> 03:49:03,808
IN THIS CASE UNLESS THERE WAS AN

6133
03:49:03,808 --> 03:49:05,510
ISSUE IN THE ORIGINAL DRUG.

6134
03:49:05,510 --> 03:49:07,479
THERE'S A LOT OF QUESTIONS CAN

6135
03:49:07,479 --> 03:49:08,913
YOU ASK AND REQUIREMENTS.

6136
03:49:08,913 --> 03:49:12,784
WE CANNOT TRAIN THIS ON

6137
03:49:12,784 --> 03:49:13,985
PROPRIETARY DATA BECAUSE THEY'RE

6138
03:49:13,985 --> 03:49:16,221
NOT ALLOWED TO GIVE US

6139
03:49:16,221 --> 03:49:17,022
PROPRIETARY DATAY SOPHISTICATED

6140
03:49:17,022 --> 03:49:21,960
WE HAVE TO FIND PROXIES FOR DRUG

6141
03:49:21,960 --> 03:49:22,727
DISCUSSIONS, FOR EXAMPLE,

6142
03:49:22,727 --> 03:49:27,532
JOURNAL ARTICLES ABOUT THIS KIND

6143
03:49:27,532 --> 03:49:28,967
OF EVALUATION AND EXISTING DRUG

6144
03:49:28,967 --> 03:49:31,870
LABELS WHICH IS A BIG 1 I WILL

6145
03:49:31,870 --> 03:49:37,609
GET TO.

6146
03:49:37,609 --> 03:49:41,179
EEIVETUALLY THEY WILL BE OPEN,

6147
03:49:41,179 --> 03:49:43,248
CHATGPT WILL NOT WORK BECAUSE

6148
03:49:43,248 --> 03:49:44,549
THEY CAN'T SEND REGULATORY

6149
03:49:44,549 --> 03:49:45,717
SUBMISSIONS FROM A COMPANY OVER

6150
03:49:45,717 --> 03:49:49,654
THE INTERNET TO A THIRD PARTY. 

6151
03:49:49,654 --> 03:49:50,889
THAT'S NOT ALLOWED.

6152
03:49:50,889 --> 03:49:53,558
AND THE FDA DOESN'T HAVE 10 OR

6153
03:49:53,558 --> 03:49:55,260
20,000 GPUs SO WE HAVE TO

6154
03:49:55,260 --> 03:49:56,361
BUILD A SMALL MODEL THEY WOULD

6155
03:49:56,361 --> 03:49:58,196
BE ABLE TO USE, THIS IS VERY

6156
03:49:58,196 --> 03:50:00,865
MUCH A PROTOTYPE PROJECT TO SEE

6157
03:50:00,865 --> 03:50:03,068
HOW CAN YOU DEAL WITH THESE

6158
03:50:03,068 --> 03:50:04,235
CONSTRAINS AND YOU CAN BUILD

6159
03:50:04,235 --> 03:50:05,003
SOMETHING USEFUL.

6160
03:50:05,003 --> 03:50:06,705
SO THE MAIN IDEA OF THIS PRONG

6161
03:50:06,705 --> 03:50:07,338
ECTOMYOSIN IS DO SOMETHING

6162
03:50:07,338 --> 03:50:08,673
OTHERS HAVE DONE THIS, IS NOT

6163
03:50:08,673 --> 03:50:10,975
OUR IDEA, AND IT'S CALLED A

6164
03:50:10,975 --> 03:50:12,944
DISTILLED MODEL AND THE IDEA IS

6165
03:50:12,944 --> 03:50:16,081
THAT YOU USE A LARGE AND CAPABLE

6166
03:50:16,081 --> 03:50:20,118
TEACHER MODEL SUCH AS GPT 4.0

6167
03:50:20,118 --> 03:50:22,387
WHICH IS FROM OPEN AI, TO

6168
03:50:22,387 --> 03:50:23,922
DISTILL KNOWLEDGE ABOUT HOW TO

6169
03:50:23,922 --> 03:50:27,926
READ A DRUG APPLICATION INTO A

6170
03:50:27,926 --> 03:50:28,960
SMALLER, MUCH SMALLER STUDENT

6171
03:50:28,960 --> 03:50:31,229
LANGUAGE MODEL THAT ONLY CAN

6172
03:50:31,229 --> 03:50:36,968
TALK ABOUT DRUGS, DRUG LABELS,

6173
03:50:36,968 --> 03:50:37,836
AND DRUG FEATURES.

6174
03:50:37,836 --> 03:50:38,703
IT CAN'T ANSWER QUESTIONS ABOUT

6175
03:50:38,703 --> 03:50:40,004
THE HISTORY OF THE WORLD, IT

6176
03:50:40,004 --> 03:50:41,473
DOESN'T KNOW WHO WON THE

6177
03:50:41,473 --> 03:50:43,608
SUPER BOWL, IT HAS NO IDEA AT

6178
03:50:43,608 --> 03:50:46,044
MANY THINGS BUT IT'S PRETTY GOOD

6179
03:50:46,044 --> 03:50:49,814
WITH DRUGS BECAUSE YOU HAVE

6180
03:50:49,814 --> 03:50:50,849
QUERIED GPT4.0000S AND THOUSANDS

6181
03:50:50,849 --> 03:50:53,184
OF TRIEMS AND USE THOSE QUESTION

6182
03:50:53,184 --> 03:50:55,420
AND ANSWER PAIRS TO FINE TUNE A

6183
03:50:55,420 --> 03:50:59,257
PUBLIC DOMAIN MUCH SMALLER

6184
03:50:59,257 --> 03:50:59,457
MODEL.

6185
03:50:59,457 --> 03:51:01,292
SO LLAMA 3.1 WITH 8 BILLION

6186
03:51:01,292 --> 03:51:02,861
PARAMETERS, BELIEVE IT OR NOT 8

6187
03:51:02,861 --> 03:51:04,462
BILLION IS A VERY SMALL MODEL.

6188
03:51:04,462 --> 03:51:06,364
SO THAT WILL BE OUR STARTING

6189
03:51:06,364 --> 03:51:06,564
MODEL.

6190
03:51:06,564 --> 03:51:08,666
A BASE LINE AND WE WILL TRY TO

6191
03:51:08,666 --> 03:51:12,637
SEE HOW GOOD CAN WE GET IT AT

6192
03:51:12,637 --> 03:51:14,572
UNDERSTANDING THE CONTENTS OF

6193
03:51:14,572 --> 03:51:14,873
DRUG LABELS.

6194
03:51:14,873 --> 03:51:16,708
SO IT TURNS OUT THAT THE REVIEW

6195
03:51:16,708 --> 03:51:18,676
PROCESS IS WELL REPRESENTED, THE

6196
03:51:18,676 --> 03:51:22,680
KIND OF LANGUAGE THAT YOU SEE IN

6197
03:51:22,680 --> 03:51:23,782
APPLICATIONS IS VERY WELL

6198
03:51:23,782 --> 03:51:24,949
REPRESENTED IN THE KIND OF

6199
03:51:24,949 --> 03:51:26,217
LANGUAGE THAT WINDS UP IN DRUG

6200
03:51:26,217 --> 03:51:27,952
LABEL ANDS THAT SHOULDN'T BE A

6201
03:51:27,952 --> 03:51:29,020
SURPRISE BECAUSE THOSE DRUG

6202
03:51:29,020 --> 03:51:30,321
LABELS ARE LITERALLY TIMES

6203
03:51:30,321 --> 03:51:31,923
WRITTEN BY THE PHOBE WHO IS

6204
03:51:31,923 --> 03:51:33,458
SUBMITTED THE DATA IN THE FIRST

6205
03:51:33,458 --> 03:51:33,658
PLACE.

6206
03:51:33,658 --> 03:51:37,462
THESE ARE ALL PUBLICLY

6207
03:51:37,462 --> 03:51:37,862
AVAILABLE.

6208
03:51:37,862 --> 03:51:38,963
2000 OR SO OR MORE DRUG LABELS

6209
03:51:38,963 --> 03:51:40,331
AND YOU CAN SEE THE PICTURE AT

6210
03:51:40,331 --> 03:51:42,700
THE BOTTOM RIGHT WHERE WE HAVE A

6211
03:51:42,700 --> 03:51:44,536
BIG MODEL LIKE CHAT GPT, AND I

6212
03:51:44,536 --> 03:51:47,972
WILL SHOW YOU CHAT GPT IS PRETTY

6213
03:51:47,972 --> 03:51:49,908
GOOD AT READING A DRUG LABEL AND

6214
03:51:49,908 --> 03:51:50,909
ANSWERING QUESTIONS ABOUT IT AND

6215
03:51:50,909 --> 03:51:53,378
THEN WE DO THIS KNOWLEDGE

6216
03:51:53,378 --> 03:51:54,512
TRANSFER MODEL INTO A SMALLER

6217
03:51:54,512 --> 03:51:55,580
MODEL AND THAT'S WHAT WE'RE

6218
03:51:55,580 --> 03:51:55,914
TRYING TO DO.

6219
03:51:55,914 --> 03:51:56,915
AND I WANT TO TELL YOU ABOUT

6220
03:51:56,915 --> 03:51:58,016
SOME STUFF AND I WANT TO TALK

6221
03:51:58,016 --> 03:51:59,184
ABOUT WHAT WE'VE DONE.

6222
03:51:59,184 --> 03:52:01,119
FIRST THING IS THESE LABELS ARE

6223
03:52:01,119 --> 03:52:02,420
PDFs, I DON'T WANT TO BORE YOU

6224
03:52:02,420 --> 03:52:05,123
BUT WE HAVE TO TAKE THE PDF AND

6225
03:52:05,123 --> 03:52:05,890
TURN IT INTO TEXT.

6226
03:52:05,890 --> 03:52:06,558
WE DO THAT.

6227
03:52:06,558 --> 03:52:07,725
IT'S CALLED A MARK DOWN AND THEN

6228
03:52:07,725 --> 03:52:10,228
WE NEED TO FIGURE OUT WHERE OR

6229
03:52:10,228 --> 03:52:12,030
HOW ARE WE GOING TO ASK

6230
03:52:12,030 --> 03:52:16,234
QUESTIONS OF THESE LABELS SO

6231
03:52:16,234 --> 03:52:17,769
THAT'S JUST A PICTURE, BY THE

6232
03:52:17,769 --> 03:52:18,837
WAY, I LOVE THIS PICTURE BECAUSE

6233
03:52:18,837 --> 03:52:21,039
1 OF THE THINGS IN IN LABEL IS

6234
03:52:21,039 --> 03:52:22,974
THE BLACK BOX WARNINGS, WELL

6235
03:52:22,974 --> 03:52:26,377
GUESS WHAT, A LOT OF TIMES WHEN

6236
03:52:26,377 --> 03:52:28,513
YOU MARKET DOWN TO ASCII TEXT,

6237
03:52:28,513 --> 03:52:29,914
YOU LOSE THE BOX, SO YOU HAVE TO

6238
03:52:29,914 --> 03:52:32,717
FIGURE OUT HOW TO RECOVER BIG

6239
03:52:32,717 --> 03:52:33,351
IMPORTANT WARNINGS THAT WOULD

6240
03:52:33,351 --> 03:52:35,954
HAVE BEEN IN A BLACK BOX WITH

6241
03:52:35,954 --> 03:52:39,090
PDF, IN A BOX ABOUT TEXT BECAUSE

6242
03:52:39,090 --> 03:52:40,825
IT'S NOT ALWAYS LABELED AS SUCH.

6243
03:52:40,825 --> 03:52:42,260
SO WE'RE DEALING WITH UP THING

6244
03:52:42,260 --> 03:52:43,194
SAYS LIKE THAT.

6245
03:52:43,194 --> 03:52:45,096
NOT EXCITING BUT CRITICAL TO GET

6246
03:52:45,096 --> 03:52:45,396
IT RIGHT.

6247
03:52:45,396 --> 03:52:47,632
WE ALSO KNOW THERE ARE ABOUT 9

6248
03:52:47,632 --> 03:52:49,400
OR 10 THINGS THAT ARE VERY

6249
03:52:49,400 --> 03:52:51,002
IMPORTANT TO EXTRACT ACCURATELY

6250
03:52:51,002 --> 03:52:53,471
FROM A LABEL AND ALSO FROM A

6251
03:52:53,471 --> 03:52:54,038
TRUG APPLICATION.

6252
03:52:54,038 --> 03:52:56,574
AND THOSE WOULD BE THE

6253
03:52:56,574 --> 03:52:59,644
INDICATION, IF THERE IS A BOXED

6254
03:52:59,644 --> 03:53:00,278
WARNING, THAT'S WHYY WOO CARE

6255
03:53:00,278 --> 03:53:01,479
ABOUT THE LAST SLIDE, TEAM

6256
03:53:01,479 --> 03:53:03,481
CENTER FOR EXCELLENCE ON AGINGS

6257
03:53:03,481 --> 03:53:05,116
AND METABOLISM, HALF LIFE,

6258
03:53:05,116 --> 03:53:05,650
MAXIMUM DAILY DOSE.

6259
03:53:05,650 --> 03:53:07,151
THERE ARE A LOT OF OTHER THINGS

6260
03:53:07,151 --> 03:53:08,219
BUT THESE ARE VERY TYPICAL

6261
03:53:08,219 --> 03:53:10,388
QUESTIONS THAT YOU WILL ASK

6262
03:53:10,388 --> 03:53:16,527
ESPECIALLY FOR GENERIC DRUG.

6263
03:53:16,527 --> 03:53:18,596
SO WE CALL THESE ATOMIC QUERIES.

6264
03:53:18,596 --> 03:53:22,634
I WILL SHOW YOU HOW WELL CHATGPT

6265
03:53:22,634 --> 03:53:23,935
DOES AT ANSWERING THESE

6266
03:53:23,935 --> 03:53:25,470
QUESTIONS, BUT EVEN IF WE COULD

6267
03:53:25,470 --> 03:53:28,740
BUILD A MODEL OF THAT, THE

6268
03:53:28,740 --> 03:53:30,174
REVIEWERS WANT TO ASK OF A PILE

6269
03:53:30,174 --> 03:53:31,743
OF DATA AND TEXT AND A LOT OF

6270
03:53:31,743 --> 03:53:33,978
QUESTIONS YOU COULD ASK OF A

6271
03:53:33,978 --> 03:53:35,880
DRUG LABEL SO WE'VE COME UP WITH

6272
03:53:35,880 --> 03:53:37,181
CLEVER WAYS TO DO THAT WHICH I

6273
03:53:37,181 --> 03:53:40,051
WILL SHOW YOU RIGHT NOW.

6274
03:53:40,051 --> 03:53:43,755
WHICH IS WE ARE USING THE

6275
03:53:43,755 --> 03:53:46,257
ABILITY OF CHATGPT, TO LOOK AT A

6276
03:53:46,257 --> 03:53:47,759
DOCUMENT, WE GIVE IT SAMPLE

6277
03:53:47,759 --> 03:53:50,828
QUESTIONS, THE DOCUMENT IS A

6278
03:53:50,828 --> 03:53:52,931
DRUG LABEL, WE GIVE IT 9 OR 10

6279
03:53:52,931 --> 03:53:55,066
QUESTIONINGS ABOUT INDICATION,

6280
03:53:55,066 --> 03:53:55,700
BOXED WARNING, BIOAVAILABILITY

6281
03:53:55,700 --> 03:53:59,737
AND THEN WE TELL IT GENERATE A

6282
03:53:59,737 --> 03:54:01,973
BUNCH OF QUERIES OF THIS LABEL

6283
03:54:01,973 --> 03:54:04,208
SIMILAR TO WHAT WE JUST SHOWED

6284
03:54:04,208 --> 03:54:06,511
YOU BASED ON CONTENT AND THIS IS

6285
03:54:06,511 --> 03:54:07,145
CALLED SELF-INSTRUCTION BECAUSE

6286
03:54:07,145 --> 03:54:08,646
IT'S GOING TO GENERATE THESE

6287
03:54:08,646 --> 03:54:09,847
QUERIES AND THEN WE'RE GOING TO

6288
03:54:09,847 --> 03:54:12,617
ASK IT TO ANSWER THE QUERY AND

6289
03:54:12,617 --> 03:54:13,685
JUST TO GIVE YOU THE BOTTOM LINE

6290
03:54:13,685 --> 03:54:17,922
AND THEN WE WILL ASK IT TO GRADE

6291
03:54:17,922 --> 03:54:18,389
ITS ANSWER.

6292
03:54:18,389 --> 03:54:19,891
AND BELIEVE IT OR NOT IT'S QUITE

6293
03:54:19,891 --> 03:54:20,992
GOOD AT ALL 3 OF THOSE.

6294
03:54:20,992 --> 03:54:23,494
SO FOR THE FIRST 1, HERE IS SOME

6295
03:54:23,494 --> 03:54:27,832
-- THESE ARE ALL QUESTIONS

6296
03:54:27,832 --> 03:54:32,537
CREATED BY CHATGPT BASED ON THE

6297
03:54:32,537 --> 03:54:32,770
LABELS.

6298
03:54:32,770 --> 03:54:33,905
THE THING OF THE LEFT THE COLON

6299
03:54:33,905 --> 03:54:35,306
IS OUR LABEL OF WHAT KIND OF

6300
03:54:35,306 --> 03:54:36,908
QUESTION IT IS SO THAT FIRST 1

6301
03:54:36,908 --> 03:54:39,644
WE SAID, THIS IS A QUESTION

6302
03:54:39,644 --> 03:54:40,378
ABOUT ADVERSE REACTIONS BUT

6303
03:54:40,378 --> 03:54:44,782
AFTER THE COLON THAT'S ALL FROM

6304
03:54:44,782 --> 03:54:50,588
CHATGPT, WHAT ARE THE ADVERSE

6305
03:54:50,588 --> 03:54:51,556
INTERACTIONS OF THE DRUG?

6306
03:54:51,556 --> 03:54:58,296
WHAT  OTHER DRUGS IRPT ACT WITH

6307
03:54:58,296 --> 03:54:59,230
THE DRUG?

6308
03:54:59,230 --> 03:55:00,698
CONTRA INDICATIONS WHAT ARE THE

6309
03:55:00,698 --> 03:55:06,104
CONDRAINDICATIONS OF THE DRUG? 

6310
03:55:06,104 --> 03:55:08,439
SO WE CREATED FROM MY STUDENT

6311
03:55:08,439 --> 03:55:11,476
BETTY, AND SHE CREATED THOSE

6312
03:55:11,476 --> 03:55:15,480
QUESTIONS AND SHE COMPARED THE

6313
03:55:15,480 --> 03:55:18,082
CHATGPT TO HER GOLD STANDARD

6314
03:55:18,082 --> 03:55:18,349
ANSWERS.

6315
03:55:18,349 --> 03:55:20,585
THEN SHE ASKED CHATGPT TO USE A

6316
03:55:20,585 --> 03:55:23,154
SIMILAR RUBRIC TO COMPARE HER

6317
03:55:23,154 --> 03:55:23,921
ANSWERS WITH ITS OWN ANSWERS AND

6318
03:55:23,921 --> 03:55:25,990
THIS IS WHERE I HAVE TO REMINE

6319
03:55:25,990 --> 03:55:28,292
THAT YOU CHATGPT IS NOT A HUMAN

6320
03:55:28,292 --> 03:55:29,494
AND HAS NO PRIDE IN ITS ANSWERS

6321
03:55:29,494 --> 03:55:31,295
AND MANY OF YOU MAY HAVE

6322
03:55:31,295 --> 03:55:32,296
EXPERIENCED THIS, BUT IT WILL

6323
03:55:32,296 --> 03:55:33,798
GIVE YOU AN ANSWER AND YOU CAN

6324
03:55:33,798 --> 03:55:36,200
SAY, DO YOU THINK THAT'S A GOOD

6325
03:55:36,200 --> 03:55:37,802
ANSWER AND IT'LL SAY NO, THAT'S

6326
03:55:37,802 --> 03:55:38,870
NOT A GOOD ANSWER.

6327
03:55:38,870 --> 03:55:40,438
SO WE TOOK ADVANTAGE OF THAT TO

6328
03:55:40,438 --> 03:55:41,572
SAY, HERE'S THE QUESTION YOU

6329
03:55:41,572 --> 03:55:42,640
CAME UP WITH, HERE'S THE ANSWER

6330
03:55:42,640 --> 03:55:45,610
YOU CAME UP WITH, IS THIS A GOOD

6331
03:55:45,610 --> 03:55:48,513
QUESTION, THAT'S THE RUBRIC.

6332
03:55:48,513 --> 03:55:49,113
CORRECT.

6333
03:55:49,113 --> 03:55:50,348
THE OUTPUT MATCHES WHAT BETTY

6334
03:55:50,348 --> 03:55:52,350
DID EXACTLY AND IT LEARNED WHAT

6335
03:55:52,350 --> 03:55:53,951
CORRECT IS AND NOW IT CAN DO

6336
03:55:53,951 --> 03:55:56,354
CORRECT FOR THINGS THAT BETTY

6337
03:55:56,354 --> 03:55:58,122
DIDN'T DO, EXTRA COME WHICH

6338
03:55:58,122 --> 03:55:59,390
MEANS IT CONTAINED THE ANSWER

6339
03:55:59,390 --> 03:56:01,426
BUT IT WAS TOO WORDY AND

6340
03:56:01,426 --> 03:56:05,196
CONTAINED A LOT OF EXTRA THINGS

6341
03:56:05,196 --> 03:56:06,864
WE DON'T NEED.

6342
03:56:06,864 --> 03:56:10,701
INCOMPLETE, WHICH IS OPPOSITE,

6343
03:56:10,701 --> 03:56:15,506
IT WAS PARTIALLY CORRECT, OR

6344
03:56:15,506 --> 03:56:18,309
INCORRECT, THE OUTPUT DOES NOT

6345
03:56:18,309 --> 03:56:20,178
MATCH THE GOLD STANDARD AT ALL.

6346
03:56:20,178 --> 03:56:23,014
SOW WHAT WE HAVE NOW IS HOW IT  

6347
03:56:23,014 --> 03:56:23,848
PERFORMS AT KUFULT ORDER OF

6348
03:56:23,848 --> 03:56:24,916
MICRONS ANSWER AND EVALUATION.

6349
03:56:24,916 --> 03:56:27,852
SO THE GREEN ARE CORRECT, BETTY

6350
03:56:27,852 --> 03:56:29,387
THINKS THEY'RE CORRECT, CHAT GPT

6351
03:56:29,387 --> 03:56:30,354
THINKS THEY'RE CORRECT AND THEY

6352
03:56:30,354 --> 03:56:36,761
ARE CORRECT AND YOU CAN SEE FOR

6353
03:56:36,761 --> 03:56:38,229
EXCRETION, RENAL, LIVER, GI,

6354
03:56:38,229 --> 03:56:40,465
FOOD EVENS PRETTY GOOD.

6355
03:56:40,465 --> 03:56:41,999
HALF LIFE PRETTY GOOD AND THAT'S

6356
03:56:41,999 --> 03:56:45,002
BECAUSE IT TIP KACCT WILY SAYS

6357
03:56:45,002 --> 03:56:48,973
THE HALF LIFE OF THIS DRUG IN A

6358
03:56:48,973 --> 03:56:52,410
HUMAN WITH HEALTHY NORMAL LIVERS

6359
03:56:52,410 --> 03:56:55,980
AND BLOOD AND IT'S GOOD AT THAT.

6360
03:56:55,980 --> 03:56:58,382
IT'S PARTIAL CREDIT ON THE REDS

6361
03:56:58,382 --> 03:56:59,884
AND ORANGES BUT I'M FOCUSING ON

6362
03:56:59,884 --> 03:57:02,186
THE GREENS, SO CAN YOU SEE ON

6363
03:57:02,186 --> 03:57:03,654
THE RIGHT SIDE, CHATGPT IS

6364
03:57:03,654 --> 03:57:05,323
PRETTY GOOD AND IF WE GENERATE

6365
03:57:05,323 --> 03:57:07,291
10S OF THOUSANDS OF EXAMPLES OF

6366
03:57:07,291 --> 03:57:10,895
THESE, WE HAVE SOME EXPECTATION

6367
03:57:10,895 --> 03:57:12,029
THAT OUR DISTILLED MODEL WILL BE

6368
03:57:12,029 --> 03:57:13,464
GOOD.

6369
03:57:13,464 --> 03:57:21,072
THE PROBLEM IS ON THE LEFT.

6370
03:57:21,072 --> 03:57:22,206
CHATGPT IS FIGURE THE OUT WHAT

6371
03:57:22,206 --> 03:57:23,841
TO DO WITH THE LEFT.

6372
03:57:23,841 --> 03:57:25,476
WE LOSE THE BOX, WE GET IT WRONG

6373
03:57:25,476 --> 03:57:27,044
AND WE'RE WORKING ON THAT AND

6374
03:57:27,044 --> 03:57:29,113
WE'RE WORKING ON, WHEN I SAY

6375
03:57:29,113 --> 03:57:31,082
WORKING, IT MEANS WE'RE WRITING

6376
03:57:31,082 --> 03:57:33,951
THE SPOKE OF PROMPTS TO THE LLM

6377
03:57:33,951 --> 03:57:40,758
TO HELP IT GET A BETTER ANSWER.

6378
03:57:40,758 --> 03:57:42,059
AND THEN BIOAVAILABILITY WAS

6379
03:57:42,059 --> 03:57:43,928
ACTUALLY SURPRISINGLY NOT VERY

6380
03:57:43,928 --> 03:57:44,128
GOOD.

6381
03:57:44,128 --> 03:57:46,264
SO THOSE ARE THE 1S WE LED IT

6382
03:57:46,264 --> 03:57:47,665
WORK ON BEFORE WE LET IT LOOSE.

6383
03:57:47,665 --> 03:57:50,768
I WANT TO SHOW YOU THIS BECAUSE

6384
03:57:50,768 --> 03:57:52,236
CHATGPT OUT OF THE BOX IS PRETTY

6385
03:57:52,236 --> 03:57:54,972
GOOD BUT IT NEEDS HELP WITH THE

6386
03:57:54,972 --> 03:57:57,175
NOW ONCE WE GET THAT IN SHAPE, I

6387
03:57:57,175 --> 03:57:58,442
CAN TELL YOU THE NEXT STEPS AND

6388
03:57:58,442 --> 03:58:02,213
I CAN BE QUIET, AND WE WILL

6389
03:58:02,213 --> 03:58:05,049
REFORM OUR AND WE WILL SCALE TO

6390
03:58:05,049 --> 03:58:06,617
THOUSANDS OF DRUG LABELS TO GET

6391
03:58:06,617 --> 03:58:13,057
TONS OF THOUSANDS OF QUERIES,

6392
03:58:13,057 --> 03:58:13,991
ANSWERS, AND EVALUATIONS EMPLOY

6393
03:58:13,991 --> 03:58:16,294
IT TURNS OUT WHEN IT DOES

6394
03:58:16,294 --> 03:58:18,229
POORLY, IT CAN BE USED IN

6395
03:58:18,229 --> 03:58:18,496
TRAINING.

6396
03:58:18,496 --> 03:58:20,264
IT'S RELATED TO PAT'S TALK,

6397
03:58:20,264 --> 03:58:21,432
NEGATIVE EXAMPLES BY THE WAI

6398
03:58:21,432 --> 03:58:24,268
THAT'S A BAD SORRY, THAT'S

6399
03:58:24,268 --> 03:58:25,736
USEFUL FOR TRAINING THE LLM.

6400
03:58:25,736 --> 03:58:27,738
SO WE WILL GIVE THAT TO THE

6401
03:58:27,738 --> 03:58:27,939
MODEL.

6402
03:58:27,939 --> 03:58:31,842
WE WILL AS I SAID FINE TUNE

6403
03:58:31,842 --> 03:58:32,610
LLAMA 3.1.

6404
03:58:32,610 --> 03:58:34,011
WHEN WILL MODIFY THE WEIGHTS, WE

6405
03:58:34,011 --> 03:58:36,614
WILL GIVE IT THE DOCUMS THAT THE

6406
03:58:36,614 --> 03:58:37,648
LABELS, THE QUESTIONS, THE

6407
03:58:37,648 --> 03:58:38,816
ANSWERS AND THE KIND OF

6408
03:58:38,816 --> 03:58:39,917
CONFIDENCE IN THE ANSWERS, AND

6409
03:58:39,917 --> 03:58:41,552
THEN WE WILL DO THE REAL, THIS

6410
03:58:41,552 --> 03:58:43,487
IS WHERE THE RUBBER WILL HIT THE

6411
03:58:43,487 --> 03:58:45,356
ROAD FOR A SUBSET OF A COUPLE

6412
03:58:45,356 --> 03:58:48,025
HUNDRED OF THESE, WE WILL GET

6413
03:58:48,025 --> 03:58:49,727
THROUGH OUR COLLABORATION WITH

6414
03:58:49,727 --> 03:58:51,896
THE FDA, FDA REVIEWERS TO

6415
03:58:51,896 --> 03:58:55,766
ACTUALLY GRID, THE RESULTS ON

6416
03:58:55,766 --> 03:58:59,370
NEW LABELS THAT HAVEN'T BEEN

6417
03:58:59,370 --> 03:59:01,772
SEEN BEFORE TO SEE HOW GOOD IT

6418
03:59:01,772 --> 03:59:01,939
IS.

6419
03:59:01,939 --> 03:59:03,474
AT THAT POINT, WE WILL PROBABLY

6420
03:59:03,474 --> 03:59:06,110
DELIVER IT TO FDA INTERNALLY FOR

6421
03:59:06,110 --> 03:59:08,312
THEM TO NOT EXPOSE IT TO DRUG

6422
03:59:08,312 --> 03:59:10,615
LABELS, BUT TO EXPOSE IT TO

6423
03:59:10,615 --> 03:59:11,315
REGULATORY SUBMISSIONS, AND THEY

6424
03:59:11,315 --> 03:59:13,451
WILL TELL US HOW WE DID, AND

6425
03:59:13,451 --> 03:59:14,919
THEN THEY'LL TELL US WHAT WE

6426
03:59:14,919 --> 03:59:17,488
HAVE TO FIX IN AN ITERATED

6427
03:59:17,488 --> 03:59:19,624
VERSION, SO, NO TIME SOON WILL

6428
03:59:19,624 --> 03:59:22,660
YOUR DRUGS BE APPROVED OR NOT

6429
03:59:22,660 --> 03:59:24,528
APPROVED BY AN LLM BUT VERY

6430
03:59:24,528 --> 03:59:27,398
EXCITING TO BE WORKING WITH OUR

6431
03:59:27,398 --> 03:59:29,433
COLLEAGUES IN REGULATORY SCIENCE

6432
03:59:29,433 --> 03:59:30,635
TO SEE HOW WE CAN -- MIGHT BE

6433
03:59:30,635 --> 03:59:33,271
ABLE TO HELP THAT PROCESS BE A

6434
03:59:33,271 --> 03:59:35,873
BIT MORE EFFICIENT AND PERHAPSIC

6435
03:59:35,873 --> 03:59:38,042
TAKEN--THEYING THE BEST OF THE

6436
03:59:38,042 --> 03:59:38,843
REVIEWERS HUMAN CAPABILITIES AND

6437
03:59:38,843 --> 03:59:45,683
THE BEST OF THE LLM'S

6438
03:59:45,683 --> 03:59:46,150
CAPABILITIES.

6439
03:59:46,150 --> 03:59:48,152
SO LET ME STOP THERE AND THE

6440
03:59:48,152 --> 03:59:51,389
OFFICE OF COLLABORATION AND

6441
03:59:51,389 --> 03:59:52,023
EVERYBODY WHO COLLABORATED AND I

6442
03:59:52,023 --> 03:59:59,263
THINK WE WILL HAVE QUESTIONS

6443
03:59:59,263 --> 03:59:59,830
LATER.

6444
03:59:59,830 --> 04:00:00,064
THANKS.

6445
04:00:00,064 --> 04:00:02,533
NTHAT WAS GREAT.

6446
04:00:02,533 --> 04:00:03,934
THAT WAS QUICK AND I THINK

6447
04:00:03,934 --> 04:00:05,469
THAT'S OKAY, SHOULD WE JUST MOVE

6448
04:00:05,469 --> 04:00:07,371
ON TO JOEL NOW I THINK SO

6449
04:00:07,371 --> 04:00:09,974
BECAUSE THEN WE CAN HAVE MORE

6450
04:00:09,974 --> 04:00:10,408
TIME FOR DISCUSSION.

6451
04:00:10,408 --> 04:00:12,743
SO JOEL, IF YOU'RE READY, I SEE

6452
04:00:12,743 --> 04:00:14,111
YOU MOVING YOUR CAMERA, THANK

6453
04:00:14,111 --> 04:00:15,746
YOU RUSS THAT WAS TERRIFIC.

6454
04:00:15,746 --> 04:00:19,984
I LOVE SORT OF THE BREDTH OF

6455
04:00:19,984 --> 04:00:23,654
WHAT WE'RE TALKING ABOUT HERE.

6456
04:00:23,654 --> 04:00:25,156
OKAY,IOLE I DON'T SEE YOUR FACE

6457
04:00:25,156 --> 04:00:26,991
BUT IS IT JUST ME, OH THERE YOU

6458
04:00:26,991 --> 04:00:31,495
ARE, YOU ARE SHARING YOUR

6459
04:00:31,495 --> 04:00:32,063
SCREEN, OKAY.

6460
04:00:32,063 --> 04:00:35,266
LET ME GO AHEAD AND INTRODUCE

6461
04:00:35,266 --> 04:00:38,402
YOU QUICKLY, SO JOEL KARPIAK, IS

6462
04:00:38,402 --> 04:00:41,806
THE DESIGN HEAD AT GSK, FOCUSING

6463
04:00:41,806 --> 04:00:43,507
ON COMPUTATIONAL CHEMIST RADIOY

6464
04:00:43,507 --> 04:00:44,709
AND PROTEIN DESIGN, WORKED ON

6465
04:00:44,709 --> 04:00:46,677
DRUG DISCOVERY, DONE A LOT WITH

6466
04:00:46,677 --> 04:00:48,779
GPC Rs WHICH IS REALLY COOL

6467
04:00:48,779 --> 04:00:51,148
AND I THINK IS NOW WORKING

6468
04:00:51,148 --> 04:00:54,452
PROBABLY PROBABLY WITH

6469
04:00:54,452 --> 04:00:55,019
INTEGRATING ARTIFICIAL

6470
04:00:55,019 --> 04:00:57,488
INTELLIGENCE FOR DRUG DORPHY AND

6471
04:00:57,488 --> 04:00:58,122
INDUSTRIAL APPLICATIONS, MAYBE,

6472
04:00:58,122 --> 04:01:00,424
BUT ANYWAY, TAKE IT AWAY.

6473
04:01:00,424 --> 04:01:00,825
>> YEAH THANKS.

6474
04:01:00,825 --> 04:01:06,397
A WIDE VARIETY OF THINGS LEADING

6475
04:01:06,397 --> 04:01:09,433
THE DATA AND PREDICTIVE SCIENCES

6476
04:01:09,433 --> 04:01:10,067
ORGANIZATION HAS BEEN

6477
04:01:10,067 --> 04:01:10,534
INTERESTING RECENTLY.

6478
04:01:10,534 --> 04:01:12,570
I WANT TO TALK ABOUT THE COST OF

6479
04:01:12,570 --> 04:01:14,238
A VARIETY OF THINGS AND HOW WE

6480
04:01:14,238 --> 04:01:16,107
CAN ALL WORK TOGETHER AS A TEAM

6481
04:01:16,107 --> 04:01:17,341
FOR TRANSLATION AND REFLECTING

6482
04:01:17,341 --> 04:01:20,311
ON WHAT TRANSLATION MEANS IN A

6483
04:01:20,311 --> 04:01:21,779
VARIETY OF DIFFERENT CONTEXT,

6484
04:01:21,779 --> 04:01:23,514
IT'S KIND OF IMPORTANT I THINK

6485
04:01:23,514 --> 04:01:26,317
AS WE'RE ACTUALLY IMPACTING HOW

6486
04:01:26,317 --> 04:01:27,351
WE ACCELERATE OR MAKE DRUG

6487
04:01:27,351 --> 04:01:28,686
DORPHY A BIT MOREI FICIENT.

6488
04:01:28,686 --> 04:01:29,854
SO JUST BEING ABLE TO TALK TO

6489
04:01:29,854 --> 04:01:32,390
EACH OTHER AND SHARE THE SAME

6490
04:01:32,390 --> 04:01:34,125
KIND OF VOCABULARY AROUND WHAT A

6491
04:01:34,125 --> 04:01:35,426
DATA SCIENTIST DOES, THE

6492
04:01:35,426 --> 04:01:37,728
ENGINEERING NEEDED FOR AN ACTUAL

6493
04:01:37,728 --> 04:01:40,164
COMPUTE CLUSTER, WHAT A

6494
04:01:40,164 --> 04:01:41,932
COMPUTATIONAL HAS FOR DESIGN OR

6495
04:01:41,932 --> 04:01:45,703
WET LAB ENABLED TO TEST IT, HOW

6496
04:01:45,703 --> 04:01:47,772
DO WE CONTRIBUTE TO EACH OTHER'S

6497
04:01:47,772 --> 04:01:49,073
DISCIPLINES, HOW DO WE IMPACT

6498
04:01:49,073 --> 04:01:49,874
AND COMMUNICATE WHAT WE'RE

6499
04:01:49,874 --> 04:01:52,309
ACTUALLY TRYING TO PREDICT, GET

6500
04:01:52,309 --> 04:01:53,077
THOSE SPRPTS DONE AND INFLUENCE

6501
04:01:53,077 --> 04:01:54,945
EACH OTHER TO WORK MORE

6502
04:01:54,945 --> 04:01:55,312
EFFICIENTLY.

6503
04:01:55,312 --> 04:01:56,514
EVEN SOMETHING VERY SIMPLE IN

6504
04:01:56,514 --> 04:01:59,383
TERMS OF HOW DID DNA CORRELATE

6505
04:01:59,383 --> 04:02:01,185
TO RNA WHICH TRANSLATES INTO

6506
04:02:01,185 --> 04:02:01,452
PROTEINS.

6507
04:02:01,452 --> 04:02:03,788
YOU KNOW PEOPLE CAN TALK ABOUT

6508
04:02:03,788 --> 04:02:07,258
YEENS AS TARGETS, VERY DIFFERENT

6509
04:02:07,258 --> 04:02:09,026
THAN AND THEREFORE A COMMON GOAL

6510
04:02:09,026 --> 04:02:11,996
AND ALSO TRANSLATION TERMS OF

6511
04:02:11,996 --> 04:02:13,397
PROTEIN TARGET AND THEN FINDING

6512
04:02:13,397 --> 04:02:15,699
A BINDER TO THAT IS DIFFERENT

6513
04:02:15,699 --> 04:02:17,134
THAN BEING EFFICACIOUS IN A

6514
04:02:17,134 --> 04:02:18,936
MODEL SYSTEM AND CERTAINLY

6515
04:02:18,936 --> 04:02:19,837
DIFFERENT THAN EFFICACIOUS IN

6516
04:02:19,837 --> 04:02:21,806
THE HUMAN SYSTEM, SO HOW DO WE

6517
04:02:21,806 --> 04:02:23,307
LINE UP MODELING TO IMPACT THESE

6518
04:02:23,307 --> 04:02:23,941
IN ANENTIOUS FICIENT WAY.

6519
04:02:23,941 --> 04:02:25,543
AND I SHOULD SAY THIS SLIDE IS

6520
04:02:25,543 --> 04:02:28,479
REALLY MESSY ON PURPOSE NOT JUST

6521
04:02:28,479 --> 04:02:29,613
BECAUSE I WEAR POWER POINT

6522
04:02:29,613 --> 04:02:32,016
SKILLS BUT BECAUSE THIS IS A

6523
04:02:32,016 --> 04:02:33,884
REALLY MISSING AREA THAT NOT

6524
04:02:33,884 --> 04:02:35,753
ANYBODY HAS RIGHT.

6525
04:02:35,753 --> 04:02:37,054
ALTHOUGH COMPUTERS THROUGHOUT

6526
04:02:37,054 --> 04:02:37,855
THIS ENTIRE SESSION CAN SPEAK

6527
04:02:37,855 --> 04:02:39,123
ALL OF THESE DIFFERENT

6528
04:02:39,123 --> 04:02:41,926
LANGUAGES, WE DON'T YET HAVE

6529
04:02:41,926 --> 04:02:43,093
COMPLETELY AUTONOMOUS AGENTS

6530
04:02:43,093 --> 04:02:43,861
DOING EVERYTHING POSSIBLE,

6531
04:02:43,861 --> 04:02:48,232
HUMANS ARE STILL IN THE LOOP,

6532
04:02:48,232 --> 04:02:50,768
STILL NEED TO MAKE DECISIONS AND

6533
04:02:50,768 --> 04:02:52,570
PRIORITIZATIONS AND HOW DO WE CO

6534
04:02:52,570 --> 04:02:52,903
THAT.

6535
04:02:52,903 --> 04:02:58,108
SOW WHEN  -- SO WHEN WE START OE

6536
04:02:58,108 --> 04:03:06,183
HAVE INFORMAT 8  -- INFORMATICI,

6537
04:03:06,183 --> 04:03:07,952
AND WE DON'T HAVE ANY KNOWLEDGE

6538
04:03:07,952 --> 04:03:08,986
GAPS OR WHITE SPACE BECAUSE

6539
04:03:08,986 --> 04:03:11,489
WE'RE DOING THINGS IN A MODULAR

6540
04:03:11,489 --> 04:03:14,225
WAY BUT TRYING TO HAVE THE

6541
04:03:14,225 --> 04:03:15,860
SEAMLESS DATA FLOW SO THAT WE'RE

6542
04:03:15,860 --> 04:03:17,595
GENERATING DATA AND MAKING

6543
04:03:17,595 --> 04:03:18,496
PREDICTIONS THAT IMPACT AND

6544
04:03:18,496 --> 04:03:20,798
LOOKING AT IT FROM A GLOBAL MIND

6545
04:03:20,798 --> 04:03:21,999
SET WHERE FOUNDATIONAL MIND SET

6546
04:03:21,999 --> 04:03:24,101
BUILDING UP THOSE DATA SETS AS

6547
04:03:24,101 --> 04:03:28,939
WE'VE TALKED ABOUT PREVIOUSLY

6548
04:03:28,939 --> 04:03:29,740
THROUGHOUT THIS WORKSHOP.

6549
04:03:29,740 --> 04:03:37,414
IMH REALLY THIS IS SEAMLESS

6550
04:03:37,414 --> 04:03:38,716
INTEGRATION BETWEEN BESPOKE,

6551
04:03:38,716 --> 04:03:39,350
DATA SCIENTISTS, AUTOMATION,

6552
04:03:39,350 --> 04:03:41,785
DOING IT AT SCALE AND THE

6553
04:03:41,785 --> 04:03:44,221
QUALITY TO MAKE PREDICTIONS AND

6554
04:03:44,221 --> 04:03:47,258
ALL OF THIS HAS UNDERPINNINGS OF

6555
04:03:47,258 --> 04:03:48,025
TRANSPARENCY AND TRUST WITHIN

6556
04:03:48,025 --> 04:03:49,894
THE TEAMS SO WE KNOW IT'S GOING

6557
04:03:49,894 --> 04:03:50,828
ON, PEOPLE UNDERSTAND WHERE

6558
04:03:50,828 --> 04:03:53,130
PREDICS ARE COMING FROM AND

6559
04:03:53,130 --> 04:03:54,565
INTERPRET THEM IN THE BEST WAY

6560
04:03:54,565 --> 04:03:56,567
TO MAKE THE BEST KINDS OF

6561
04:03:56,567 --> 04:03:57,101
DECISIONS.

6562
04:03:57,101 --> 04:03:58,602
SO YOU'VE HEARD A LOT ABOUT

6563
04:03:58,602 --> 04:03:59,770
DIFFERENT KINDS OF MACHINE

6564
04:03:59,770 --> 04:04:05,442
LEARNING AND DRUG DISCOVERY AND

6565
04:04:05,442 --> 04:04:08,946
THINGS LIKE THIS, THERE'S A WIDE

6566
04:04:08,946 --> 04:04:09,747
VARIETY OF COMPUTATIONAL IMPACT.

6567
04:04:09,747 --> 04:04:11,749
SO I WANT TO FOCUS ON WHAT

6568
04:04:11,749 --> 04:04:13,117
HAPPENS IN SMALL MOLECULE

6569
04:04:13,117 --> 04:04:15,019
DESIGN, TO PUT ALL THESE THINGS

6570
04:04:15,019 --> 04:04:15,286
TOGETHER.

6571
04:04:15,286 --> 04:04:16,620
SO THAT'S A ROUND GENERATING

6572
04:04:16,620 --> 04:04:18,422
DATA, HOW DO WE DO IT

6573
04:04:18,422 --> 04:04:20,257
INTENTIONALLY, HOW DO WE

6574
04:04:20,257 --> 04:04:22,293
INKERPERATE IT INTO MODELS AND

6575
04:04:22,293 --> 04:04:23,460
DEPLOY THOSE IN RESPONSIBLE WAY

6576
04:04:23,460 --> 04:04:25,763
FOR PEOPLE TO BE MABEL TO MAKE

6577
04:04:25,763 --> 04:04:29,967
GOOD DECISIONS AND A BIT OF A

6578
04:04:29,967 --> 04:04:32,136
SAND BOX ENVIRONMENT AND

6579
04:04:32,136 --> 04:04:33,270
INTERPRET WHAT WE'RE TRYING TO

6580
04:04:33,270 --> 04:04:34,371
SAY INTO THE EXPERIMENTAL

6581
04:04:34,371 --> 04:04:35,639
PRABLGHTIS EMPLOY SO A LOT OF

6582
04:04:35,639 --> 04:04:42,813
THIS GOES INTO A BIT OF A

6583
04:04:42,813 --> 04:04:45,282
FRAMEWORK AT GSK,ERB HAS THEIR

6584
04:04:45,282 --> 04:04:45,950
SHOP FOR.

6585
04:04:45,950 --> 04:04:46,850
SO STARTING WITH THIS AND FOR

6586
04:04:46,850 --> 04:04:49,286
THE REST OF THE TALK, IN

6587
04:04:49,286 --> 04:04:50,554
PRINCIPLE THERE'S REALLY NOTHING

6588
04:04:50,554 --> 04:04:51,388
DIFFERENT BETWEEN SMALL

6589
04:04:51,388 --> 04:04:53,524
MOLECULES AND LARGE MOLECULES OR

6590
04:04:53,524 --> 04:04:54,658
OLIGOS OR ANYTHING ELSE WE MIGHT

6591
04:04:54,658 --> 04:04:55,993
BE INTERESTED INY DID SIGNING,

6592
04:04:55,993 --> 04:04:57,127
JUST HAPPENED TO SELECT THIS TO

6593
04:04:57,127 --> 04:04:58,929
GET INTO THE BIT OF THE

6594
04:04:58,929 --> 04:05:00,431
NITTY-GRITTY ABOUT HOW TEAMS

6595
04:05:00,431 --> 04:05:00,731
WORK.

6596
04:05:00,731 --> 04:05:02,333
SO HERE WE'VE GOT SORT OF ALL

6597
04:05:02,333 --> 04:05:03,934
THE COMPONENTS WE HEARD ABOUT

6598
04:05:03,934 --> 04:05:04,435
ALREADY THROUGHOUT THESE

6599
04:05:04,435 --> 04:05:06,870
SESSIONS IN EMERGING ITS OF

6600
04:05:06,870 --> 04:05:09,640
MOLECULE GENERATORS, MODEL

6601
04:05:09,640 --> 04:05:11,775
CLICKS, THAT HAVE A VARIETY OF

6602
04:05:11,775 --> 04:05:13,110
DIFFERENT AMOUNT OF DATA BEHIND

6603
04:05:13,110 --> 04:05:15,179
THEM TO ENABLE THE QUALITY MODEL

6604
04:05:15,179 --> 04:05:18,782
BUILDING AND ALSO SELECTION OR

6605
04:05:18,782 --> 04:05:19,350
ULTIMATELY PRIORITIZATION OF

6606
04:05:19,350 --> 04:05:20,784
MOLECULES TO MAKE AND HOW DO WE

6607
04:05:20,784 --> 04:05:24,088
DO THAT ON AUTOMATED CHEMICAL

6608
04:05:24,088 --> 04:05:25,389
PLATFORMS AND THE EMPHASIS IN A

6609
04:05:25,389 --> 04:05:27,157
WAY THAT REALLY DRIVES THE

6610
04:05:27,157 --> 04:05:29,126
PROGRAM FORWARD AND WITH THE

6611
04:05:29,126 --> 04:05:30,928
MAXIMAL AMOUNT OF IMPACT AND

6612
04:05:30,928 --> 04:05:31,762
GENERATEDDA THE EACH STAGE.

6613
04:05:31,762 --> 04:05:33,664
SO I WANT TO KIND OF GIVE A

6614
04:05:33,664 --> 04:05:35,532
COUPLE OF EXAMPLES AROUND WHAT

6615
04:05:35,532 --> 04:05:36,667
THIS MIGHT LOOK LIKE.

6616
04:05:36,667 --> 04:05:37,801
SO THE VARIETY OF DIFFERENT

6617
04:05:37,801 --> 04:05:43,741
TYPES OF DATA ARE GOING TO BE

6618
04:05:43,741 --> 04:05:46,010
IMPORTANT FOR THE KINDS OF

6619
04:05:46,010 --> 04:05:48,912
MODELSY WANT TO BUILD BUT

6620
04:05:48,912 --> 04:05:50,581
ON-TARGET, OFF-TARGET AND HOW DO

6621
04:05:50,581 --> 04:05:52,116
WE DEPLOY THOSE FOR BETTER

6622
04:05:52,116 --> 04:05:52,916
DECISIONINGS.

6623
04:05:52,916 --> 04:06:00,124
IT COULD BE DATA IMAGES, OR IT

6624
04:06:00,124 --> 04:06:02,626
COULD BE ANYTHING, SO HAVING A

6625
04:06:02,626 --> 04:06:05,596
FLEXIBLE SYSTEM FOR MODEL AND

6626
04:06:05,596 --> 04:06:07,398
DEPLOYMENT IS IMPORTANT BUT

6627
04:06:07,398 --> 04:06:08,365
STANDARDIZATION ABOUT HOW WE

6628
04:06:08,365 --> 04:06:10,200
BUILD THOSE MODELS AND BEING

6629
04:06:10,200 --> 04:06:12,436
ABLE TO SELECT THE VAST 1S THAT

6630
04:06:12,436 --> 04:06:13,704
ARE GBLIZABLE GOING FORWARD.

6631
04:06:13,704 --> 04:06:23,147
AND SO, I JUST WANT TO MAKE

6632
04:06:23,147 --> 04:06:31,121
ANOTHER COMPARISON TO A BIT, --

6633
04:06:31,121 --> 04:06:33,757
FROM A VARIETY OF SCREENING

6634
04:06:33,757 --> 04:06:36,093
CASCADES WHICH IS SIMILAR ACROSS

6635
04:06:36,093 --> 04:06:38,796
THE INDUSTRY, WHAT'S PUBLISH IS

6636
04:06:38,796 --> 04:06:39,863
IT'S TYPICALLY POSITIVE SO BEING

6637
04:06:39,863 --> 04:06:42,533
ABLE TO ESTABLISH CAREFUL 3

6638
04:06:42,533 --> 04:06:43,634
CURATION ESTABLISH BALANCE DATA

6639
04:06:43,634 --> 04:06:46,403
SETS FOR MODEL BUILDING IS

6640
04:06:46,403 --> 04:06:47,371
PRETTY IMPORTANT AND AGAIN I

6641
04:06:47,371 --> 04:06:48,439
WANT TO EMPHASIZE THE IMPORTANCE

6642
04:06:48,439 --> 04:06:50,140
OF BEING ABLE TO INCORPORATE

6643
04:06:50,140 --> 04:06:53,444
THAT OPEN SOURCE DATA GOING

6644
04:06:53,444 --> 04:06:53,677
FORWARD.

6645
04:06:53,677 --> 04:06:56,280
SO, JUST SOME THINGS TO TAKE

6646
04:06:56,280 --> 04:06:57,648
INTO CONSIDERATION AGAIN BECAUSE

6647
04:06:57,648 --> 04:06:59,249
THIS IS A THEME.

6648
04:06:59,249 --> 04:07:01,819
IN TERMS OF EVALUATING

6649
04:07:01,819 --> 04:07:03,587
EXPERIMENTAL DATA FOR MODEL

6650
04:07:03,587 --> 04:07:05,622
BUILDING, THIS REQUIRES REALLY

6651
04:07:05,622 --> 04:07:06,824
CLOSE COLLABORATION IN

6652
04:07:06,824 --> 04:07:09,626
COMMUNICATION WITH EXPERIMENTAL

6653
04:07:09,626 --> 04:07:11,729
SCIENCES TO MAKE SURE YOU'RE

6654
04:07:11,729 --> 04:07:12,429
INTERPRETING EXPERIMENTAL DATA

6655
04:07:12,429 --> 04:07:14,164
TO BUILD THE BEST KINDS OF

6656
04:07:14,164 --> 04:07:15,599
MODELS, SO WITHIN THE

6657
04:07:15,599 --> 04:07:18,235
EXPERIMENTAL ASSAYS, IT COULD BE

6658
04:07:18,235 --> 04:07:19,236
[INDISCERNIBLE] MEASURE WANTS

6659
04:07:19,236 --> 04:07:21,138
WHICH IMPACT THINGS, ERROR

6660
04:07:21,138 --> 04:07:22,072
PREDICTION, IMBALANCED DAILY

6661
04:07:22,072 --> 04:07:24,241
BASIS AT SETY DID PENDING  ON

6662
04:07:24,241 --> 04:07:30,147
THE RESULTS, AND THE FACT THAT

6663
04:07:30,147 --> 04:07:30,881
YOU'RE SIMPLIFYING THE

6664
04:07:30,881 --> 04:07:31,749
PRESENTATIONS AND HOW YOU BUILD

6665
04:07:31,749 --> 04:07:33,217
THE MODEL AND THE QUALITY OF IT.

6666
04:07:33,217 --> 04:07:35,652
THE ABILITY OF THE MODEL TO

6667
04:07:35,652 --> 04:07:37,454
EXTRAPOLATE BASED ON DATA AND

6668
04:07:37,454 --> 04:07:37,788
TRAINING SET.

6669
04:07:37,788 --> 04:07:38,388
ULTIMATE UNCERTAINTY IN THE

6670
04:07:38,388 --> 04:07:39,857
MODEL AND HOW WE QUANTIFY THAT

6671
04:07:39,857 --> 04:07:42,693
AND THEN ALSO BUILD AN

6672
04:07:42,693 --> 04:07:44,628
INTERPRETIVE METHOD IF WE CAN

6673
04:07:44,628 --> 04:07:47,364
FOR EXPERT ANDS NONEXPERTS TO

6674
04:07:47,364 --> 04:07:48,866
MAKE DECISIONS OFF OF AND

6675
04:07:48,866 --> 04:07:49,767
HAPPENED WHAT'S GOING ON UNDER

6676
04:07:49,767 --> 04:07:50,701
THE HOOD.

6677
04:07:50,701 --> 04:07:52,603
AND OVER TIME SURPRISINGLY AS

6678
04:07:52,603 --> 04:07:59,109
CHEMIST RADIOY CHANGES THE

6679
04:07:59,109 --> 04:08:00,410
PROGRAM EVOLVES THE ABILITY FOR

6680
04:08:00,410 --> 04:08:02,479
ANY KIND OF SCAFFOLD OR GROUPS

6681
04:08:02,479 --> 04:08:05,649
WILL SCRIEWBT KNCHTLY SO THERE'S

6682
04:08:05,649 --> 04:08:07,050
A DEPENDENCY IN ORDER TO

6683
04:08:07,050 --> 04:08:09,686
GENERALIZE, MISSING DATA SETS,

6684
04:08:09,686 --> 04:08:10,487
SPARKS MATRIXES, TONS OF

6685
04:08:10,487 --> 04:08:11,588
PROBLEMS YOU HAVE TO TAKE INTO

6686
04:08:11,588 --> 04:08:12,523
CONSIDERATION WHEN WE LOOK AT IT

6687
04:08:12,523 --> 04:08:15,526
TO DO THIS IN A RIGOROUS WAY.

6688
04:08:15,526 --> 04:08:17,828
SO BEING ABLE TO THEN BOTH AT A

6689
04:08:17,828 --> 04:08:19,229
GLOBAL LEVEL, BUILD DATA SETS

6690
04:08:19,229 --> 04:08:20,230
INTENTIONALLY TO BE ABLE TO

6691
04:08:20,230 --> 04:08:22,599
BUILD THE BEST KIND OF MODELS

6692
04:08:22,599 --> 04:08:23,700
THAT EXTRAPOLATE ACROSS CHEMICAL

6693
04:08:23,700 --> 04:08:25,836
SPACE WHILE AT THE SAME TIME

6694
04:08:25,836 --> 04:08:27,037
HAPPENING THE LIMITATIONS ON A

6695
04:08:27,037 --> 04:08:29,273
POTENTIAL SERIES THAT YOU'RE

6696
04:08:29,273 --> 04:08:30,340
WORKING ON AND THEN NEED TO

6697
04:08:30,340 --> 04:08:32,209
BUILD A MODEL TO FILL THE 1

6698
04:08:32,209 --> 04:08:34,478
SPACE AND ALSO COMPUTE IT TO

6699
04:08:34,478 --> 04:08:36,814
NATURAL PROJECT THAT YOU'RE

6700
04:08:36,814 --> 04:08:38,015
WORKING ON IN ORDER TO JUST

6701
04:08:38,015 --> 04:08:38,982
DRIVE IT.

6702
04:08:38,982 --> 04:08:41,118
SO BEDDING WHETHER OR NOT

6703
04:08:41,118 --> 04:08:42,286
THERE'S GLOBAL MODELS ARE

6704
04:08:42,286 --> 04:08:43,687
SUITABLE, IF YOU NEED TO BUILD A

6705
04:08:43,687 --> 04:08:46,089
LOCAL MODEL AND PUSH IT MORE, OR

6706
04:08:46,089 --> 04:08:48,392
IF YOU CAN GENERATE DATA IN

6707
04:08:48,392 --> 04:08:49,760
ORDER TO MAKE BOTH BETTER AT THE

6708
04:08:49,760 --> 04:08:52,930
SAME TIME AND WHAT ARE THE

6709
04:08:52,930 --> 04:08:53,597
STRATEGIES IN ORDER TO DO THAT

6710
04:08:53,597 --> 04:08:57,701
EMPLOY I SHOULD SAY, JUST TO

6711
04:08:57,701 --> 04:08:59,837
GIVE A SCALE ON THE KIND OF DATA

6712
04:08:59,837 --> 04:09:01,371
AND MODELS NECESSARY FOR DRIVING

6713
04:09:01,371 --> 04:09:02,773
THESE KINDS OF PROJECTS FORWARD

6714
04:09:02,773 --> 04:09:05,175
TO GIVE AN EXAMPLE HERE FOR

6715
04:09:05,175 --> 04:09:06,777
OFFTARGET ASSAY PANEL, LOTS OF

6716
04:09:06,777 --> 04:09:09,146
NASTY THINGS YOU WANT TO AVOID

6717
04:09:09,146 --> 04:09:10,480
FOR VARIOUS VENE REASONS AND

6718
04:09:10,480 --> 04:09:11,048
COUNSEL STREAM CENTER FOR

6719
04:09:11,048 --> 04:09:12,249
EXCELLENCE ON AGINGS, HERE WE'RE

6720
04:09:12,249 --> 04:09:14,051
SHOWING A SAMPLING OF 98 MODELS

6721
04:09:14,051 --> 04:09:16,954
BASED ON A PANEL OF TARGETS,

6722
04:09:16,954 --> 04:09:18,355
THAT'S PRETTY STANDARD.

6723
04:09:18,355 --> 04:09:20,490
AND IN SHOWS DIFFERENT LEVELS OF

6724
04:09:20,490 --> 04:09:22,059
MODEL CALLING BASED ON THE DATA

6725
04:09:22,059 --> 04:09:23,327
THAT EXISTS AND WHETHER OR NOT

6726
04:09:23,327 --> 04:09:25,329
WE CAN IMPROVE THAT WITH THE

6727
04:09:25,329 --> 04:09:26,530
INGTENSIONAL DATA GENERATION

6728
04:09:26,530 --> 04:09:26,864
GOING FORWARD.

6729
04:09:26,864 --> 04:09:28,232
SO AGAIN JUST TO GIVE A SENSE OF

6730
04:09:28,232 --> 04:09:31,635
SCALE OF WHAT NEEDS TO BE DONE

6731
04:09:31,635 --> 04:09:33,170
IN ORDER TO DESIGN A GOOD

6732
04:09:33,170 --> 04:09:34,838
MOLECULE AND THESE SORT OF DATA

6733
04:09:34,838 --> 04:09:39,009
GAPS WE HAVE IN AND AROUND THAT.

6734
04:09:39,009 --> 04:09:41,211
AND GETTING BACK TO TEAM WORK

6735
04:09:41,211 --> 04:09:42,246
AND COLLABORATION, TALK ABOUT

6736
04:09:42,246 --> 04:09:44,014
WORKING WITH THE

6737
04:09:44,014 --> 04:09:45,015
EXPERIMENTALISTS IN ORDER TO

6738
04:09:45,015 --> 04:09:47,084
INTERPRET THE DATA CORRECTLY FOR

6739
04:09:47,084 --> 04:09:49,419
MODEL BUILDING NOW IT'S ABOUT

6740
04:09:49,419 --> 04:09:50,621
INTERPRETING MODEL BUILDING

6741
04:09:50,621 --> 04:09:52,656
CORRECTLY AND DEPLOYING THOSE

6742
04:09:52,656 --> 04:09:54,725
MODELS FOR EXPERIMENTALISTS OR

6743
04:09:54,725 --> 04:09:55,359
CHEMISTS WHOMEVER ON THE OTHER

6744
04:09:55,359 --> 04:09:59,329
SIDE TO BE ABLE TO USE

6745
04:09:59,329 --> 04:09:59,863
APPROPRIATELY.

6746
04:09:59,863 --> 04:10:01,698
UNDERSUPER VISION AND SAND BOX,

6747
04:10:01,698 --> 04:10:03,667
DESIGN AND HOW DO WE POOL ALL OF

6748
04:10:03,667 --> 04:10:06,236
THOSE IDEAS TOGETHER TO MAKE THE

6749
04:10:06,236 --> 04:10:06,603
BEST DECISIONS.

6750
04:10:06,603 --> 04:10:08,238
SO 1 POTENTIAL WAY TO DO THIS IS

6751
04:10:08,238 --> 04:10:09,273
THROUGH A SYSTEM FOR INSTANCE

6752
04:10:09,273 --> 04:10:11,642
THIS IS THE COMMERCIAL 1 WITH

6753
04:10:11,642 --> 04:10:14,645
THE BY DESIGN, WHERE IT'S A

6754
04:10:14,645 --> 04:10:16,880
CLOUD BASED SYSTEM, CLOUD BASED

6755
04:10:16,880 --> 04:10:18,115
PLATFORM DESIGN THAT'S

6756
04:10:18,115 --> 04:10:19,616
ACCESSIBLE, YOU CAN DEPLOY

6757
04:10:19,616 --> 04:10:21,385
MODELS THROUGH IT, GET IMMEDIATE

6758
04:10:21,385 --> 04:10:22,586
FEEDBACK ON DIFFERENT DESIGN

6759
04:10:22,586 --> 04:10:23,220
IDEAS AND INCORPORATE ANYTHING

6760
04:10:23,220 --> 04:10:24,988
ELSE THAT YOU WANT TO IN ON

6761
04:10:24,988 --> 04:10:26,590
ORDER TO ACTUALLY PUSH THESE

6762
04:10:26,590 --> 04:10:29,192
KINDS OF PROJECTS ALONG.

6763
04:10:29,192 --> 04:10:31,128
SO FOR INSTANCE YOU CAN BUILD

6764
04:10:31,128 --> 04:10:32,362
CUSTOM TEMPLATES FOR USERS TO BE

6765
04:10:32,362 --> 04:10:35,198
ABLE TO LOOK AT DATA IN A

6766
04:10:35,198 --> 04:10:36,800
CERTAIN WAY AND MAKE SURE IT'S

6767
04:10:36,800 --> 04:10:38,635
INTERPRETED IN A CERTAIN WAY AND

6768
04:10:38,635 --> 04:10:40,604
INCORPORATE ALL OF THOSE OTHER

6769
04:10:40,604 --> 04:10:42,139
KIND OF END POINTS IN HERE THAT

6770
04:10:42,139 --> 04:10:42,673
MIGHT BE IMPORTANT.

6771
04:10:42,673 --> 04:10:44,675
SO NOT JUST A PREDICTION BUT THE

6772
04:10:44,675 --> 04:10:46,076
UNCERTAINTY RELATED TO IT, HOW

6773
04:10:46,076 --> 04:10:48,879
IT SHOULD BE INTERPRETED AND THE

6774
04:10:48,879 --> 04:10:49,513
APPROPRIATE VIECIALIZATIONS IF

6775
04:10:49,513 --> 04:10:52,082
NECESSARY IN ORDER TO ACTUALLY

6776
04:10:52,082 --> 04:10:52,482
PUSH THINGS ALONG.

6777
04:10:52,482 --> 04:10:53,884
AND DEPENDING ON THE THINGS YOU

6778
04:10:53,884 --> 04:10:55,786
MIGHT BE INTERESTED IN, THERE

6779
04:10:55,786 --> 04:10:57,921
WOULD BE A PARTICULAR VIEW FOR

6780
04:10:57,921 --> 04:11:00,824
SAFETY OR OFFTARGET EFFECTS,

6781
04:11:00,824 --> 04:11:02,292
ONTARGET POTENCY ANYTHING YOU

6782
04:11:02,292 --> 04:11:03,560
MIGHT BE INTERESTED IN CAN BE

6783
04:11:03,560 --> 04:11:04,227
DEPLOYED HERE.

6784
04:11:04,227 --> 04:11:14,504
SO THIS IS REALLY A WAY TO HAVE

6785
04:11:14,504 --> 04:11:16,440
TRANSPARENCY ESPECIALLY WITH

6786
04:11:16,440 --> 04:11:17,407
COMPUTATIONALLISTS AND HOW THEY

6787
04:11:17,407 --> 04:11:18,709
MAKE DECISIONS TO DESIGN THE

6788
04:11:18,709 --> 04:11:19,876
TEST CYCLE AND AGAIN JUST TO

6789
04:11:19,876 --> 04:11:22,913
GIVE A SENSE OF SCALE, SO, ON

6790
04:11:22,913 --> 04:11:25,449
EACH PROJECT WE HAVE OVER 150

6791
04:11:25,449 --> 04:11:27,150
GLOBAL MODELS THAT ARE

6792
04:11:27,150 --> 04:11:29,119
CONTINUOUSLY REBUILT THAT ARE OF

6793
04:11:29,119 --> 04:11:30,220
VARIOUS DIFFERENT PROPERS THAT

6794
04:11:30,220 --> 04:11:31,288
WOULD BE INTERESTING OR

6795
04:11:31,288 --> 04:11:32,789
IMPORTANT FOR ACTUALLY DRIVING

6796
04:11:32,789 --> 04:11:35,292
THE PROGECS FORWARD IN THAT

6797
04:11:35,292 --> 04:11:38,061
PARTICULAR SERIES AND THEN

6798
04:11:38,061 --> 04:11:39,463
UNSURPRISINGLY, MANY CUSTOM

6799
04:11:39,463 --> 04:11:40,297
MODELS DEPENDING ON THE ACTUAL

6800
04:11:40,297 --> 04:11:41,832
SERIES OR THE TARGET SPECIFIC

6801
04:11:41,832 --> 04:11:43,166
THAT CAN BE DEPLOYED THROUGH

6802
04:11:43,166 --> 04:11:45,068
HERE AND THIS IS ALWAYS A

6803
04:11:45,068 --> 04:11:47,437
COLLABORATIVE EFFORT FOR

6804
04:11:47,437 --> 04:11:48,939
INITANCE WITH DMPK WHERE OTHER

6805
04:11:48,939 --> 04:11:50,707
ASSAY SCIENTISTS ARE ABLE TO DO

6806
04:11:50,707 --> 04:11:51,074
THIS.

6807
04:11:51,074 --> 04:11:52,442
SO AGAIN, MANY COOKS IN THE

6808
04:11:52,442 --> 04:11:55,278
KITCHEN IN ORDER TO TRY TO DO

6809
04:11:55,278 --> 04:11:58,515
THIS, BUT THE CLOSE

6810
04:11:58,515 --> 04:11:59,549
COLLABORATION BETWEEN THE

6811
04:11:59,549 --> 04:12:00,617
DIFFERENT FUNCTIONS AND THE

6812
04:12:00,617 --> 04:12:02,886
TRANSPARENCY AND THE TRUST THERE

6813
04:12:02,886 --> 04:12:03,754
IS ABSOLUTELY ESSENTIAL FROM THE

6814
04:12:03,754 --> 04:12:06,289
BEGINNING SO WE TRY TO DO THAT

6815
04:12:06,289 --> 04:12:07,824
BOTHOT EXPERIMENTAL AND ON THE

6816
04:12:07,824 --> 04:12:08,825
COMPUTATIONAL SIDE.

6817
04:12:08,825 --> 04:12:12,229
AGAIN, ANOTHER ASPECT OF

6818
04:12:12,229 --> 04:12:13,130
TRANSPARENCY IS AUTOMATED MODEL

6819
04:12:13,130 --> 04:12:15,065
TRIP FOR AT LEAST THESE TYPES OF

6820
04:12:15,065 --> 04:12:16,466
GLOBAL ASSAYS OR GLOBAL

6821
04:12:16,466 --> 04:12:18,435
PROPERTIES THAT WE MIGHT BE

6822
04:12:18,435 --> 04:12:19,369
INTERESTED IN, ANYBODY AT ANY

6823
04:12:19,369 --> 04:12:23,140
TIME CAN LOOK AT HOW THAT MODEL

6824
04:12:23,140 --> 04:12:25,308
IS   PERFORMING WITH AN

6825
04:12:25,308 --> 04:12:27,944
AUTOMATED SYSTEM FLAGGING IF IT

6826
04:12:27,944 --> 04:12:29,646
STARTS TO GET MORE COMPOUNDS

6827
04:12:29,646 --> 04:12:31,281
MADE AND SUGGIEST A THAT THE

6828
04:12:31,281 --> 04:12:32,716
MODEL NEEDS TO BE REMADE OR WE

6829
04:12:32,716 --> 04:12:34,418
CAN CONSIDER AN ACTIVE LEARNING

6830
04:12:34,418 --> 04:12:36,153
STRATEGY TO MAKE IT BETTER.

6831
04:12:36,153 --> 04:12:38,055
BUT AGAIN, IN THE INTEREST OF

6832
04:12:38,055 --> 04:12:38,922
TRANSPARENCY, THESE ARE THINGS

6833
04:12:38,922 --> 04:12:41,258
THAT ANYBODY CAN LOOK @ ANY 1

6834
04:12:41,258 --> 04:12:43,360
TIME EVEN THOUGH THERE'S AN

6835
04:12:43,360 --> 04:12:44,628
AUTOMATED SYSTEM WITH FLAG

6836
04:12:44,628 --> 04:12:44,995
ISSUES.

6837
04:12:44,995 --> 04:12:47,397
JUST TO GROUND IT IN REALITY A

6838
04:12:47,397 --> 04:12:48,498
BIT, FOR INSTANCE 1 PROJECT JUST

6839
04:12:48,498 --> 04:12:49,966
TO GIVE AN EXAMPLE OF HOW THIS

6840
04:12:49,966 --> 04:12:50,901
WORKS ON INDIVIDUAL PROPERTIES

6841
04:12:50,901 --> 04:12:53,470
AND THEN ALSO HOW IT COULD WORK

6842
04:12:53,470 --> 04:12:55,238
IN COMBINATION, SO HERE THERE IS

6843
04:12:55,238 --> 04:12:56,306
AN OFFTARGET LIABILITY HERE FOR

6844
04:12:56,306 --> 04:12:59,910
A PARTICULAR SERIES ON A PROJECT

6845
04:12:59,910 --> 04:13:01,878
AND IN ORDER TO GET AROUND IT WE

6846
04:13:01,878 --> 04:13:03,547
HAVE FOCUSED DATA GENERATION,

6847
04:13:03,547 --> 04:13:05,482
BASED ON THE MODEL THAT THE

6848
04:13:05,482 --> 04:13:07,551
FLOABAL 1 IS NOT TRANSLATING AND

6849
04:13:07,551 --> 04:13:10,987
SO BEING ABLE TO DECREASE THAT

6850
04:13:10,987 --> 04:13:13,256
OVER TIME THROUGH THESE

6851
04:13:13,256 --> 04:13:16,126
DIFFERENT KINDS OF ACTIVE

6852
04:13:16,126 --> 04:13:17,561
LEARNING ITERATIONS WAS PRETTY

6853
04:13:17,561 --> 04:13:19,729
KEY TO THE PROJECT AND AT THE

6854
04:13:19,729 --> 04:13:21,264
SAME TIME PROVIDING THE

6855
04:13:21,264 --> 04:13:22,165
VISUALIZATIONS AND

6856
04:13:22,165 --> 04:13:23,100
JUSTIFICATIONS FOR WHY WE NEED

6857
04:13:23,100 --> 04:13:24,134
TO DO THIS.

6858
04:13:24,134 --> 04:13:26,636
SO FOR INITANCEOT RIGHT SHOWING

6859
04:13:26,636 --> 04:13:28,371
THE GLOBAL MONO PREDICTIONS WERE

6860
04:13:28,371 --> 04:13:30,640
CLUSTER INDEED A SPACE THAT WERE

6861
04:13:30,640 --> 04:13:31,508
NOT APPLICABLE, TO THE LOCAL

6862
04:13:31,508 --> 04:13:32,642
THINGS THAT WERE NOT INTERESTED

6863
04:13:32,642 --> 04:13:36,313
IN YOU CAN SEE IN THE FIRST 2

6864
04:13:36,313 --> 04:13:36,880
COLUMNS.

6865
04:13:36,880 --> 04:13:37,514
CHEMISTRY'S VERY DIFFERENT.

6866
04:13:37,514 --> 04:13:41,518
SO THINGS LIKE THIS ARE PRETTY

6867
04:13:41,518 --> 04:13:43,019
ESSENTIAL FOR COMMUNICATION

6868
04:13:43,019 --> 04:13:43,487
ABOUT PRIORITIZATION,

6869
04:13:43,487 --> 04:13:45,188
EXPERIMENTS THAT NEED TO BE DONE

6870
04:13:45,188 --> 04:13:47,124
AS OPPOSE TO JUST SAYING IT OR

6871
04:13:47,124 --> 04:13:49,893
TAKING SOMEBODY'S WORD FOR IT.

6872
04:13:49,893 --> 04:13:51,895
YOU KNOW IT'S A CENTRAL PART OF

6873
04:13:51,895 --> 04:13:53,763
THE PROJECT TEAM MOVING FORWARD.

6874
04:13:53,763 --> 04:13:56,166
AND SO LOOKING AT THIS, IS MORE

6875
04:13:56,166 --> 04:14:00,070
THAN AN INDIVIDUAL PROPERTY, SO

6876
04:14:00,070 --> 04:14:01,538
LOOKING BACK AT THAT OFF TARGET

6877
04:14:01,538 --> 04:14:04,708
I MENTIONED BEFORE VERSUS

6878
04:14:04,708 --> 04:14:06,042
ON-TARGET POTENCY, BEING ABLE TO

6879
04:14:06,042 --> 04:14:07,344
DECREASE OR TUNE OUT THE

6880
04:14:07,344 --> 04:14:08,979
ACTIVITY THAT WE DON'T WANT,

6881
04:14:08,979 --> 04:14:10,881
WHILE KEEPING THE POTENCYOT

6882
04:14:10,881 --> 04:14:12,816
ACTUAL TARGET AT THE SAME TIME.

6883
04:14:12,816 --> 04:14:16,686
SO USING THAT EXAMPLE IS JUST AN

6884
04:14:16,686 --> 04:14:17,988
OFFTARGETS, IT'S A PRETTY GOOD

6885
04:14:17,988 --> 04:14:19,756
PROOF OF CONCEPT AND GETTING

6886
04:14:19,756 --> 04:14:23,293
BACK TO HOW WE ACTUALLY DO THIS

6887
04:14:23,293 --> 04:14:25,595
IN A PRACTICAL WAY, WOWRKING ON

6888
04:14:25,595 --> 04:14:27,197
MULTIPLE SEARS AT A TIME FOR

6889
04:14:27,197 --> 04:14:28,832
INSTANCE, WE TRY TO MAP OUT

6890
04:14:28,832 --> 04:14:30,066
GETTING BACK TO THAT DATA

6891
04:14:30,066 --> 04:14:31,234
STRATEGY AND BEING INTENTIONAL

6892
04:14:31,234 --> 04:14:34,604
ABOUT WHAT WE'RE DOING, MAP OUT

6893
04:14:34,604 --> 04:14:36,673
HOW LONG THESE CYCLES ACTUALLY

6894
04:14:36,673 --> 04:14:38,308
TAKE BASED ON THE CHEMISTRY

6895
04:14:38,308 --> 04:14:39,776
WE'RE INTERESTED IN AND STAGGER

6896
04:14:39,776 --> 04:14:40,977
THE CHEMISTRY RESOURCE WITH THE

6897
04:14:40,977 --> 04:14:43,346
SCREENING RESOURCE, WITH THE

6898
04:14:43,346 --> 04:14:44,147
COMPUTATIONAL RESOURCE ALL AT

6899
04:14:44,147 --> 04:14:45,916
THE SAME TIME SO THIS PROCESS

6900
04:14:45,916 --> 04:14:47,517
CAN BE ASENTIOUS FICIENT AS

6901
04:14:47,517 --> 04:14:49,019
POSSIBLE WHILE GENERATING THE

6902
04:14:49,019 --> 04:14:50,220
MAXIMUM AMOUNT OF IMPACTFUL DATA

6903
04:14:50,220 --> 04:14:51,755
WE CAN DO AT THE SAME TIME SO

6904
04:14:51,755 --> 04:14:53,790
THIS IS PRETTY MUCH ESSENTIAL

6905
04:14:53,790 --> 04:14:55,458
WHEN MAINTAINING A LARNG

6906
04:14:55,458 --> 04:14:56,993
PORTFOLIO OF PROGECS AS OPPOSE

6907
04:14:56,993 --> 04:14:59,729
TO A START UPPER SOMETHING.

6908
04:14:59,729 --> 04:15:00,931
WE REALLY FOCUSED ON 1 TARGET

6909
04:15:00,931 --> 04:15:08,438
AND PUTTING ALL THE RESOURCES

6910
04:15:08,438 --> 04:15:08,738
BEHIND IT.

6911
04:15:08,738 --> 04:15:12,442
SO I SHOULD SAY TOO, THIS IS

6912
04:15:12,442 --> 04:15:14,844
DONE ON A LARGER SCALE.

6913
04:15:14,844 --> 04:15:16,146
SO NOT SURPRISINGLY, WHEN

6914
04:15:16,146 --> 04:15:18,415
STARTING A PROJECT YOU WANT

6915
04:15:18,415 --> 04:15:20,784
THINGS THAT ARE SOLUBLE AND SO

6916
04:15:20,784 --> 04:15:24,354
FOR AN ORAL DRUG FOR INSTANCE

6917
04:15:24,354 --> 04:15:26,289
AND AN EASY WAY TO DO THAT IS

6918
04:15:26,289 --> 04:15:28,458
TAKE GENERAL GUIDE LIVENS ABOUT

6919
04:15:28,458 --> 04:15:30,126
WHAT'S GOD THERE, TRY TO DESIGN

6920
04:15:30,126 --> 04:15:33,763
THEM FOR THE FULTERS OR

6921
04:15:33,763 --> 04:15:34,431
MULTIPROPERTY OPTIMIZATION

6922
04:15:34,431 --> 04:15:37,567
TECHNIQUES, AND DO THOSE OVER

6923
04:15:37,567 --> 04:15:38,301
TIME, THAT'S GREAT.

6924
04:15:38,301 --> 04:15:38,635
RIGHT?

6925
04:15:38,635 --> 04:15:40,503
WE CAN DO THAT AND THIS SHOWS AN

6926
04:15:40,503 --> 04:15:41,805
EXAMPLE ABOUT DIFFERENT KINDINGS

6927
04:15:41,805 --> 04:15:43,273
OF PROPERTIES THAT HAVE BEEN

6928
04:15:43,273 --> 04:15:45,642
IMPORTANT BASED ON THE SERIES

6929
04:15:45,642 --> 04:15:48,078
AND FLAGS OVER A VARIETY OF

6930
04:15:48,078 --> 04:15:48,878
ITERATIONS AND OVERTIME EMILY

6931
04:15:48,878 --> 04:15:52,048
THE ON A PARTICULAR PROJECT.

6932
04:15:52,048 --> 04:15:52,983
BUT ON UNSURPRISINGLY, WHAT THIS

6933
04:15:52,983 --> 04:15:55,652
DOES IS IT POTENTIA WILY GIVES

6934
04:15:55,652 --> 04:15:57,687
YOU G GOODISH COMPOUNDS.

6935
04:15:57,687 --> 04:15:58,888
IF GENERAL BUT THERE MIGHT BE

6936
04:15:58,888 --> 04:16:00,056
SOMETHING SPECIFIC FOR YOUR

6937
04:16:00,056 --> 04:16:00,957
PARTICULAR TARGET, FOR YOUR

6938
04:16:00,957 --> 04:16:02,759
PARTICULAR DISEASE OR FOR YOUR

6939
04:16:02,759 --> 04:16:05,161
PARTICULAR MECHANISM OF ACTION,

6940
04:16:05,161 --> 04:16:06,363
THAT IS IMPORTANT SO

6941
04:16:06,363 --> 04:16:11,668
UNDERSTANDING WHAT A GOOD

6942
04:16:11,668 --> 04:16:14,271
MOLECULE OR CANDIDATE MOLECULE

6943
04:16:14,271 --> 04:16:17,073
IN THAT PARTICULAR THING IS

6944
04:16:17,073 --> 04:16:18,241
POSSIBLE AND IMPORTANT TO SAVE

6945
04:16:18,241 --> 04:16:19,676
YOU FROM MAKING MODELS YOU DON'T

6946
04:16:19,676 --> 04:16:22,779
HAVE AND MAKING IT BETTER FOR TO

6947
04:16:22,779 --> 04:16:24,047
YOU HAVE THIS.

6948
04:16:24,047 --> 04:16:25,515
IT ALSO HELPS SET UP

6949
04:16:25,515 --> 04:16:26,182
MULTIOBLIGATIONS YECTIVE OPTICAL

6950
04:16:26,182 --> 04:16:27,317
IMAGES MIERSATION THAT AT THE

6951
04:16:27,317 --> 04:16:29,286
SAME TIME THAT ARE SPECIFIC TO

6952
04:16:29,286 --> 04:16:30,787
YOUR DISEASE OR EVEN YOUR

6953
04:16:30,787 --> 04:16:32,789
SERIES, OR MODALITY AT THE SAME

6954
04:16:32,789 --> 04:16:32,989
TIME.

6955
04:16:32,989 --> 04:16:36,860
AND SO 1 POTENTIAL WAY TO DO

6956
04:16:36,860 --> 04:16:40,030
THAT, IS ACTUALLY GOING AND

6957
04:16:40,030 --> 04:16:42,666
TAKING THIS FROM A CLINICAL

6958
04:16:42,666 --> 04:16:45,101
MODELING AND GOING BACK TO

6959
04:16:45,101 --> 04:16:46,903
TRANSLATION WHICH IS THESE

6960
04:16:46,903 --> 04:16:51,207
COURSE MECHANISTIC MODELS, AND

6961
04:16:51,207 --> 04:16:52,342
QUANTITATIVE SYSTEMS

6962
04:16:52,342 --> 04:16:53,009
PHARMACOLOGY, AND WHERE EACH OF

6963
04:16:53,009 --> 04:16:54,344
THE DISEASES WHERE YOU KNOW FOR

6964
04:16:54,344 --> 04:16:56,913
INSTANCE BY THE TIME YOU HAVE A

6965
04:16:56,913 --> 04:16:58,148
CANDIDATE AND HOW THESE ARE

6966
04:16:58,148 --> 04:17:01,418
FORMED HOW THESE ARE   PERFORMED

6967
04:17:01,418 --> 04:17:04,120
IN IN VIVO, OR MAYBE EVEN HAVE

6968
04:17:04,120 --> 04:17:05,221
CLINICAL DISEASE THAT YOU'VE

6969
04:17:05,221 --> 04:17:06,656
BEEN INVESTING IN OVER TIME OR

6970
04:17:06,656 --> 04:17:09,192
BOUGHT THAT DATA FOR INSTANCE

6971
04:17:09,192 --> 04:17:12,929
AND THEN ALSO UNDERSTAND THE PBK

6972
04:17:12,929 --> 04:17:16,199
PROPERTIES FOR THE SERIES AND

6973
04:17:16,199 --> 04:17:18,101
AND SO WHAT WORKING BACKWARDS

6974
04:17:18,101 --> 04:17:20,637
FROM THOSE, GOING BACKWARDS FROM

6975
04:17:20,637 --> 04:17:22,706
THE CLINIC, SO A REVERSE

6976
04:17:22,706 --> 04:17:24,641
TRANSLATION APPROACH, CAN YOU

6977
04:17:24,641 --> 04:17:25,942
PREDICT THESE ADME PROPERTIES

6978
04:17:25,942 --> 04:17:27,477
WITH REALLY GOOD MODELS, SO

6979
04:17:27,477 --> 04:17:29,846
UNDERSTANDING HOW THE INDIVIDUAL

6980
04:17:29,846 --> 04:17:33,016
COMPONENTS OF PERMIABILITY,

6981
04:17:33,016 --> 04:17:33,516
SOLVABILITY, CLEARANCE,

6982
04:17:33,516 --> 04:17:35,051
BIOAVAILABILITY, ALM THESE

6983
04:17:35,051 --> 04:17:35,952
THINGS THAT ARE IMPORTANT

6984
04:17:35,952 --> 04:17:37,987
INDIVIDUALLY HOW THEY MAY

6985
04:17:37,987 --> 04:17:39,389
COMBINE INTO THE PROPERTIES AND

6986
04:17:39,389 --> 04:17:41,491
THEY MAY TRANSLATE INTO THESE TO

6987
04:17:41,491 --> 04:17:44,861
BE EFFICACIOUS IN HUMANS.

6988
04:17:44,861 --> 04:17:47,197
SO BEING ABLE TO TAKE THEY--THE

6989
04:17:47,197 --> 04:17:48,431
KRETICALLY BEING ABLE TO TAKE

6990
04:17:48,431 --> 04:17:49,666
THESE MODELS OF THESE MOLECULES

6991
04:17:49,666 --> 04:17:51,267
THAT WERE INTERESTED IN, AND

6992
04:17:51,267 --> 04:17:53,970
THEN AGAIN WORK BACKWARDS SO,

6993
04:17:53,970 --> 04:17:57,173
BEING ABLE TO VIRTUALLY PROFILE

6994
04:17:57,173 --> 04:17:59,042
HUNDREDS OF THOUSANDS OF

6995
04:17:59,042 --> 04:18:00,076
POTENTIAL MOLECULES, BED THE

6996
04:18:00,076 --> 04:18:01,478
DRIVERS OF THE EFFICACY THAT ARE

6997
04:18:01,478 --> 04:18:04,114
CONTRIBUTING TO THE PK, SO THAT

6998
04:18:04,114 --> 04:18:06,082
COULD BE SOLIEWBILITYD, COULD BE

6999
04:18:06,082 --> 04:18:07,550
PERMIABILITY OR COULD BE A

7000
04:18:07,550 --> 04:18:10,987
REALLY COMPLICATED KRIEWK OF ALL

7001
04:18:10,987 --> 04:18:13,323
OF THEM INTO ONCE BUT USING

7002
04:18:13,323 --> 04:18:15,825
DIFFERENT KINDS OF MACHINE

7003
04:18:15,825 --> 04:18:16,860
LEARNING, AND DESIGN STRATEGY

7004
04:18:16,860 --> 04:18:22,465
SHOULD BE GOING FORWARD, AND FOR

7005
04:18:22,465 --> 04:18:24,834
THE NEXT ENUMERATION, YOU CAN

7006
04:18:24,834 --> 04:18:26,503
BUILD BETTER MODELS AND KEEP

7007
04:18:26,503 --> 04:18:27,170
ITERATING THROUGH THE SERIES SO

7008
04:18:27,170 --> 04:18:33,510
IT'S NOT JUST MAKING OR PLAYING

7009
04:18:33,510 --> 04:18:36,646
WHACK-A-MOLE, AND SAYING I WANT

7010
04:18:36,646 --> 04:18:37,781
TO OPTIMIZE PERMIABILITY BUT

7011
04:18:37,781 --> 04:18:39,082
LOOKING LARGELY IN A TRANSLATION

7012
04:18:39,082 --> 04:18:43,253
POINT OF VIEW, HOW DO WE DO

7013
04:18:43,253 --> 04:18:44,721
HUMAN DOZE THAT WE'RE TRYING TO

7014
04:18:44,721 --> 04:18:48,358
OPTIMIZE AS EARLY AS POSSIBLE.

7015
04:18:48,358 --> 04:18:50,260
YOU CAN HAVE 2 DIFFERENT

7016
04:18:50,260 --> 04:18:55,932
COMPOUNDS LABELED A AND B HERE,

7017
04:18:55,932 --> 04:18:57,667
BUT DIFFERENT BALANCE OF AND

7018
04:18:57,667 --> 04:18:59,369
THIS IS THE PROBABILITY OF WHAT

7019
04:18:59,369 --> 04:19:01,171
THE HUMAN DOSE MIGHT END UP

7020
04:19:01,171 --> 04:19:02,305
LOOKING LIKE, KIND OF HELPS MAKE

7021
04:19:02,305 --> 04:19:03,907
IT CLEAR ABOUT WHAT THE DESIGN

7022
04:19:03,907 --> 04:19:05,341
STRATEGY SHOULD BE AND WHAT

7023
04:19:05,341 --> 04:19:08,178
SERIES WE MAYBE WANT TO

7024
04:19:08,178 --> 04:19:09,145
PRIORITIZE WITH MORE RESOURCE

7025
04:19:09,145 --> 04:19:10,880
GOING FORWARD, SO DO WE THINK

7026
04:19:10,880 --> 04:19:13,183
IT'S EASIER TO OPTIMIZE POTENCY

7027
04:19:13,183 --> 04:19:15,218
BUT KEEP ADME THE SAME, IF WE

7028
04:19:15,218 --> 04:19:17,287
REDUCE IT BY 5 X, WE CAN

7029
04:19:17,287 --> 04:19:18,755
INCREASE THE PROBABILITY OF A

7030
04:19:18,755 --> 04:19:19,856
LOWER DOSE MUCH MORE OR IS IT

7031
04:19:19,856 --> 04:19:22,392
GOING TO BE HARDER TO OPTIMIZE

7032
04:19:22,392 --> 04:19:26,663
THE ADME PROPERTIES FOR INSTANCE

7033
04:19:26,663 --> 04:19:30,700
AND THEN WE USE AND BREAK DOWN

7034
04:19:30,700 --> 04:19:31,501
WHICH PHYSICAL PROPERTIES

7035
04:19:31,501 --> 04:19:34,404
CONTRIBUTE TO THE OVERALL ADME

7036
04:19:34,404 --> 04:19:37,974
SCORE TO OPTIMIZE FOR FURTHER

7037
04:19:37,974 --> 04:19:38,942
DESIGN ROUNDS.

7038
04:19:38,942 --> 04:19:40,944
AND I SHOULD SAY TOO, THERE ARE

7039
04:19:40,944 --> 04:19:45,982
OTHER KINDS OF PROPERTIES THAN

7040
04:19:45,982 --> 04:19:48,151
THAT IMPACT THAT KIND OF ARK

7041
04:19:48,151 --> 04:19:50,687
SUSMGZ WHERE THERE MIGHT BE

7042
04:19:50,687 --> 04:19:51,688
BIOLOGICAL VARIABLES WE DON'T

7043
04:19:51,688 --> 04:19:53,089
UNDERSTAND, WHETHER IT'S A NEW

7044
04:19:53,089 --> 04:19:54,357
DISEASE OR SOMETHING LIKE THIS

7045
04:19:54,357 --> 04:19:56,159
SO WE CAN MAKE ASSUMPTIONS AND

7046
04:19:56,159 --> 04:19:57,560
SEE HOW THEY LOOK OVER TIME, SO

7047
04:19:57,560 --> 04:19:58,962
THE PLOTS ON THE RIGHT ARE

7048
04:19:58,962 --> 04:20:04,467
DIFFERENT LEVELS OF EFFLUX RATIO

7049
04:20:04,467 --> 04:20:07,036
IN THE SAME PLOT AND SHOW HOW

7050
04:20:07,036 --> 04:20:08,137
THAT CHANGES TIME AND THE

7051
04:20:08,137 --> 04:20:09,506
PROBLEMMABILITY OF SUCCESS MIGHT

7052
04:20:09,506 --> 04:20:12,041
CALL OUT AN IMPORTANT PARAMETER,

7053
04:20:12,041 --> 04:20:14,344
FOR INSTANCE, THAT WE NEED THAT

7054
04:20:14,344 --> 04:20:16,045
WE NEED TO PRIORITIZE WITH

7055
04:20:16,045 --> 04:20:16,713
EXPERIMENTAL TESTING TO

7056
04:20:16,713 --> 04:20:17,480
UNDERSTAND WHAT OUR COMPOUNDS

7057
04:20:17,480 --> 04:20:18,948
ARE DOING WHETHER OR NOT A

7058
04:20:18,948 --> 04:20:23,286
SERIES IS WORTH PURSUING.

7059
04:20:23,286 --> 04:20:24,287
SO IT HIGHLIGHTS BASICALLY THE

7060
04:20:24,287 --> 04:20:27,957
THINGS WE DON'T KNOW AND SHOULD

7061
04:20:27,957 --> 04:20:28,858
PRIORITIZE THE RESOURCE AROUND

7062
04:20:28,858 --> 04:20:29,959
CERTAIN EXPERIMENTS THAT ARE

7063
04:20:29,959 --> 04:20:31,561
CLEAR NO GOES FOR THE PROJECT OR

7064
04:20:31,561 --> 04:20:34,163
FOR THE PARTICULAR SERIES MUCH

7065
04:20:34,163 --> 04:20:36,366
SAME SORT OF THING JUST AROUND

7066
04:20:36,366 --> 04:20:38,067
TURNOVER RATE OF THE ACTUAL

7067
04:20:38,067 --> 04:20:39,836
TARGET IN A PD SENSE AND SO YOU

7068
04:20:39,836 --> 04:20:43,406
CAN HAVE THE SAME MOLECULES IN

7069
04:20:43,406 --> 04:20:44,741
TERMS OF POTENCY, ADME PROPERTY

7070
04:20:44,741 --> 04:20:45,875
ANDS THINGS LIKE THIS BUT

7071
04:20:45,875 --> 04:20:47,043
DEPENDING ON WHAT THE ACTUAL

7072
04:20:47,043 --> 04:20:49,979
PROPERTIES OF THE MOLECULE AND

7073
04:20:49,979 --> 04:20:51,548
THE TARGET ARE?

7074
04:20:51,548 --> 04:20:52,849
IT COULD HAVE 0 PERCENT CHANCE

7075
04:20:52,849 --> 04:20:55,451
OF SUCCESS OR CLINICAL

7076
04:20:55,451 --> 04:20:56,853
EFFICACIOUS RESPONSE RATE, OR IT

7077
04:20:56,853 --> 04:20:58,855
COULD HAVE A 50% CHANCE OF

7078
04:20:58,855 --> 04:21:01,024
SUCCESS SO BEING ABLE TO

7079
04:21:01,024 --> 04:21:03,960
HIGHLIGHT THE PROPER EXPERIMENTS

7080
04:21:03,960 --> 04:21:05,361
TO DO TO INTERPRET TO TELL YOU

7081
04:21:05,361 --> 04:21:09,933
WHAT TO DO NEXT, IS IMPORTANT

7082
04:21:09,933 --> 04:21:12,035
FOR PRIORITIZING RESOURCE AND SO

7083
04:21:12,035 --> 04:21:13,836
GOING BACK TO COMPOUND DESIGN,

7084
04:21:13,836 --> 04:21:15,338
WHERE THIS COULD COME INTO PLAY

7085
04:21:15,338 --> 04:21:17,106
ESPECIALLY AS EARLY AS POSSIBLE,

7086
04:21:17,106 --> 04:21:21,177
WHERE YOU WANT TO IT TOO, IS

7087
04:21:21,177 --> 04:21:21,945
TAKING INITIAL [INDISCERNIBLE]

7088
04:21:21,945 --> 04:21:23,379
SO HERE IN THE MIDDLE IS THAT

7089
04:21:23,379 --> 04:21:27,317
RED, IS THAT BLACK BOX, AND

7090
04:21:27,317 --> 04:21:29,319
DOING DIFFERENT KINDS OF

7091
04:21:29,319 --> 04:21:30,019
MOLECULE GENERATORS, AROUND

7092
04:21:30,019 --> 04:21:34,958
THOSE  SCAFFOLDS, DIFFERENT

7093
04:21:34,958 --> 04:21:37,026
SCREENS, THINGS LIKE THIS TO

7094
04:21:37,026 --> 04:21:38,061
UNDERSTAND THE OPTIMIZE ABILITY

7095
04:21:38,061 --> 04:21:41,364
OR POTENTIAL OF THAT SERIES TO

7096
04:21:41,364 --> 04:21:42,932
REACH ANY PROBABILITY OF SUCCESS

7097
04:21:42,932 --> 04:21:47,303
OF BEING EFFICACIOUS OR WITH A

7098
04:21:47,303 --> 04:21:47,570
LOW DOSE.

7099
04:21:47,570 --> 04:21:50,039
AND SO BEING ABLE TO THEN TAKE

7100
04:21:50,039 --> 04:21:51,774
THOSE SERIES LOOK AT THE COMMON

7101
04:21:51,774 --> 04:21:52,875
PROPERTIES OF ANY OF THOSE

7102
04:21:52,875 --> 04:21:54,477
MOLECULES THAT ARE PREDICTED TO

7103
04:21:54,477 --> 04:21:58,748
HAVE A LORER DOSE, UNDERSTAND

7104
04:21:58,748 --> 04:22:02,819
WHERE THOSE KEY PHARMA 4S ARE

7105
04:22:02,819 --> 04:22:04,187
AND UTILIZING THOSE OR IF WHAT

7106
04:22:04,187 --> 04:22:05,688
WE DO, HOW MUCH DESIGN EFFORTS

7107
04:22:05,688 --> 04:22:07,490
WE PUT INTO IT IS NEVER GOING TO

7108
04:22:07,490 --> 04:22:09,459
BE PREDICTED LOWER THAT KNOW

7109
04:22:09,459 --> 04:22:11,961
THAT THAN EITHER WE CAN KILL A

7110
04:22:11,961 --> 04:22:13,730
SERIES, WE CAN REPRIORITIZE,

7111
04:22:13,730 --> 04:22:16,232
TEST 1 OR 2 THINGS THAT ARE THE

7112
04:22:16,232 --> 04:22:17,300
VIRTUAL NAIL IN THE COFFIN TO

7113
04:22:17,300 --> 04:22:19,836
MAKE SURE WE HAVE CONFIDENCE IN

7114
04:22:19,836 --> 04:22:21,771
THE MODEL PREDICS, IT INFORMS

7115
04:22:21,771 --> 04:22:25,308
THE NEXT DESIGN FOR THE DESIGN

7116
04:22:25,308 --> 04:22:25,575
STRATEGY.

7117
04:22:25,575 --> 04:22:29,145
SO JUST TRYING TO THINK AGAIN

7118
04:22:29,145 --> 04:22:29,679
GLOBALLY HOW MODELING AND

7119
04:22:29,679 --> 04:22:33,049
TRANSLATE ALL THE WAY FROM THE

7120
04:22:33,049 --> 04:22:35,551
VERY BEGINNING TO THE CLINIC AND

7121
04:22:35,551 --> 04:22:36,686
FEEDBACK PRACTICES THAT TO

7122
04:22:36,686 --> 04:22:39,222
ANSWER QUESTIONS ABOUT HOW DO WE

7123
04:22:39,222 --> 04:22:42,258
MODULATE THIS TARGET IN AN

7124
04:22:42,258 --> 04:22:42,992
EFFICACIOUS WAY, WHO MODALITY

7125
04:22:42,992 --> 04:22:48,297
YOU WANT TO CHOOSE, SO VALENCEY

7126
04:22:48,297 --> 04:22:50,566
VERSUS SYNTAX, VERSUS FUNCTIONAL

7127
04:22:50,566 --> 04:22:52,235
INHIBITION AND THEN PRIORITIZING

7128
04:22:52,235 --> 04:22:53,536
DIFFERENT KINDS OF COMPOUNDS, 1

7129
04:22:53,536 --> 04:22:55,038
OF THE GAPS TO BUILD A ABOUT

7130
04:22:55,038 --> 04:22:56,806
THER MODEL WE MIGHT NEED OR MAKE

7131
04:22:56,806 --> 04:22:58,307
MORE ACCURATE PREDICTIONS AND

7132
04:22:58,307 --> 04:22:59,208
PRYERITIZE THOSE EXPERIMENTS AND

7133
04:22:59,208 --> 04:23:02,979
THEN AGAIN WHARKS IS ACTUALLY

7134
04:23:02,979 --> 04:23:04,280
TRIEFING THE BIOLOGICAL EFFICACY

7135
04:23:04,280 --> 04:23:06,883
AND WHAT ARE THE PHYSICAL

7136
04:23:06,883 --> 04:23:08,384
PROPHECIES THAT ARE CLEAR FOR US

7137
04:23:08,384 --> 04:23:08,918
TO DESIGN TO.

7138
04:23:08,918 --> 04:23:11,587
SO WITH THAT I JUST WANT TO SAY

7139
04:23:11,587 --> 04:23:13,056
THANK YOU TO EVERYBODY AND WAY

7140
04:23:13,056 --> 04:23:14,991
TOO MANY PEOPLE AT GSK AND

7141
04:23:14,991 --> 04:23:15,591
ELSEWHERE TO THINK.

7142
04:23:15,591 --> 04:23:21,764
SO THIS IS JUST A GENERAL 1.

7143
04:23:21,764 --> 04:23:23,533
APPRECIATE IT.

7144
04:23:23,533 --> 04:23:25,635
HOPEFULLY I CAUGHT US ON TIME.

7145
04:23:25,635 --> 04:23:26,035
THAT.

7146
04:23:26,035 --> 04:23:33,176
>> THAT WAS PERFECT. 

7147
04:23:33,176 --> 04:23:33,843
THAT WAS GREAT.

7148
04:23:33,843 --> 04:23:34,143
>> GREAT.

7149
04:23:34,143 --> 04:23:34,744
>> AT THIS POINT WHAT
WE'RE

7150
04:23:34,744 --> 04:23:37,447
GOING TO DO IS IF THE PANELISTS,

7151
04:23:37,447 --> 04:23:39,282
THERE'S MARTY AND RUSS IF YOU

7152
04:23:39,282 --> 04:23:44,587
COULD COME BACK TO US THAT WOULD

7153
04:23:44,587 --> 04:23:44,921
BE GREAT.

7154
04:23:44,921 --> 04:23:46,889
HOW ARE YOU DOING, WE WILL JUST

7155
04:23:46,889 --> 04:23:47,790
DISCUSS THINGS I HAVE QUESTIONS

7156
04:23:47,790 --> 04:23:49,726
THAT I HAVE SORT OF COME UP

7157
04:23:49,726 --> 04:23:51,227
WITH, I'M SURE -- AND I -- YOU

7158
04:23:51,227 --> 04:23:53,396
KNOW I WANT TO KEEP IT OPEN, WE

7159
04:23:53,396 --> 04:23:54,764
DO HAVE SOME SORT OF MORE

7160
04:23:54,764 --> 04:24:01,504
TECHNICAL QUESTIONS IN THE
Q&A

7161
04:24:01,504 --> 04:24:03,639
I'M GOING TO START WITH AND I

7162
04:24:03,639 --> 04:24:05,475
WILL INVITE THE UDIENCE, IF YOU

7163
04:24:05,475 --> 04:24:07,710
WANT TO PUT QUESTIONS IN, MAYBE

7164
04:24:07,710 --> 04:24:10,246
I CAN JUST START IT OFF THOUGH,

7165
04:24:10,246 --> 04:24:12,148
I THINK SO WE HEARD, WE REALLY

7166
04:24:12,148 --> 04:24:14,083
HEARD, I THINK LIKE A REALLY

7167
04:24:14,083 --> 04:24:16,052
GREAT RANGE OF TALKS, YOU KNOW

7168
04:24:16,052 --> 04:24:19,222
TAWRCHING ON MANY DEFINITE

7169
04:24:19,222 --> 04:24:20,289
EXCITING THICKS, HAPPENING

7170
04:24:20,289 --> 04:24:24,527
INSILICO DRUG DISCOVERY AND

7171
04:24:24,527 --> 04:24:24,761
DESIGN.

7172
04:24:24,761 --> 04:24:26,295
I'M CURIOUS SORT OF A BASIC

7173
04:24:26,295 --> 04:24:35,571
QUESTION OF WHAT ARE YOU MOST

7174
04:24:35,571 --> 04:24:36,606
EXCITED ABOUT IN THIS LAND SCAPE

7175
04:24:36,606 --> 04:24:38,274
AND I KNOW IT WILL BE DIFFERENT

7176
04:24:38,274 --> 04:24:41,410
FOR EVERYBODY BECAUSE WE ALL

7177
04:24:41,410 --> 04:24:48,818
HAVE DIFFERENT THINGS WE ALL

7178
04:24:48,818 --> 04:24:49,018
LIKE.

7179
04:24:49,018 --> 04:24:50,987
PAT, CAN I START WITH YOU SINCE

7180
04:24:50,987 --> 04:24:52,088
YOU'RE UNMUTED.

7181
04:24:52,088 --> 04:24:52,455
>> SURE.

7182
04:24:52,455 --> 04:24:54,857
THE MOST EXCITING PART FOR ME IS

7183
04:24:54,857 --> 04:24:56,592
THE PART WE'RE ENCOUNTERING

7184
04:24:56,592 --> 04:25:04,600
BETWEEN PHYSICS AND MACHINE

7185
04:25:04,600 --> 04:25:05,234
LEARNING.

7186
04:25:05,234 --> 04:25:06,602
AND I THINKA BECAUSE OF THAT, WE

7187
04:25:06,602 --> 04:25:08,704
CAN DO THINGS AT A MUCH HIGHER

7188
04:25:08,704 --> 04:25:09,939
RESOLUTION THAN WE COULD DO

7189
04:25:09,939 --> 04:25:11,641
BEFORE, BUT THE THING THAT

7190
04:25:11,641 --> 04:25:14,710
EXCITING ME THE MOST IS WE HAVE

7191
04:25:14,710 --> 04:25:17,280
THIS PROTEIN STRUCKURE AT 1 END,

7192
04:25:17,280 --> 04:25:18,948
YOU HAVE THE ALPHA FOLD

7193
04:25:18,948 --> 04:25:21,284
TECHNIQUES AND AT THE OTHER END,

7194
04:25:21,284 --> 04:25:22,618
WE HAVE MOLECULAR DYNAMICS AND

7195
04:25:22,618 --> 04:25:24,787
IT'S ALL KIND OF COMING TOGETHER

7196
04:25:24,787 --> 04:25:26,856
AND I THINK, YOU KNOW IN A

7197
04:25:26,856 --> 04:25:28,157
COUPLE OF YEARS, WE'RE ALL GOING

7198
04:25:28,157 --> 04:25:30,026
TO BE DOING THESE MULTISCALE

7199
04:25:30,026 --> 04:25:31,127
SIMMULESS THAT ARE GOING TO

7200
04:25:31,127 --> 04:25:32,762
ENABLE US TO DO THINGS THAT WE

7201
04:25:32,762 --> 04:25:42,638
COULD NEVER EVEN THINK ABOUT

7202
04:25:42,638 --> 04:25:42,839
DOING.

7203
04:25:42,839 --> 04:25:44,006
>> I'M SMILING TO HEAR YOU
SAY

7204
04:25:44,006 --> 04:25:46,976
THAT, I THINK THE IDEA OF

7205
04:25:46,976 --> 04:25:48,444
RUNNING MULTIPLE SIMULATIONS IS

7206
04:25:48,444 --> 04:25:48,811
PRETTY COOL.

7207
04:25:48,811 --> 04:25:50,413
I HAVE A QUESTION BUT I'M SAYING

7208
04:25:50,413 --> 04:25:52,481
IT FOR THE END.

7209
04:25:52,481 --> 04:25:55,218
MARTY YOU ARE UNMUTED AND

7210
04:25:55,218 --> 04:25:56,719
ANYTHING YOU'RE PARTICULARLY

7211
04:25:56,719 --> 04:25:57,520
EXCITED ABOUT?

7212
04:25:57,520 --> 04:25:59,388
>> THOSE WHO ATTENDED
HEARD SOME

7213
04:25:59,388 --> 04:26:01,023
OF WHAT I HAD TO SAY IN THE

7214
04:26:01,023 --> 04:26:02,325
KICKOFF SO YOU KNOW A BIT ABOUT

7215
04:26:02,325 --> 04:26:04,727
THAT BUT IT ECHOES SOME OF WHAT

7216
04:26:04,727 --> 04:26:07,597
PAT'S SAYING, IT ECHOES SOME OF

7217
04:26:07,597 --> 04:26:09,332
THE THEMES THAT JOEL WAS PULLING

7218
04:26:09,332 --> 04:26:12,668
OUT AT THE END WHICH IS IT'S NOT

7219
04:26:12,668 --> 04:26:14,470
JUST 1 HAMMER, I'M ANYTHING GO

7220
04:26:14,470 --> 04:26:15,805
OUT AND FIND NAILS TO HAMMER,

7221
04:26:15,805 --> 04:26:22,411
IT'S HOW DO I BRING ALL OF THESE

7222
04:26:22,411 --> 04:26:22,945
THINGS TOGETHER BOIGHTSD

7223
04:26:22,945 --> 04:26:24,146
COMPUTATIONALLY AND IN THE WET

7224
04:26:24,146 --> 04:26:24,747
LAB.

7225
04:26:24,747 --> 04:26:26,749
SO ON THE MULTIOMICS SIDE, WHAT

7226
04:26:26,749 --> 04:26:29,619
ARE THE DATA SETS THAT INFORM US

7227
04:26:29,619 --> 04:26:33,256
ACROSS ALL OF THESE THINGS.

7228
04:26:33,256 --> 04:26:35,758
AND THEN WHAT ARE THE

7229
04:26:35,758 --> 04:26:36,592
COMPUTATIONAL METHODS THAT

7230
04:26:36,592 --> 04:26:38,694
INTEGRATE THOSE DATA TOGETHER TO

7231
04:26:38,694 --> 04:26:40,630
UNDERSTAND DISEASE BIOLOGY

7232
04:26:40,630 --> 04:26:43,966
BETTER AND TO UNDERSTAND WHETHER

7233
04:26:43,966 --> 04:26:47,470
OUR MOLECULES ARE IMPACTING THE

7234
04:26:47,470 --> 04:26:49,038
SYSTEM IN THE RIGHT WAY AND YOU

7235
04:26:49,038 --> 04:26:52,675
WOULD FALL OFF YOUR CHAIR IF I

7236
04:26:52,675 --> 04:26:54,644
SAID WE'RE GOING TO DO MORE MD

7237
04:26:54,644 --> 04:26:59,115
SIMULATIONS SO I'M NOT GOING TO

7238
04:26:59,115 --> 04:27:00,583
SAY THAT.

7239
04:27:00,583 --> 04:27:00,983
>> THAT'S GREAT.

7240
04:27:00,983 --> 04:27:05,221
YOU CAN TELL I'M EXCITED.

7241
04:27:05,221 --> 04:27:09,659
>> SO RUSS OR PAT?

7242
04:27:09,659 --> 04:27:11,227
>> THE THEY SAID THE MOST

7243
04:27:11,227 --> 04:27:13,429
EFFICIENT PRPLT IS NO EXPERIMENT

7244
04:27:13,429 --> 04:27:14,664
AT ALL AND WHAT I'M EXCITED

7245
04:27:14,664 --> 04:27:16,599
ABOUT IS THAT WE WILL BE ABLE ON

7246
04:27:16,599 --> 04:27:18,868
DO A TON OF PRETTY CHEAP

7247
04:27:18,868 --> 04:27:20,036
COMPUTATION, IT'S NOT FREE BUT

7248
04:27:20,036 --> 04:27:21,070
COMPARED TO A LOT OF EXPERIMENTS

7249
04:27:21,070 --> 04:27:22,538
SO THAT WHEN WE'RE DOING OUR

7250
04:27:22,538 --> 04:27:24,240
EXPERIMENTS THEY WILL BE

7251
04:27:24,240 --> 04:27:25,908
CONFIRMATORY OF THINGS THAT WERE

7252
04:27:25,908 --> 04:27:27,476
PRETTY DARN SURE OF AND I'VE

7253
04:27:27,476 --> 04:27:28,544
SEEN IT ALREADY HAPPEN IN SOME

7254
04:27:28,544 --> 04:27:30,012
OF OUR WORK WHERE REALLY, WE

7255
04:27:30,012 --> 04:27:31,547
WENT TO A MOUSE AND WE IMU WE

7256
04:27:31,547 --> 04:27:33,015
HAD TO DO THE MOUSE BUT WE KNEW

7257
04:27:33,015 --> 04:27:34,150
IT WAS GOING TO WORK AND WHEN

7258
04:27:34,150 --> 04:27:35,918
THAT HAPPENS YOU ARE IN A VERY

7259
04:27:35,918 --> 04:27:38,487
GOOD MOOD AND YOU DON'T SPEND A

7260
04:27:38,487 --> 04:27:40,323
LOT OF MONEY, SO MAYBE IT WAS

7261
04:27:40,323 --> 04:27:41,958
SHERYL WHO SAID IT, I DON'T KNOW

7262
04:27:41,958 --> 04:27:43,359
WHO SAID IT, BUT I THINK THAT

7263
04:27:43,359 --> 04:27:45,027
THAT'S WHERE WE'RE HEADED AND

7264
04:27:45,027 --> 04:27:46,862
THAT'S GOING TO REALLY OPEN UP A

7265
04:27:46,862 --> 04:27:47,964
LOT OF OPPORTUNITIES, BECAUSE

7266
04:27:47,964 --> 04:27:50,933
YOU REALLY SHOULD DO EVERYTHING

7267
04:27:50,933 --> 04:27:51,968
COMPUTATIONAL EVEN WITH THE

7268
04:27:51,968 --> 04:27:53,402
AWARDS OF THE COMPUTATION AND WE

7269
04:27:53,402 --> 04:27:55,404
KNOW ABOUT THE AWARDS, IN MANY

7270
04:27:55,404 --> 04:27:59,375
WAYS IT'S BEEN A 2 DAY MEETING

7271
04:27:59,375 --> 04:28:00,509
ABOUT AWARDS BUT YOU SHOULD DO

7272
04:28:00,509 --> 04:28:01,610
ALL THAT BEFORE EXPERIMENTS

7273
04:28:01,610 --> 04:28:02,812
BECAUSE THEN HAVE YOU SOME

7274
04:28:02,812 --> 04:28:04,113
EXPECTATION THAT YOU'RE NOT

7275
04:28:04,113 --> 04:28:05,081
DOING REALLY SILLY EXPERIMENTS

7276
04:28:05,081 --> 04:28:07,383
THAT ARE NOT WORTH WHILE.

7277
04:28:07,383 --> 04:28:08,784
SO I'M EXCITED ABOUT THAT

7278
04:28:08,784 --> 04:28:11,220
BECAUSE IN COMPUTER FOLKS ARE

7279
04:28:11,220 --> 04:28:15,324
ACTUALLY CONTRIBUTING, GOD FOR

7280
04:28:15,324 --> 04:28:15,558
BID.

7281
04:28:15,558 --> 04:28:17,126
>> JOEL, HOW ABOUT ANY
THOUGHTS

7282
04:28:17,126 --> 04:28:19,028
ON WHAT YOU FIND PARTICULARLY

7283
04:28:19,028 --> 04:28:19,295
EXCITING.

7284
04:28:19,295 --> 04:28:25,668
I LOVED YOUR TALK BY THE WAY

7285
04:28:25,668 --> 04:28:25,935
NTHANKS.

7286
04:28:25,935 --> 04:28:27,403
YEAH, I THINK THAT THE

7287
04:28:27,403 --> 04:28:28,437
ENGINEERING AND TECHNICAL

7288
04:28:28,437 --> 04:28:32,508
PLATFORMS ARE CATCHING UP TO THE

7289
04:28:32,508 --> 04:28:33,275
COMPUTATION AND WE'RE CO

7290
04:28:33,275 --> 04:28:35,144
CREATING THOSE FOR THE SAME TIME

7291
04:28:35,144 --> 04:28:37,246
FOR THE TIME, SCALE AND QUALITY

7292
04:28:37,246 --> 04:28:38,781
OF DATA AND REPRODUCIBILITY THAT

7293
04:28:38,781 --> 04:28:39,915
WE ACTUALLY NEED AND MARTY WAS

7294
04:28:39,915 --> 04:28:41,917
TALKING ABOUT JUST THE FACT THAT

7295
04:28:41,917 --> 04:28:44,286
WE'RE IN THE SPACE TO INTEGRATE

7296
04:28:44,286 --> 04:28:46,155
THESE KINDS OF MULTIMODE AT

7297
04:28:46,155 --> 04:28:47,590
TYPES, TO MAKE STUFF THAT REALLY

7298
04:28:47,590 --> 04:28:51,327
POWERFUL IN TERMS OF MULTIOMICS,

7299
04:28:51,327 --> 04:28:53,129
GENETICS, IME NOAMICS ALL AT THE

7300
04:28:53,129 --> 04:28:54,864
SAME TIME TAKEN--THEY MAKE MORE

7301
04:28:54,864 --> 04:28:55,965
BRNT DECISIONS THERE, 'S A LOT

7302
04:28:55,965 --> 04:28:59,869
TO DO, IT'S EXCITING.

7303
04:28:59,869 --> 04:29:01,303
>> YEAH, VERY COOL.

7304
04:29:01,303 --> 04:29:05,141
SO A FLIP OF THAT QUESTION IS

7305
04:29:05,141 --> 04:29:06,675
THEN, -- OH MY GOD I HAVE SO

7306
04:29:06,675 --> 04:29:08,344
MANY THINGS TO TALK ABOUT -- BUT

7307
04:29:08,344 --> 04:29:09,879
A FLIP OF THAT QUESTION, MAYBE

7308
04:29:09,879 --> 04:29:11,580
YOU EACH TOUCHED ON THIS BUT

7309
04:29:11,580 --> 04:29:13,416
JUST SORT OF STEPPING BACK, WHAT

7310
04:29:13,416 --> 04:29:15,317
DO YOU VIEW AS THE BIGGEST

7311
04:29:15,317 --> 04:29:18,054
CHALLENGE TO YOU KNOW MAKING

7312
04:29:18,054 --> 04:29:18,921
PROGRESS, I EMBASSY IN THE

7313
04:29:18,921 --> 04:29:21,657
PARTICULAR SPACE OF WHAT YOU'RE

7314
04:29:21,657 --> 04:29:28,030
EXCITED ABOUT IN IS IT THE DATA?

7315
04:29:28,030 --> 04:29:28,330
ALGORITHMS?

7316
04:29:28,330 --> 04:29:28,998
IS IT THE TEAM?

7317
04:29:28,998 --> 04:29:31,300
WHAT DO YOU THINK?

7318
04:29:31,300 --> 04:29:36,372
>> IF I COULD START FROM

7319
04:29:36,372 --> 04:29:37,840
ACADEMIC   PERSPECT.

7320
04:29:37,840 --> 04:29:40,009
COMPUTE ISICLING US, MY

7321
04:29:40,009 --> 04:29:41,644
COLLEAGUES MIGHT HAVE A

7322
04:29:41,644 --> 04:29:43,746
DIFFERENT REACTION, BUT I KNOW

7323
04:29:43,746 --> 04:29:45,314
THEY HAVE SOME PAIN AS WELL, FOR

7324
04:29:45,314 --> 04:29:48,517
THE FIRST TIME IN MY CAREER IS

7325
04:29:48,517 --> 04:29:49,819
HOW ARE STUDENTING GOING TO PAY

7326
04:29:49,819 --> 04:29:51,320
FOR THE COMPUTE, WE HAVE

7327
04:29:51,320 --> 04:29:53,489
CLUSTERS AND WE HAVE SUPER

7328
04:29:53,489 --> 04:29:54,824
COMPUTERS BUT WE FIGURED OUT A

7329
04:29:54,824 --> 04:29:57,493
WAY BUT I AM CONCERNED AS AN

7330
04:29:57,493 --> 04:29:58,627
ACADEMIC ABOUT OUR ABILITY TO

7331
04:29:58,627 --> 04:29:59,795
MAKE CONTRIBUTIONS AT SCALE AND

7332
04:29:59,795 --> 04:30:01,297
SO THAT'S FINE, WE CAN TRY TO

7333
04:30:01,297 --> 04:30:03,566
MAKE OUR CONTRIBUTIONS AND BE

7334
04:30:03,566 --> 04:30:04,967
STRATEGIC PLAN TEEJIC BUT I DO

7335
04:30:04,967 --> 04:30:06,235
WORRY THAT IT'S FUNNY BECAUSE

7336
04:30:06,235 --> 04:30:09,305
YOU'RE TALKING ABOUT THE

7337
04:30:09,305 --> 04:30:10,473
EMERGING OF MOLECULAR DYNAMICS

7338
04:30:10,473 --> 04:30:13,909
AND AI ML, THE 2 MOST EXPENSIVE

7339
04:30:13,909 --> 04:30:15,411
COMPUTE -- LITERALLY THE THINGS

7340
04:30:15,411 --> 04:30:16,145
THAT USE --

7341
04:30:16,145 --> 04:30:17,646
>> WHY NOT BRING THEM
TOGETHER,

7342
04:30:17,646 --> 04:30:17,847
RIGHT?

7343
04:30:17,847 --> 04:30:19,014
>> AT THE NATIONAL LABS
THE MOST

7344
04:30:19,014 --> 04:30:20,916
AND WE WANT MORE OF THAT.

7345
04:30:20,916 --> 04:30:22,485
AND WE WANT TO DO A CROSS

7346
04:30:22,485 --> 04:30:24,253
PRODUCT OF AI AND BY THE WAY, I

7347
04:30:24,253 --> 04:30:28,257
TOTALLY AGREE WITH THAT, BUT WE

7348
04:30:28,257 --> 04:30:29,925
REAMLY DO HAVE COMPUTE

7349
04:30:29,925 --> 04:30:31,093
CHALLENGES, I THINK AND I LOLLED

7350
04:30:31,093 --> 04:30:34,864
TO HEAR FROM COLLEAGUES AND

7351
04:30:34,864 --> 04:30:35,798
INDUSTRY, THIS THAT'S A PECULIAR

7352
04:30:35,798 --> 04:30:36,665
PROBLEM FOR ME.

7353
04:30:36,665 --> 04:30:38,801
>> I WOULD SAY RUSS IT'S
NOT A

7354
04:30:38,801 --> 04:30:41,704
UNIQUE AND PECULIAR PROBLEM FOR

7355
04:30:41,704 --> 04:30:41,937
YOU.

7356
04:30:41,937 --> 04:30:44,573
I'M GOING TO THROW IN A

7357
04:30:44,573 --> 04:30:45,774
NONSEQUITER BECAUSE IT'S ON MY

7358
04:30:45,774 --> 04:30:47,610
MIND WHEN WE TALK ABOUT COMPUTE

7359
04:30:47,610 --> 04:30:49,345
BUT I'VE HEARD SO MANY PEOPLE IN

7360
04:30:49,345 --> 04:30:51,914
THE AI WORLD IN THE PUBLIC

7361
04:30:51,914 --> 04:30:54,683
DOMAIN TALKING ABOUT HOW THE

7362
04:30:54,683 --> 04:30:59,889
SINGULARITY IS COMING, AND THE

7363
04:30:59,889 --> 04:31:01,390
COMPUTERS ARE GOING TO BE ABLE

7364
04:31:01,390 --> 04:31:02,358
TO THINK.

7365
04:31:02,358 --> 04:31:03,692
YOU CANNOT GENERATE ENOUGH POWER

7366
04:31:03,692 --> 04:31:08,297
TO RUN A BIG ENOUGH COMPUTER TO

7367
04:31:08,297 --> 04:31:10,633
HAVE THE SINGULARITY HAPPEN.

7368
04:31:10,633 --> 04:31:14,803
SO THAT'S CRAZY TALK EMPLOY AND

7369
04:31:14,803 --> 04:31:16,372
ROLLING IT BACK TO THE SIMPLER

7370
04:31:16,372 --> 04:31:19,341
QUESTIONS HERE, SOME OF OUR

7371
04:31:19,341 --> 04:31:21,277
ASPIRATIONS ARE PROBABLY BIGGER

7372
04:31:21,277 --> 04:31:23,746
THAN THE COMPUTE THAT WE HAVE IN

7373
04:31:23,746 --> 04:31:28,951
HAND RIGHT NOW AND SO, IT IS THE

7374
04:31:28,951 --> 04:31:30,920
NEXT BIG INTUITIVE LEAP PROBABLY

7375
04:31:30,920 --> 04:31:34,089
DOES NEED OTHER COMPUTE

7376
04:31:34,089 --> 04:31:35,357
INFRASTRUCTURES, PROBABLY NEEDS

7377
04:31:35,357 --> 04:31:37,426
COMPUTE INFRASTRUCTURES THAT ARE

7378
04:31:37,426 --> 04:31:39,395
NOT QUANTUM COMPUTING BECAUSE

7379
04:31:39,395 --> 04:31:41,430
THOSE ARE EVEN MORE ENERGY HOGS

7380
04:31:41,430 --> 04:31:45,634
THAN THE TOUGH WE'VE RUN RIGHT

7381
04:31:45,634 --> 04:31:46,202
NOW.

7382
04:31:46,202 --> 04:31:48,404
SO THAT'S PERHAPS A TANGENT, SO

7383
04:31:48,404 --> 04:31:49,705
JUST SAYING RUSS I KIND OF AGREE

7384
04:31:49,705 --> 04:31:51,340
WITH YOU AND YES, I AM THE

7385
04:31:51,340 --> 04:31:54,877
PERSON WHO HAS TO FIGURE OUT THE

7386
04:31:54,877 --> 04:31:56,178
BUDGET FOR THE COMPUTE WE WANT

7387
04:31:56,178 --> 04:31:59,081
TO DO ACROSS ALL OF THE

7388
04:31:59,081 --> 04:32:00,282
DIFFERENT RESOURCES ACROSS ALL

7389
04:32:00,282 --> 04:32:05,120
OF THE PROJECTS GOING ON AT

7390
04:32:05,120 --> 04:32:06,655
AMGEN SO IT IS A PERENNIAL

7391
04:32:06,655 --> 04:32:09,258
PROBLEM AND CONTINUES TO BE 1

7392
04:32:09,258 --> 04:32:10,793
EMPLOY I DON'T KNOW THAT IT'S A

7393
04:32:10,793 --> 04:32:17,633
SCIENTIFIC PROBLEM BUT IT IS A

7394
04:32:17,633 --> 04:32:18,667
LOGISTICAL AND OPERATIONAL

7395
04:32:18,667 --> 04:32:18,934
CHALLENGE.

7396
04:32:18,934 --> 04:32:21,804
>> I DON'T SEE IT THAT
WAY.

7397
04:32:21,804 --> 04:32:22,838
>> IT'S INTERESTING.

7398
04:32:22,838 --> 04:32:26,609
>> THE VAST MAJORITY OF
THE

7399
04:32:26,609 --> 04:32:28,577
ESPECIALLY THE MACHINE LEARNING

7400
04:32:28,577 --> 04:32:30,512
MODELS I'M BUILDING FOR PROPERTY

7401
04:32:30,512 --> 04:32:31,947
PREDICTION, I DON'T NEED THIS ON

7402
04:32:31,947 --> 04:32:33,749
MY LAB TOP TO DO A LOT OF THIS,

7403
04:32:33,749 --> 04:32:36,752
I KNOW THE BIGGEST THING WE'RE

7404
04:32:36,752 --> 04:32:38,854
LACKING IS DATA, YOU KNOW.

7405
04:32:38,854 --> 04:32:42,258
IT HURTS ME TO SEE ALL OF THESE

7406
04:32:42,258 --> 04:32:44,326
SMART KIDS AT MIT, OUT THERE

7407
04:32:44,326 --> 04:32:45,894
BUILDING MACHINE LEARNING MODELS

7408
04:32:45,894 --> 04:32:49,431
ON TERRIBLE DATA SETS.

7409
04:32:49,431 --> 04:32:50,366
I THINK IF ANYTHING, ESPECIALLY

7410
04:32:50,366 --> 04:32:52,101
IN TERPS OF JUST PROPERTY

7411
04:32:52,101 --> 04:32:54,503
PREDICTION FOR DRUG DISCOVERY,

7412
04:32:54,503 --> 04:32:55,271
THE MOST SIGNIFICANT THING WE

7413
04:32:55,271 --> 04:32:58,173
BEING DO IS HAVE A PUBLICLY

7414
04:32:58,173 --> 04:32:59,708
FUNDED EFFORT TO GENERATE HIGH

7415
04:32:59,708 --> 04:33:01,310
QUALITY DATA, IN A CONSISTENT

7416
04:33:01,310 --> 04:33:07,249
FASHION AND MAKE THIS PUBLICLY

7417
04:33:07,249 --> 04:33:07,516
AVAILABLE.

7418
04:33:07,516 --> 04:33:08,517
>> DON'T DISAGREE ABOUT
THE DATA

7419
04:33:08,517 --> 04:33:10,452
QUESTION, I WOULD SAY THERE'S A

7420
04:33:10,452 --> 04:33:12,521
RANGE OF MODELS THAT ARE

7421
04:33:12,521 --> 04:33:14,123
HAPPENING ON MY TEAM AND IF

7422
04:33:14,123 --> 04:33:22,531
SOMEONE IS TRYING TO BUILD A

7423
04:33:22,531 --> 04:33:24,333
MODEL PROCESSING I BUNCH OF

7424
04:33:24,333 --> 04:33:27,136
IMAGES COMING OUT OF SINGLE CELL

7425
04:33:27,136 --> 04:33:29,238
EXPERIMENTS AND LAYER ON TOP OF

7426
04:33:29,238 --> 04:33:31,407
THAT THE TRANSCRIPTOMICS AND THE

7427
04:33:31,407 --> 04:33:32,508
PROTEOMICS THAT BECOMES A BIG

7428
04:33:32,508 --> 04:33:34,143
COMPUTE PROBLEM FOR ME.

7429
04:33:34,143 --> 04:33:37,846
WHEREAS SOME OF THE OTHER MODELS

7430
04:33:37,846 --> 04:33:39,214
ABSOLUTELY, JUST, YOU KNOW LET

7431
04:33:39,214 --> 04:33:40,983
ME GIVE YOU MORE MEMORY FOR YOUR

7432
04:33:40,983 --> 04:33:43,519
LAPTOP SO THAT IT'S RUNNING

7433
04:33:43,519 --> 04:33:44,920
EFFICIENTLY ENOUGH FOR YOU SO

7434
04:33:44,920 --> 04:33:47,423
IT'S THAT SPAN, THAT BEING SAID,

7435
04:33:47,423 --> 04:33:49,058
YOU ARE EXACTLY RIGHT.

7436
04:33:49,058 --> 04:33:50,392
I CAN'T EVEN TRAIN THOSE MODELS

7437
04:33:50,392 --> 04:33:54,963
UNLESS I HAVE THE DATA FOR THEM

7438
04:33:54,963 --> 04:33:56,465
-- UNLESS I GENERATE THOSE

7439
04:33:56,465 --> 04:33:57,800
IMAGES AND I CAN'T IMENERATE

7440
04:33:57,800 --> 04:33:59,868
THOSE IMAGES IN THE RIGHT WAY IF

7441
04:33:59,868 --> 04:34:01,904
I HAVEN'T PROACTIVELY THOUGHT

7442
04:34:01,904 --> 04:34:04,073
ABOUT WHAT'S THE QUESTION I'M

7443
04:34:04,073 --> 04:34:05,274
TRYING TO ANSWER, WHAT ARE THE

7444
04:34:05,274 --> 04:34:07,209
DATA THAT WILL HELP ME TO ANSWER

7445
04:34:07,209 --> 04:34:10,479
IT, AND HOW DO WE CONSTRUCT AN

7446
04:34:10,479 --> 04:34:12,081
EXPERIMENT GOING BACK TO SOME OF

7447
04:34:12,081 --> 04:34:16,118
MY  COMMENTS THE OTHER DAY, THAT

7448
04:34:16,118 --> 04:34:18,454
IS AN APPROPRIATE MODEL AT THE

7449
04:34:18,454 --> 04:34:19,288
RIGHT LEVEL OF SPECIFICITY FOR

7450
04:34:19,288 --> 04:34:20,723
THE QUESTION WE'RE TRYING TO

7451
04:34:20,723 --> 04:34:25,661
ASK, SO I AGREE WITH YOU PAT. 

7452
04:34:25,661 --> 04:34:27,930
I JUST THINK THAT THERE'S SOME

7453
04:34:27,930 --> 04:34:30,032
SPAN ACROSS THE KINDS OF COMPUTE

7454
04:34:30,032 --> 04:34:33,001
WE DO AND REALLY IF ROMMIE WANTS

7455
04:34:33,001 --> 04:34:34,870
TO DO ALL THESE ATOM SIMULATIONS

7456
04:34:34,870 --> 04:34:36,405
AND TRAIN A BUNCH OF TOUGH ON

7457
04:34:36,405 --> 04:34:40,943
TOP OF IT SHE WILL NEED A BUNCH

7458
04:34:40,943 --> 04:34:41,477
OF [INDISCERNIBLE]

7459
04:34:41,477 --> 04:34:44,446
>> YEAH IT TAKES A LOT OF
BIG

7460
04:34:44,446 --> 04:34:44,680
COMPUTE.

7461
04:34:44,680 --> 04:34:46,148
JOEL ANY COMMENTS SOME.

7462
04:34:46,148 --> 04:34:48,917
>> NO, I WOULD AGREE WITH
MARTI,

7463
04:34:48,917 --> 04:34:50,085
JUST FROM THE VAST VARIETY OF

7464
04:34:50,085 --> 04:34:52,654
STUFF YOU WOULD DO IN TARGET ID,

7465
04:34:52,654 --> 04:34:54,256
PRECISION MEDICINE, COMPARED TO

7466
04:34:54,256 --> 04:34:59,528
ACTUAL MOLECULE DESIGN, YOU

7467
04:34:59,528 --> 04:35:02,765
KNOW, REALLY FAR OUTWEIGHS THE

7468
04:35:02,765 --> 04:35:04,466
RESULT AND AS CUTCHES TRANSITION

7469
04:35:04,466 --> 04:35:06,368
TO THE CLOUD AND BACK TO A

7470
04:35:06,368 --> 04:35:08,437
HYBRID MODEL I DO NOT ENVY HER

7471
04:35:08,437 --> 04:35:10,506
FOR TRYING TO FIGURE OUT THE

7472
04:35:10,506 --> 04:35:11,540
PUDGEET FOR EVERYTHING NTHERE

7473
04:35:11,540 --> 04:35:12,975
WAS TRANSITION OUT OF THAT, YOU

7474
04:35:12,975 --> 04:35:14,977
MEAN IN I DIDN'T REALIZE THAT

7475
04:35:14,977 --> 04:35:15,611
NYEAH.

7476
04:35:15,611 --> 04:35:16,311
>> YEAH, IT'S SOMETIMES
CHEAPER

7477
04:35:16,311 --> 04:35:17,880
TO DO I THINK THISS BACK ON

7478
04:35:17,880 --> 04:35:19,948
PRIMARY REVIEWER AND --

7479
04:35:19,948 --> 04:35:22,050
>> SO, SURE, YEAH, SO IT'S
LIKE

7480
04:35:22,050 --> 04:35:28,190
SIMILAR TO US.

7481
04:35:28,190 --> 04:35:28,524
>> OKAY.

7482
04:35:28,524 --> 04:35:32,728
>> YEAH, THERE'S ALWAYS
GOING TO

7483
04:35:32,728 --> 04:35:35,597
BE SECTIONS OF THIS, AND CLOUD

7484
04:35:35,597 --> 04:35:36,765
COMPUTING IS COMPUTING IT'S JUST

7485
04:35:36,765 --> 04:35:40,936
ON SOMEONE ELSE'S PREMISES.

7486
04:35:40,936 --> 04:35:41,870
>> SO IT'S NOT MY PAPER
WEIGHT

7487
04:35:41,870 --> 04:35:45,474
AT THE END OF IT.

7488
04:35:45,474 --> 04:35:45,774
>> RIGHT.

7489
04:35:45,774 --> 04:35:46,742
>> SOPHISTICATED A
QUESTION,

7490
04:35:46,742 --> 04:35:48,210
TOO, SO PAT YOU MENTIONED THAT

7491
04:35:48,210 --> 04:35:50,312
YOU KNOW AND IT'S TRUE, THINGS

7492
04:35:50,312 --> 04:35:54,216
ARE LIKE, THEY FEEL LIKE THEY'RE

7493
04:35:54,216 --> 04:35:57,820
MOVING, YOU SAID A HUNDRED MILES

7494
04:35:57,820 --> 04:36:00,689
AN HOUR AND I FEEL LIKE IT'S

7495
04:36:00,689 --> 04:36:02,191
FASTER, AND I WAS LISTENING TO A

7496
04:36:02,191 --> 04:36:04,860
TALK AND I WAS THINKING HOW IS

7497
04:36:04,860 --> 04:36:06,495
THIS HAPPENING  ON PLANET EARTH?

7498
04:36:06,495 --> 04:36:08,664
HOW DO YOU ALL KEEP UP WITH

7499
04:36:08,664 --> 04:36:09,765
THINGS MOVING FAST?

7500
04:36:09,765 --> 04:36:11,567
HOW ARE YOU HANDLING IT IN

7501
04:36:11,567 --> 04:36:14,369
KEEPING PACE WITH THE ADVANCES

7502
04:36:14,369 --> 04:36:15,003
BECAUSE IT'S CRAZY.

7503
04:36:15,003 --> 04:36:19,374
>> YEAH, I THINK IT'S --
IT'S

7504
04:36:19,374 --> 04:36:22,277
HAVING A GROUP OF PEOPLE WHO ARE

7505
04:36:22,277 --> 04:36:24,480
REALLY EXCITED ABOUT THE FIELD

7506
04:36:24,480 --> 04:36:26,615
AND WHO ARE CONSTANTLY READING,

7507
04:36:26,615 --> 04:36:28,083
GOING TO MEETINGS, TALKING TO

7508
04:36:28,083 --> 04:36:30,652
PEOPLE AND THEN HAVING

7509
04:36:30,652 --> 04:36:31,820
CONVERSATIONS --

7510
04:36:31,820 --> 04:36:32,921
>> FILTERING OUT --

7511
04:36:32,921 --> 04:36:34,289
>> YEAH, AMONG THAT GROUP.

7512
04:36:34,289 --> 04:36:34,590
YEAH.

7513
04:36:34,590 --> 04:36:35,257
BUT IT'S TOUGH.

7514
04:36:35,257 --> 04:36:39,428
YOU KNOW BECAUSE THERE'S

7515
04:36:39,428 --> 04:36:40,529
PROBABLY MACHINE LEARNING AND

7516
04:36:40,529 --> 04:36:41,797
DRUG DISCOVERY PAPERS THAT SHOW

7517
04:36:41,797 --> 04:36:43,332
UP ON ARCHIVE EVERY WEEK SO

7518
04:36:43,332 --> 04:36:45,200
YEAH, ASTERISKS JUST GETTING

7519
04:36:45,200 --> 04:36:49,605
RIDICULOUS TO TRY TO KEEP UP.

7520
04:36:49,605 --> 04:36:51,006
>> YEAH, YEAH.

7521
04:36:51,006 --> 04:36:52,574
>> ON OUR LAB SLACK WE
HAVE AN

7522
04:36:52,574 --> 04:36:53,275
INTERESTING PAPERS CHANNEL AND

7523
04:36:53,275 --> 04:36:55,410
USED TO TRY TO GIVE A 1 LINER

7524
04:36:55,410 --> 04:36:57,913
LIKE AN IMPRESSION OF THE PAPER

7525
04:36:57,913 --> 04:36:59,281
TO HELP THE STUDENTS LIKE

7526
04:36:59,281 --> 04:37:02,951
PRACTICE THEIR OWN CRITICAL AND

7527
04:37:02,951 --> 04:37:07,823
FORGET ABOUT IT, LIKE RIGHT NOW

7528
04:37:07,823 --> 04:37:09,291
IT'S A SCROLLING, IT'S A FEED

7529
04:37:09,291 --> 04:37:10,526
AND SO FOCUS AND KNOWING WHAT

7530
04:37:10,526 --> 04:37:12,661
WERE YOU AWARE TRYING TO DOA

7531
04:37:12,661 --> 04:37:14,129
BECOMES MORE IMPORTANT BECAUSE

7532
04:37:14,129 --> 04:37:16,365
YOU CAN'T BE JUST OPPORTUNISTIC

7533
04:37:16,365 --> 04:37:18,233
BECAUSE YOU WILL BE CONTEXT

7534
04:37:18,233 --> 04:37:19,134
SWITCHING EVERY TIN MINUTES SO

7535
04:37:19,134 --> 04:37:20,269
YOU HAVE TO HAVE A PLAN, YOU

7536
04:37:20,269 --> 04:37:22,037
HAVE TO REALIZE THAT THE PLAN

7537
04:37:22,037 --> 04:37:23,705
MAY HAVE TO CHANGE BUT THE FOCUS

7538
04:37:23,705 --> 04:37:25,707
HAS TO BE THERE AND THIS IS LIKE

7539
04:37:25,707 --> 04:37:28,143
FOR Ph.D. STUDENTS BUT ALSO

7540
04:37:28,143 --> 04:37:29,945
FOR START UPS AND FOR BIG PHARMA

7541
04:37:29,945 --> 04:37:31,680
THE FOCUS AND THE PLAN, YOU HAVE

7542
04:37:31,680 --> 04:37:33,549
TO BELIEVE IN IT AND STICK TO

7543
04:37:33,549 --> 04:37:35,050
IT, EVERY NOW AND THEN THERE

7544
04:37:35,050 --> 04:37:37,452
WILL BE A VERY DISRUPTIVE

7545
04:37:37,452 --> 04:37:38,387
TECHNOLOGY BUT MOST TECHNOLOGIES

7546
04:37:38,387 --> 04:37:44,493
ARE ONLY A LITTLE DISRUPTIVE TO

7547
04:37:44,493 --> 04:37:45,193
THE PLAN.

7548
04:37:45,193 --> 04:37:47,262
BUT SOMETIMES THEY BLOW YOUR

7549
04:37:47,262 --> 04:37:48,630
PLAN OUT OF THE WATER AND THAT'S

7550
04:37:48,630 --> 04:37:50,365
THE 1 YOU WORRY ABOUT, RIGHT IN.

7551
04:37:50,365 --> 04:37:51,633
>> YEAH, THERE WAS A
QUESTION

7552
04:37:51,633 --> 04:37:53,535
FROM THE AUDIENCE THAT I DID --

7553
04:37:53,535 --> 04:37:56,772
THAT I DID EXTRACT, AND I'M

7554
04:37:56,772 --> 04:37:58,407
SORRY, I'M NOT -- I'M NOT SURE

7555
04:37:58,407 --> 04:38:03,412
WHAT I ASKED IT MIGHT HAVE BEEN

7556
04:38:03,412 --> 04:38:06,949
AN  ANONYMOUS ATTENDEE WHAT THE

7557
04:38:06,949 --> 04:38:08,917
QUESTION WAS WHAT ARE THE

7558
04:38:08,917 --> 04:38:14,523
CHALLENGES YOU CAN SEE ON THIS

7559
04:38:14,523 --> 04:38:16,291
FIELD AS A YOUNG SCIENTIST?

7560
04:38:16,291 --> 04:38:18,560
I CAN PUT ON MY PROFESSOR HAT

7561
04:38:18,560 --> 04:38:21,396
AND THE STUDENTS AND DIFFERENT

7562
04:38:21,396 --> 04:38:22,531
DISCIPLINES AND THINGS THAT YOU

7563
04:38:22,531 --> 04:38:24,099
FEEL LIKE ARE LACKING OR NEED TO

7564
04:38:24,099 --> 04:38:26,168
BE ADDRESSED THAT ARE MAYBE A

7565
04:38:26,168 --> 04:38:27,336
LITTLE BIT DIFFERENT NOW OR ARE

7566
04:38:27,336 --> 04:38:35,344
THERE -- WHAT ARE THE THOUGHTS

7567
04:38:35,344 --> 04:38:35,844
ON THAT?

7568
04:38:35,844 --> 04:38:38,113
STHRKS 1 THING THAT WORRIES ME

7569
04:38:38,113 --> 04:38:40,215
ARE DATA SCIENTISTS THAT ARE NOT

7570
04:38:40,215 --> 04:38:41,350
CONNECKED TO A DOMAIN OF

7571
04:38:41,350 --> 04:38:41,617
EXPERTISE.

7572
04:38:41,617 --> 04:38:42,818
IF YOU WANT TO BE IN THE FIELD,

7573
04:38:42,818 --> 04:38:46,722
IT'S GO TO KNOW A LOT OF CHEMIST

7574
04:38:46,722 --> 04:38:48,323
RADIOY OR BIOLOGY, AND SOME

7575
04:38:48,323 --> 04:38:49,758
PEOPLE ARE BEING SOLD THIS

7576
04:38:49,758 --> 04:38:51,760
GENERAL PURPOSE -- I CAN'T

7577
04:38:51,760 --> 04:38:53,028
IMAGINE THERE'S ANY DOMAIN WHERE

7578
04:38:53,028 --> 04:38:55,931
YOU WILL BE HAPPY HIRING A KIND

7579
04:38:55,931 --> 04:38:58,033
OF HIRED GUN DATA SCIENTIST WHO

7580
04:38:58,033 --> 04:38:59,234
DOESN'T KNOW ANYTHING ABOUT YOUR

7581
04:38:59,234 --> 04:39:00,836
DOMAIN BUT IT'S FOR SURE, I

7582
04:39:00,836 --> 04:39:02,904
WON'T SPEAK FOR OTHER DOMAINS IN

7583
04:39:02,904 --> 04:39:05,140
BIOLOGY, CHEMIST RADIOY AND

7584
04:39:05,140 --> 04:39:06,308
MEDICINE, FOR SURE, YOU'RE NOT

7585
04:39:06,308 --> 04:39:13,415
THAT USEFUL IF YOU'RE JUST GOING

7586
04:39:13,415 --> 04:39:22,357
TO TURN THE CRANK -- ABOUT

7587
04:39:22,357 --> 04:39:26,094
SOMETHING THAT IS IN THE WORMED.

7588
04:39:26,094 --> 04:39:26,528
>> RIGHT.

7589
04:39:26,528 --> 04:39:28,163
MAYBE COMPLETELY, I THINK, I

7590
04:39:28,163 --> 04:39:29,531
ALWAYS KIND OF URGE PEOPLE TO

7591
04:39:29,531 --> 04:39:33,402
SIT AT THE CENTER OF A THEN

7592
04:39:33,402 --> 04:39:34,870
DIAGRAM CONSISTISTING OF YES,

7593
04:39:34,870 --> 04:39:36,938
YOU NEED THE HACKING SKILLS, YOU

7594
04:39:36,938 --> 04:39:38,974
NEED TO UNDERSTAND MACHINE

7595
04:39:38,974 --> 04:39:40,375
LEARNING, YOU DEFINITELY NEED

7596
04:39:40,375 --> 04:39:41,176
THAT DOMAIN EXPERTISE AND THEN

7597
04:39:41,176 --> 04:39:44,379
YOU NEED TO KNOW ENOUGH MATH

7598
04:39:44,379 --> 04:39:48,483
STATISTICS TO KNOW WHETHER

7599
04:39:48,483 --> 04:39:49,418
YOU'RE FOOLING YOURSELF EMPLOY.

7600
04:39:49,418 --> 04:39:51,053
>> I KNEW YOU WERE GOING
TO SAY

7601
04:39:51,053 --> 04:39:51,453
THAT.

7602
04:39:51,453 --> 04:39:52,587
>> I'M GOING TO BE HONEST
WITH

7603
04:39:52,587 --> 04:39:55,257
YOU THAT EVEN AS THE PACE OF

7604
04:39:55,257 --> 04:39:57,759
LIFE CHANGES, EVEN AS NEW, AS 6

7605
04:39:57,759 --> 04:39:58,727
BILLION PAPERS COME OUT EVERY

7606
04:39:58,727 --> 04:40:05,133
WEEK IN ALL OF THAT, THE CORE

7607
04:40:05,133 --> 04:40:08,937
SKILLS THAT OUR NEW HIREEES NEED

7608
04:40:08,937 --> 04:40:10,172
TO HAVE, HAVEN'T CHANGED ANY,

7609
04:40:10,172 --> 04:40:10,806
RIGHT?

7610
04:40:10,806 --> 04:40:15,210
I STILL WANT THAT PERSON WHO IS

7611
04:40:15,210 --> 04:40:16,745
AN EXPERT IN SOMETHING BUT HAS

7612
04:40:16,745 --> 04:40:18,046
THE ABILITY TO LEARN NEW THINGS

7613
04:40:18,046 --> 04:40:19,815
AND TO TEACH TO OTHERS.

7614
04:40:19,815 --> 04:40:23,018
I STILL WANT THAT PERSON WHO IS

7615
04:40:23,018 --> 04:40:25,487
RIGOROUS ENOUGH IN THEIR

7616
04:40:25,487 --> 04:40:30,392
THINKING AND NOT FOOLING

7617
04:40:30,392 --> 04:40:33,095
THEMSELVES ENOUGH IN

7618
04:40:33,095 --> 04:40:34,429
INTERPRETING THEIR RESULTS THAT

7619
04:40:34,429 --> 04:40:36,398
THEY WILL EVEN MAKE PAT WALTERS

7620
04:40:36,398 --> 04:40:38,233
HAPPY, I STILL WANT SOMEONE WHO

7621
04:40:38,233 --> 04:40:43,238
CAN COME IN AND COMMUNICATE

7622
04:40:43,238 --> 04:40:44,206
THEIR CONCEPTS AND THEIR DEPTH

7623
04:40:44,206 --> 04:40:46,208
OF KNOWLEDGE TO A NONEXPERT

7624
04:40:46,208 --> 04:40:48,477
AUDIENCE IN A USEFUL WAY.

7625
04:40:48,477 --> 04:40:50,779
SO, AT THE ROOT OF IT, I DON'T

7626
04:40:50,779 --> 04:40:53,281
THINK THE CORE BEHAVIORS HAVE

7627
04:40:53,281 --> 04:40:56,017
CHANGED AT ALL IN WHAT WE NEED.

7628
04:40:56,017 --> 04:40:58,120
THERE'S MAYBE SOME ADDITIONS

7629
04:40:58,120 --> 04:41:03,425
AROUND FOR ALL OF US, INCLUDING

7630
04:41:03,425 --> 04:41:05,327
-- HERE AROUND HOW DO WE RIDE

7631
04:41:05,327 --> 04:41:10,398
THE WAVE OF CHANGE AND MAINTAIN

7632
04:41:10,398 --> 04:41:10,699
RESILIENCE?

7633
04:41:10,699 --> 04:41:12,667
BUT IT'S THE SAME SET OF SKILLS

7634
04:41:12,667 --> 04:41:14,936
THAT I'M LOOKING FOR AND I

7635
04:41:14,936 --> 04:41:15,771
COULDN'T AGREE MORE WITH RUSS

7636
04:41:15,771 --> 04:41:19,541
AND PAT THAT YOU DO WANT -- YOU

7637
04:41:19,541 --> 04:41:22,878
DO WANT TO BE ABLE TO KEEP

7638
04:41:22,878 --> 04:41:24,913
CHENLIST RADIOY TO A MEDICINAL

7639
04:41:24,913 --> 04:41:27,449
CHEMIST IF YOU WANT TO DO THAT.

7640
04:41:27,449 --> 04:41:28,083
>> JOEL, YOU UNMUTED
YOURSELF

7641
04:41:28,083 --> 04:41:29,918
WHAT DO YOU THINK?

7642
04:41:29,918 --> 04:41:32,187
>> I MEAN I WOULD AGREE
WITH

7643
04:41:32,187 --> 04:41:33,722
EVERYBODY ESPECIALLY RESONATE

7644
04:41:33,722 --> 04:41:35,524
WITH WHAT MARTI WAS SAYING

7645
04:41:35,524 --> 04:41:37,926
AROUND THE CORE SKILLS WE STILL

7646
04:41:37,926 --> 04:41:39,427
SEE THE COMPUTATIONAL SCIENTISTS

7647
04:41:39,427 --> 04:41:41,463
BEING THAT TRANSLATOR BETWEEN

7648
04:41:41,463 --> 04:41:43,865
THE DEEP, DEEP, DATA ENGINEERS

7649
04:41:43,865 --> 04:41:45,200
AND TECHNICAL EXPERTISES TO BE

7650
04:41:45,200 --> 04:41:46,201
ABLE TO COMMUNICATE REQUIREMENTS

7651
04:41:46,201 --> 04:41:47,702
AND THEN ALSO WORKING WITH THE

7652
04:41:47,702 --> 04:41:49,538
AUTOMATION TEAMS, WORKING WITH

7653
04:41:49,538 --> 04:41:50,505
THE ACTUAL EXPERIMENTAL

7654
04:41:50,505 --> 04:41:51,807
SCIENTISTS AND REALLY BEING THAT

7655
04:41:51,807 --> 04:41:52,974
KIND OF HUB, SO IF YOU DON'T

7656
04:41:52,974 --> 04:41:54,543
KNOW SOMETHING THAT YOU'RE

7657
04:41:54,543 --> 04:41:56,278
TALKING ABOUT, THAT WILL BECOME

7658
04:41:56,278 --> 04:41:57,846
OBVIOUS AND THAT'S A PRETTY POOR

7659
04:41:57,846 --> 04:42:00,081
FIT FOR THE JOB.

7660
04:42:00,081 --> 04:42:02,818
AND TO GET TO THE POINT AROUND

7661
04:42:02,818 --> 04:42:05,320
SORT OF THE GENERAL DATA

7662
04:42:05,320 --> 04:42:06,721
SCIENTISTS AND MACHINE LEARNING

7663
04:42:06,721 --> 04:42:07,622
TESTIFIES I CAN'T COUNT THE

7664
04:42:07,622 --> 04:42:12,327
NUMBER OF TIMES WHO DEPLOYED

7665
04:42:12,327 --> 04:42:13,495
MODELS AND ASKED THEM WHAT THE

7666
04:42:13,495 --> 04:42:14,930
IMPACT OF THAT WAS, AND THEY

7667
04:42:14,930 --> 04:42:17,666
CAN'T TELL ME BECAUSE THEY WERE

7668
04:42:17,666 --> 04:42:19,067
JUST TOLD TO BUILD THE MODEL AND

7669
04:42:19,067 --> 04:42:20,368
DON'T CARE ABOUT WHAT HAPPENS

7670
04:42:20,368 --> 04:42:21,937
AFTERWARDS SO THE PEOPLE WHO

7671
04:42:21,937 --> 04:42:23,138
HAVE THE SUBJECT MATTER

7672
04:42:23,138 --> 04:42:24,339
EXPERTISE OR AT LEAST THE

7673
04:42:24,339 --> 04:42:26,641
AGILITY TO TALK TO PROJECT

7674
04:42:26,641 --> 04:42:28,343
MANAGER TEAMS AND TALK ABOUT

7675
04:42:28,343 --> 04:42:29,477
DRUG DISCOVERY IS, AND TO

7676
04:42:29,477 --> 04:42:31,880
UNDERSTAND THE ROLE AND MAKE AN

7677
04:42:31,880 --> 04:42:35,884
IMPACT IN THINGS TOTALLY KEY TO

7678
04:42:35,884 --> 04:42:36,184
SUCCESS.

7679
04:42:36,184 --> 04:42:42,324
>> YEAH, YEAH, I AGREE.

7680
04:42:42,324 --> 04:42:43,792
SO KEEPING MY PROFESSOR HAT ON,

7681
04:42:43,792 --> 04:42:45,760
MAKING DRUG SYSTEM NOT REALLY

7682
04:42:45,760 --> 04:42:48,763
THE PURSUE OF ACADEMIA, RIGHT WE

7683
04:42:48,763 --> 04:42:50,799
CAN CONTRIBUTE TO LITTLE BITS OF

7684
04:42:50,799 --> 04:42:52,100
THIS BUT THAT'S AN INDUSTRIAL

7685
04:42:52,100 --> 04:42:53,168
SCALE THING, YOU KNOW IN THE

7686
04:42:53,168 --> 04:42:55,437
PAST, I MEAN TYPICALLY ACADEMICS

7687
04:42:55,437 --> 04:42:57,873
MIGHT BE DEVELOPING NEW METHODS,

7688
04:42:57,873 --> 04:42:59,774
RIGHT, DOING MORE SORT OF SLOWER

7689
04:42:59,774 --> 04:43:01,877
LONG TIME KALE PROGEC TYPE WORK,

7690
04:43:01,877 --> 04:43:05,480
MAYBE FOCUSING ON LEARNING

7691
04:43:05,480 --> 04:43:06,648
FUNDAMENTAL BIOLOGY, YOU KNOW,

7692
04:43:06,648 --> 04:43:09,084
REGARDING METHODS I HAD AN

7693
04:43:09,084 --> 04:43:11,152
INTERESTING SORT OF TWITTER CHAT

7694
04:43:11,152 --> 04:43:11,786
WITH BJ [INDISCERNIBLE].

7695
04:43:11,786 --> 04:43:13,889
WHO WAS MAKING THE ARGUMENT THAT

7696
04:43:13,889 --> 04:43:15,223
WAS SO MUCH GREAT METHOD STUFF

7697
04:43:15,223 --> 04:43:16,625
HAPPENING IN INDUSTRY YOU SHOULD

7698
04:43:16,625 --> 04:43:17,826
COME OVER BECAUSE YOUR BUDGET'S

7699
04:43:17,826 --> 04:43:19,294
GOING TO BE BIGGER AND YOU COULD

7700
04:43:19,294 --> 04:43:20,795
DO MORE THINGS WHICH I THOUGHT

7701
04:43:20,795 --> 04:43:24,366
WAS REALLY INTERESTING BUT SO I

7702
04:43:24,366 --> 04:43:26,067
JUST WANTED TO ASK LIKE SO HOW

7703
04:43:26,067 --> 04:43:28,904
DO YOU VIEW THE ROLE OF ACADEMIA

7704
04:43:28,904 --> 04:43:29,671
IN THIS SPACE.

7705
04:43:29,671 --> 04:43:31,106
ARE THERE PARTICULAR THINGS YOU

7706
04:43:31,106 --> 04:43:34,276
LOOK FORWARDS US FOR OTHER THAN

7707
04:43:34,276 --> 04:43:35,644
THE PIPELINE OF LIKE YOUR NEXT

7708
04:43:35,644 --> 04:43:38,513
TEAM OR IN TERPS OF LIKE METHODS

7709
04:43:38,513 --> 04:43:45,887
AND THOUGHT DIRECTION?

7710
04:43:45,887 --> 04:43:46,855
OR NO?

7711
04:43:46,855 --> 04:43:47,689
>> I THINK IN ESPECIALLY
IN

7712
04:43:47,689 --> 04:43:49,124
TERMS OF MACHINE LEARNING, I

7713
04:43:49,124 --> 04:43:52,794
LOVE THE FACT THAT ACADEMIA IS

7714
04:43:52,794 --> 04:43:54,229
DOING A LOT OF FUNDAMENTAL

7715
04:43:54,229 --> 04:43:55,297
RESEARCH, YOU KNOW PEOPLE ARE

7716
04:43:55,297 --> 04:43:56,364
DOING RESEARCH ON THINGS LIKE

7717
04:43:56,364 --> 04:43:58,767
HOW CAN I USE A DIFFUSION MODEL

7718
04:43:58,767 --> 04:44:00,402
TO GENERATE A MOLECULE?

7719
04:44:00,402 --> 04:44:06,641
BUT THEN, YOU KNOW, I FEEL LIKE

7720
04:44:06,641 --> 04:44:08,276
WE IN INDUSTRY HAVE A

7721
04:44:08,276 --> 04:44:09,811
RESPONSIBILITY TO TRY THAT OUT

7722
04:44:09,811 --> 04:44:10,712
AND PROVIDE FEEDBACK AND 1 OF

7723
04:44:10,712 --> 04:44:12,747
THE THINGS I TROO TO DO AT LEAST

7724
04:44:12,747 --> 04:44:14,049
IF SOMEBODY WILL PUBLISH A PAPER

7725
04:44:14,049 --> 04:44:16,651
AND THEN THEY WILL HAVE A GET--I

7726
04:44:16,651 --> 04:44:18,453
REPOE WITH THEIR CODE I THINK

7727
04:44:18,453 --> 04:44:22,057
IT'S UNDUM BENT ON US THEN TO

7728
04:44:22,057 --> 04:44:22,857
SAY THAT'S GREAT.

7729
04:44:22,857 --> 04:44:24,125
LET'S TRY THAT OUT AND MAYBE WE

7730
04:44:24,125 --> 04:44:25,760
CAPTAIN GIVE YOU BACK THE EXACT

7731
04:44:25,760 --> 04:44:28,697
STRUCTURES BUT MAYBE WE CAN GIVE

7732
04:44:28,697 --> 04:44:32,033
YOU BACK ADVICE ON HOW YOU CAN

7733
04:44:32,033 --> 04:44:34,235
MAKE THIS MORE PRACTICAL USE

7734
04:44:34,235 --> 04:44:34,569
GOING FORWARD.

7735
04:44:34,569 --> 04:44:39,140
>> I DO THINK THERE'S SOME
OF

7736
04:44:39,140 --> 04:44:41,876
THE INNOVATION AROUND

7737
04:44:41,876 --> 04:44:42,877
EXPERIMENTAL TECHNIQUES THAT

7738
04:44:42,877 --> 04:44:45,847
GENERATE CERTAIN KINDS OF DATA

7739
04:44:45,847 --> 04:44:47,615
THAT THEN CAN FEED MODELS WHERE

7740
04:44:47,615 --> 04:44:52,420
THERE'S A BIG ASPECT OF THAT.

7741
04:44:52,420 --> 04:44:54,956
THAT FOSTERS IDEAS INSIDE OF

7742
04:44:54,956 --> 04:44:56,891
ACADEMIA, THAT WE DON'T

7743
04:44:56,891 --> 04:44:58,159
NECESSARILY HAVE THE DAY IN AND

7744
04:44:58,159 --> 04:45:03,064
DAY OUT  TIME TO DO IN INDUSTRY.

7745
04:45:03,064 --> 04:45:04,599
>> YEAH, OR MAYBE THE
PARTICULAR

7746
04:45:04,599 --> 04:45:06,868
EXPERTISE AND SO FORTH WITH

7747
04:45:06,868 --> 04:45:10,005
DIFFERENT MICROSCOPIES AND SO

7748
04:45:10,005 --> 04:45:10,872
ON.

7749
04:45:10,872 --> 04:45:11,306
>> AND SO --

7750
04:45:11,306 --> 04:45:13,608
>> I WOULD SAY THERE'S --
NI

7751
04:45:13,608 --> 04:45:15,310
DIDN'T MEAN TO INTERRUPT, JOEL,

7752
04:45:15,310 --> 04:45:16,011
YOU GO AHEAD.

7753
04:45:16,011 --> 04:45:18,480
>> NO, NO, I WAS GOING TO
GO --

7754
04:45:18,480 --> 04:45:20,615
I FEEL LIKE I'M AGREEING WITH

7755
04:45:20,615 --> 04:45:23,752
YOU A LOT MARTI, AND ALSO EXPABD

7756
04:45:23,752 --> 04:45:26,254
ON THE ACADEMIA MODEL ISN'T ALSO

7757
04:45:26,254 --> 04:45:28,023
JUST LIKE, OH PARTNERSHIP WITH 1

7758
04:45:28,023 --> 04:45:30,392
LAB, RIGHT IN THERE'S ALL KINDS

7759
04:45:30,392 --> 04:45:31,626
OF CONSORTIA OPINION AN ACTUAL

7760
04:45:31,626 --> 04:45:32,560
UNIVERSITY WHERE WIEWR ALL

7761
04:45:32,560 --> 04:45:33,762
WORKING TOGETHER AND

7762
04:45:33,762 --> 04:45:34,429
CONTRIBUTING DIFFERENT METHODS

7763
04:45:34,429 --> 04:45:37,465
THAT MAY BE, YOU KNOW, A

7764
04:45:37,465 --> 04:45:38,700
PARTICULAR COMPANY WOULD WANT TO

7765
04:45:38,700 --> 04:45:39,801
INVEST IN THAT, BECAUSE THE

7766
04:45:39,801 --> 04:45:41,369
GROUP THING THAT YOU'RE PUTTING

7767
04:45:41,369 --> 04:45:41,636
TOGETHER.

7768
04:45:41,636 --> 04:45:45,407
SO I THINK THERE'S A BUNCH OF

7769
04:45:45,407 --> 04:45:46,408
DIFFERENT MODELS, ESPECIALLY AS

7770
04:45:46,408 --> 04:45:48,043
TECHNOLOGY GETS BIGGER AND AS

7771
04:45:48,043 --> 04:45:52,580
RESEARCH GETS MORE

7772
04:45:52,580 --> 04:45:54,949
COLLABORATIVE. 

7773
04:45:54,949 --> 04:45:56,484
>> ALL RIGHT, COOL.

7774
04:45:56,484 --> 04:46:04,159
LET'S SEE, WHAT ELSE.

7775
04:46:04,159 --> 04:46:05,994
ON TEAMS.

7776
04:46:05,994 --> 04:46:07,762
IT'S EVIDENT, DRUG DISCOVERY IS

7777
04:46:07,762 --> 04:46:10,999
A HUGE TEAM, IT'S TEAM SCIENCE

7778
04:46:10,999 --> 04:46:12,000
ACROSS THE BOARD.

7779
04:46:12,000 --> 04:46:13,501
DO YOU HAVE AND THAT'S SO

7780
04:46:13,501 --> 04:46:14,769
IMPORTANT AND IT'S SO HARD, IT'S

7781
04:46:14,769 --> 04:46:18,373
NOT ALWAYS THAT PEOPLE COME

7782
04:46:18,373 --> 04:46:20,375
TOGETHER AND THEY GEL IN A WAY

7783
04:46:20,375 --> 04:46:21,709
WHERE EVERYBODY'S RUNNING IN THE

7784
04:46:21,709 --> 04:46:23,511
SAME DIRECTION, IF YOU CAN GET

7785
04:46:23,511 --> 04:46:24,879
THAT, CULTIVATE THAT IN ANY

7786
04:46:24,879 --> 04:46:27,315
ORGANIZATION I THINK WHETHER

7787
04:46:27,315 --> 04:46:28,616
IT'S ACADEMIC, LAB, OWN GROUP,

7788
04:46:28,616 --> 04:46:32,120
WHOLE PLACE, YOU KNOW IT CAN BE

7789
04:46:32,120 --> 04:46:34,689
AWESOME, DO YOU HAVE THOUGHTS OR

7790
04:46:34,689 --> 04:46:37,292
ADVICE FOR FOLKS, LIKE WHAT YOU

7791
04:46:37,292 --> 04:46:43,364
LEARNED ABOUT MAKING THOSE

7792
04:46:43,364 --> 04:46:43,798
STRONG TEAMS?

7793
04:46:43,798 --> 04:46:45,333
>> WOULD YOU LIKE TO TALK
FOR

7794
04:46:45,333 --> 04:46:55,310
SEVERAL HOURS ON END ON THIS

7795
04:46:55,310 --> 04:46:55,810
TOPIC?

7796
04:46:55,810 --> 04:46:56,077
[LAUGHTER]

7797
04:46:56,077 --> 04:46:57,679
I JUST THINK IT'S SUCH AN

7798
04:46:57,679 --> 04:46:58,613
IMPORTANT TOPIC.

7799
04:46:58,613 --> 04:47:01,416
MANY OF YOU HAVE HEARD ME WHINE

7800
04:47:01,416 --> 04:47:03,985
ABOUT THAT'S WHY I CAME BACK TO

7801
04:47:03,985 --> 04:47:06,054
INDUSTRY IS BECAUSE I MISSED

7802
04:47:06,054 --> 04:47:07,889
WORKING ON INTERDEPENDENT TEAMS

7803
04:47:07,889 --> 04:47:09,491
THAT COULD RELY ON EACH OTHER

7804
04:47:09,491 --> 04:47:10,992
AND WERE RELIABLE FOR EACH

7805
04:47:10,992 --> 04:47:13,194
OTHER.

7806
04:47:13,194 --> 04:47:15,964
>> TRUST.

7807
04:47:15,964 --> 04:47:17,132
>> THAT IS THE FUNCTION
THAT

7808
04:47:17,132 --> 04:47:17,499
NEEDS TO HAPPEN.

7809
04:47:17,499 --> 04:47:19,167
ONE OF IT IS JUST A MIND SET

7810
04:47:19,167 --> 04:47:19,934
THING.

7811
04:47:19,934 --> 04:47:21,336
I HAVE SOME OF THEM IN THE ROOM

7812
04:47:21,336 --> 04:47:24,305
WILL HAVE HEARD MY ANALOGY

7813
04:47:24,305 --> 04:47:25,673
BEFORE OF WAVE PARTICLE DUALITY

7814
04:47:25,673 --> 04:47:33,448
AS THE WAY WE SHOULD APPROACH

7815
04:47:33,448 --> 04:47:35,049
OUR LIFE, AND BEING MINDFUL

7816
04:47:35,049 --> 04:47:38,219
ABOUT WHEN I'M A PARTICLE

7817
04:47:38,219 --> 04:47:39,087
CONTRIBUTING MY EXPERTISE VERSUS

7818
04:47:39,087 --> 04:47:44,259
WHEN I'M A MEMBER OF A TEAM

7819
04:47:44,259 --> 04:47:45,260
INTEGRATING TOGETHER.

7820
04:47:45,260 --> 04:47:47,262
I ALSO THINK IT'S IMPORTANT AND

7821
04:47:47,262 --> 04:47:48,630
MOST COMPANIES SPEND A LOT OF

7822
04:47:48,630 --> 04:47:51,833
TIME ON SKILLS BUILDING OR

7823
04:47:51,833 --> 04:47:54,169
BUILDING TEAMS, HOW CAN I BE

7824
04:47:54,169 --> 04:47:56,971
AWARE OF MY PREFERENCES, MY

7825
04:47:56,971 --> 04:47:59,674
WIMBERLYS OF PAIRING MYSELF IN

7826
04:47:59,674 --> 04:48:01,075
THE WORLD, HOW AM I AWARE OF

7827
04:48:01,075 --> 04:48:02,343
THAT IN RELATIONSHIP TO THE

7828
04:48:02,343 --> 04:48:07,582
OTHER PEOPLE AROUND ME AND THEN

7829
04:48:07,582 --> 04:48:08,783
BEYOND THAT SKILLS TRAINING

7830
04:48:08,783 --> 04:48:11,119
AROUND WHAT ARE ORGANIZATIONAL

7831
04:48:11,119 --> 04:48:12,987
AND PROCESS TOOLS WE CAN USE TO

7832
04:48:12,987 --> 04:48:15,190
MAKE SURE THAT OUR MATRIX TEAMS

7833
04:48:15,190 --> 04:48:17,959
KNOW WHAT THE GOAL IS WHAT

7834
04:48:17,959 --> 04:48:19,394
THEY'RE CONTRIBUTING AND HOW IT

7835
04:48:19,394 --> 04:48:21,196
WINDS UP TOGETHER AND SO FORTH.

7836
04:48:21,196 --> 04:48:22,864
AND SO, IT REALLY IS A TOPIC

7837
04:48:22,864 --> 04:48:25,800
THAT I COULD TALK ON FOR WAY TOO

7838
04:48:25,800 --> 04:48:28,903
LONG, SOIME GOING TO SHUT UP AND

7839
04:48:28,903 --> 04:48:30,738
TURN IT OVER TO MY COLLEAGUES

7840
04:48:30,738 --> 04:48:36,711
PAT AND JOEL IN PARTICULAR SINCE

7841
04:48:36,711 --> 04:48:39,214
WE COME FROM AN INDUSTRY

7842
04:48:39,214 --> 04:48:42,617
PERSPECTIVE ABOUT HOW YOU THINK

7843
04:48:42,617 --> 04:48:43,952
IT AFFECTS TEAMS?

7844
04:48:43,952 --> 04:48:48,690
PAT YOU'VE BEEN IN SEVERAL

7845
04:48:48,690 --> 04:48:55,129
SIZES, JOEL, YOU'VE BEEN IN THE

7846
04:48:55,129 --> 04:48:56,397
MONSTROSITY OF GSK ALL YOUR LIFE

7847
04:48:56,397 --> 04:48:58,833
SO HOW DOES IT PLAY OUT?

7848
04:48:58,833 --> 04:49:01,603
>> I AGREE WITH EVERYTHING
MARTI

7849
04:49:01,603 --> 04:49:05,607
SAID, THERE'S SO MUCH THAT'S

7850
04:49:05,607 --> 04:49:06,941
COMMUNICATION, IT'S ALL ABOUT

7851
04:49:06,941 --> 04:49:08,676
APPRECIATION, IT'S ABOUT

7852
04:49:08,676 --> 04:49:09,544
CRAFTSMANSHIP AND APPRECIATING

7853
04:49:09,544 --> 04:49:10,912
THE CRAFTSMAN SHN OF OTHERS BUT

7854
04:49:10,912 --> 04:49:12,880
I THINK MORE THAN ANYTHING, AT

7855
04:49:12,880 --> 04:49:14,849
LEAST FOR ME, IT'S BEEN JUST

7856
04:49:14,849 --> 04:49:16,284
TRYING TO FIND PEOPLE WITH THE

7857
04:49:16,284 --> 04:49:23,224
RIGHT MIND SET AND BUILDING THE

7858
04:49:23,224 --> 04:49:23,791
RIGHT TEAM.

7859
04:49:23,791 --> 04:49:26,527
AND TRYING TO BUILD A TEAM OF

7860
04:49:26,527 --> 04:49:29,897
PEOPLE WHO NOT ONLY APPRECIATIVE

7861
04:49:29,897 --> 04:49:31,833
OF OTHER ASPECTS BUT HIRING

7862
04:49:31,833 --> 04:49:36,237
ENTHUSIASTIC PEOPLE.

7863
04:49:36,237 --> 04:49:40,141
HIRINGMEDICINAL CHEMISTS WHO ARE

7864
04:49:40,141 --> 04:49:41,009
ENTHUSIASTIC ABOUT PHARMACOLOGY.

7865
04:49:41,009 --> 04:49:42,443
IT'S GREAT IF YOU CAN CREATE

7866
04:49:42,443 --> 04:49:43,077
THIS SITUATION WHERE YOU HAVE

7867
04:49:43,077 --> 04:49:44,612
PEOPLE WHO ARE EAGER TO LEARN

7868
04:49:44,612 --> 04:49:46,080
FROM 1 ANOTHER, TO ME THAT'S

7869
04:49:46,080 --> 04:49:49,150
WHERE THE MAGIC IS.

7870
04:49:49,150 --> 04:49:50,351
>> YEAH, I WOULD AGREE AND
IT'S

7871
04:49:50,351 --> 04:49:52,854
HARD TO DO, I FEEL LIKE IT'S A

7872
04:49:52,854 --> 04:49:53,888
PARTICULAR CHALLENGE FOR

7873
04:49:53,888 --> 04:49:55,790
ACADEMIA BECAUSE WE HAVE A BIT

7874
04:49:55,790 --> 04:49:57,191
MORE TURNOVER, ALSO, SO THE

7875
04:49:57,191 --> 04:49:58,259
TEAMS BECAUSE WE'RE TRAINING AND

7876
04:49:58,259 --> 04:50:00,028
PEOPLE LEAVE, BUT THEN AGAIN, I

7877
04:50:00,028 --> 04:50:01,095
RECOGNIZE THAT PEOPLE DO LEAVE

7878
04:50:01,095 --> 04:50:03,631
INDUSTRY AND SO FORTH, BUT,IOLE,

7879
04:50:03,631 --> 04:50:07,302
SORRY, I SHOULD LET YOU SAY

7880
04:50:07,302 --> 04:50:08,870
SOMETHING EMPLOY >>  I
WOULD

7881
04:50:08,870 --> 04:50:12,206
OBVIOUSLY AGREE WITH PAT AND

7882
04:50:12,206 --> 04:50:13,775
MARTI'S EXPERIENCES, IT STARTS

7883
04:50:13,775 --> 04:50:15,376
AT HIRING, LOOKING AT PEOPLE

7884
04:50:15,376 --> 04:50:18,479
SPECIFICALLY WITH THOSE KINDS OF

7885
04:50:18,479 --> 04:50:20,014
QUALITIES AS OPPOSE TO LETTING

7886
04:50:20,014 --> 04:50:21,015
SMK WHO DOESN'T HAVE THOSE

7887
04:50:21,015 --> 04:50:22,283
VALUES AND FIGURE OUT WHAT DO I

7888
04:50:22,283 --> 04:50:24,085
DO WITH THEM NOW AND BRING THE

7889
04:50:24,085 --> 04:50:26,421
WHOLE TEAM DOWN, SO WE TRY TO

7890
04:50:26,421 --> 04:50:27,755
SCREEN FOR THOSE LIABILITIES

7891
04:50:27,755 --> 04:50:30,491
UNSURPRISINGLY BUT BEING

7892
04:50:30,491 --> 04:50:32,393
INTENGAL ABOUT HIGH   PERFORMING

7893
04:50:32,393 --> 04:50:33,795
TEAM EXERCISES REALLY

7894
04:50:33,795 --> 04:50:36,097
UNDERSTANDING PEOPLE'S

7895
04:50:36,097 --> 04:50:36,964
COMMUNICATION STYLES.

7896
04:50:36,964 --> 04:50:37,732
ABSOLUTELY THE INVESTMENT IS

7897
04:50:37,732 --> 04:50:39,133
THERE AND SO AGAIN GETTING BACK

7898
04:50:39,133 --> 04:50:40,468
TO ADVISE FOR YOUNG PEOPLE INTO

7899
04:50:40,468 --> 04:50:41,736
THE FIELD, BEING ABLE TO

7900
04:50:41,736 --> 04:50:42,837
COMMUNICATE AND GET ALONG WITH

7901
04:50:42,837 --> 04:50:49,344
OTHERS IS JUST AS IMPORTANT AS

7902
04:50:49,344 --> 04:50:52,080
YOUR TECHNICAL KNOWLEDGE.

7903
04:50:52,080 --> 04:50:54,749
>> I WOULD ALSO AGREE WITH
JOEL

7904
04:50:54,749 --> 04:50:56,984
ON THAT 1 AS WELL, I WOULD SAY

7905
04:50:56,984 --> 04:51:07,662
THE REWART  RE-- REWARD STRUCTUE

7906
04:51:14,135 --> 04:51:16,971
AWARD STRUCTURES IN ACADEMIA ARE

7907
04:51:16,971 --> 04:51:27,448
DISINCENTIVE, I KNOW, I KNOW

7908
04:51:33,521 --> 04:51:36,591
SEVERAL. 

7909
04:51:36,591 --> 04:51:37,792
>> RUSS ANY COMMENTS?

7910
04:51:37,792 --> 04:51:38,426
>> IT WASN'T LIKE THE STAR

7911
04:51:38,426 --> 04:51:39,694
SYSTEM WHERE IT WAS ABOUT

7912
04:51:39,694 --> 04:51:41,295
INDIVIDUAL STARS WHO WERE -- AND

7913
04:51:41,295 --> 04:51:42,930
THEN A STAR IN THE MIDDLE AND

7914
04:51:42,930 --> 04:51:44,165
THEN THAT WAS ALL OUT THE DOOR

7915
04:51:44,165 --> 04:51:46,601
AND IT WAS ALL ABOUT THE MISSION

7916
04:51:46,601 --> 04:51:48,703
OF THE COMPANY, AND THAT STAGE

7917
04:51:48,703 --> 04:51:50,338
IT WAS JUST VERY INVIGORATING

7918
04:51:50,338 --> 04:51:51,906
AND I SAID TO MYSELF, FOR THE

7919
04:51:51,906 --> 04:51:53,374
FIRST TIME IN 35 YEARS BECAUSE I

7920
04:51:53,374 --> 04:51:59,113
REALLY, I NEVER -- I NEVER HAD A

7921
04:51:59,113 --> 04:52:02,016
REAL JOB, NOW I SEE WHY BEING IN

7922
04:52:02,016 --> 04:52:03,050
INDUSTRY CAN BE VERY SATISFYING

7923
04:52:03,050 --> 04:52:04,218
BECAUSE I WAS ON THEEZ

7924
04:52:04,218 --> 04:52:05,653
FUNCTIONAL TEAMS WHERE THERE

7925
04:52:05,653 --> 04:52:09,524
WASN'T TOO MUCH BS, WE WERE

7926
04:52:09,524 --> 04:52:10,324
LUCKY POLITICS HADN'T OCCURRED

7927
04:52:10,324 --> 04:52:11,626
IN THE COMPANY AND THEY WERE ALL

7928
04:52:11,626 --> 04:52:12,894
ROW NOTHING THE SAME DIRECTION,

7929
04:52:12,894 --> 04:52:14,629
I SAID THIS IS WHY IT'S COOL

7930
04:52:14,629 --> 04:52:16,230
THEY WILL GET STUFF DONE AND

7931
04:52:16,230 --> 04:52:17,565
THEY DID, THAT WE WOULD NEVER BE

7932
04:52:17,565 --> 04:52:21,669
ABLE TO DO IN ACADEMIA.

7933
04:52:21,669 --> 04:52:24,839
WELL THAT'S 1 THING MARTI CAN

7934
04:52:24,839 --> 04:52:26,340
TALK ABOUT FOR A LONG TIME AND

7935
04:52:26,340 --> 04:52:27,542
PUBLISHING AND EVERYTHING, AND I

7936
04:52:27,542 --> 04:52:28,209
WILL JUST MAKE A COMMENT BECAUSE

7937
04:52:28,209 --> 04:52:30,445
I WANT TO MAKE SURE WE STILL

7938
04:52:30,445 --> 04:52:30,812
HAVE TIME.

7939
04:52:30,812 --> 04:52:33,014
WE WILL DO A COUPLE MORE

7940
04:52:33,014 --> 04:52:34,248
QUESTIONS BUT DURING OSKS, ALSO

7941
04:52:34,248 --> 04:52:36,451
FOR ME, THAT WAS A TIME, TOO

7942
04:52:36,451 --> 04:52:40,588
WHERE WE HAD THIS SHEDDING WELL,

7943
04:52:40,588 --> 04:52:41,789
EVERYWHERE BUT ALSO IN ACADEMIA,

7944
04:52:41,789 --> 04:52:43,724
THE SHEDDING OF ALL OF OUR KIND

7945
04:52:43,724 --> 04:52:45,526
OF LIKE NORMAL ROUTINES AND SO

7946
04:52:45,526 --> 04:52:54,168
PEOPLE WERE SHARING AND TALKING

7947
04:52:54,168 --> 04:52:56,270
AND YOU DID LOSE IT QUITE A BIT,

7948
04:52:56,270 --> 04:53:06,814
I DON'T MISS, I DON'T MISS THAT,

7949
04:53:17,758 --> 04:53:19,293
THE THINGS THAT YOU'RE

7950
04:53:19,293 --> 04:53:20,528
DESCRIBING THERE ARE EXACTLY THE

7951
04:53:20,528 --> 04:53:22,063
THEMES THAT WE'RE DESCRIBING

7952
04:53:22,063 --> 04:53:23,130
FROM PAT ANDIOLE.

7953
04:53:23,130 --> 04:53:26,300
WHEN WE WALK INTO THIS COMPANY,

7954
04:53:26,300 --> 04:53:29,070
OUR GOAL IS TO DRIVE MEDICINES

7955
04:53:29,070 --> 04:53:31,706
TOWARDS PATIENTS SO THEY FEEL

7956
04:53:31,706 --> 04:53:32,640
BETTER.

7957
04:53:32,640 --> 04:53:36,511
WE'RE ALL ALIGNED, AND GOING

7958
04:53:36,511 --> 04:53:36,978
TOWARDS --

7959
04:53:36,978 --> 04:53:37,612
>> THAT'S REALLY FUN.

7960
04:53:37,612 --> 04:53:39,580
>> WE ALL KNOW THERE'S
THIS ARC

7961
04:53:39,580 --> 04:53:40,982
FORWARD THAT WE'RE ALL INVESTED

7962
04:53:40,982 --> 04:53:43,317
IN TOGETHER AND SO, YES, IN ANY

7963
04:53:43,317 --> 04:53:45,386
TIME YOU GET 2 PEOPLE TOGETHER,

7964
04:53:45,386 --> 04:53:47,355
THERE WILL BE SOME POLITICS,

7965
04:53:47,355 --> 04:53:50,725
THAT GOES WITHOUT SPEAKING BUT

7966
04:53:50,725 --> 04:53:51,993
THE FACT THAT IT IS SX HEAT

7967
04:53:51,993 --> 04:53:53,427
THAT'S WHY I CAME BACK TO

7968
04:53:53,427 --> 04:53:56,964
INDUSTRY, IS BECAUSE I DID THAT

7969
04:53:56,964 --> 04:54:02,537
WORK WITH THE D. O. E. COVID

7970
04:54:02,537 --> 04:54:03,804
DISCOVERY PROJECT.

7971
04:54:03,804 --> 04:54:04,405
>> VERY DIFFERENT.

7972
04:54:04,405 --> 04:54:06,874
>> NINE NATIONAL LABS AND
TO SEE

7973
04:54:06,874 --> 04:54:10,711
THEM MERGE INTO A TEAM AND STEP

7974
04:54:10,711 --> 04:54:13,748
INTO ACTION WAS JUST

7975
04:54:13,748 --> 04:54:14,715
INVIGORATING TO AND BROUGHT ME

7976
04:54:14,715 --> 04:54:24,158
BACK HOME WHERE I BELONG.

7977
04:54:24,158 --> 04:54:25,760
SMRKS OKAY, COOL, I LOST TRACK,

7978
04:54:25,760 --> 04:54:27,628
THIS IS WHY I HAVE TO PAY

7979
04:54:27,628 --> 04:54:27,895
ATTENTION.

7980
04:54:27,895 --> 04:54:29,730
OKAY, OKAY, IS IT TIME FOR

7981
04:54:29,730 --> 04:54:30,031
PREDICTIONS?

7982
04:54:30,031 --> 04:54:37,738
OR IS IT TOO EARLY IN NI DO WANT

7983
04:54:37,738 --> 04:54:39,040
YOU ALL TO MAKE A PREDICTION

7984
04:54:39,040 --> 04:54:40,942
ABOUT THE FUTURE, WHATEVER THEY

7985
04:54:40,942 --> 04:54:42,076
SAY ABOUT THAT, ABOUT, YOU KNOW

7986
04:54:42,076 --> 04:54:43,744
WHAT DO YOU THINK THE DRUG

7987
04:54:43,744 --> 04:54:44,845
DISCOVERY PROCESS IS GOING TO

7988
04:54:44,845 --> 04:54:46,480
LOOK LIKE 10 YEARS FROM NOW.

7989
04:54:46,480 --> 04:54:49,617
YOU THINK IT WILL BE RADICALLY

7990
04:54:49,617 --> 04:54:49,884
DIFFERENT.

7991
04:54:49,884 --> 04:54:51,352
IS EVERYTHING GOING TO BE

7992
04:54:51,352 --> 04:54:53,087
SELF-DRIVING LABS LIKE ROBOTS.

7993
04:54:53,087 --> 04:54:57,892
WILL PEOPLE KNOW HOW TO PIPETTE,

7994
04:54:57,892 --> 04:55:00,027
OR ARE WE STILL HOPING FORMORE

7995
04:55:00,027 --> 04:55:10,571
DATA MAYBE 10 YEARS SO THAT PUTS

7996
04:55:12,406 --> 04:55:13,774
AT 2034?

7997
04:55:13,774 --> 04:55:15,042
>> I'M NOT TALKING FIRST.

7998
04:55:15,042 --> 04:55:17,111
>> WHO WANTS TO BITE FIRST
LIKE

7999
04:55:17,111 --> 04:55:18,846
SOMETIMES YOU GET THE

8000
04:55:18,846 --> 04:55:19,513
IMPRESSION, YEAH, IT'S CHANGING

8001
04:55:19,513 --> 04:55:21,349
SO FAST BUT YEAH, ACTUALLY.

8002
04:55:21,349 --> 04:55:24,619
>> OKAY, I'LL MAKE A
PREDICTION.

8003
04:55:24,619 --> 04:55:27,588
I AM WORRIED THAT MY LOVE AND

8004
04:55:27,588 --> 04:55:29,523
EXPERTISE ABOUT SMALL MOLECULES

8005
04:55:29,523 --> 04:55:32,727
IS A FADING THING BECAUSE OF

8006
04:55:32,727 --> 04:55:35,596
CRSPR AND BECAUSE OF -- IT'S NOT

8007
04:55:35,596 --> 04:55:37,398
REALLY A WORRY BUT I AM SO

8008
04:55:37,398 --> 04:55:40,001
EXCITED, LET'S PUT IT AS A

8009
04:55:40,001 --> 04:55:41,736
POSITIVE THAT NOW WHEN YOU HAVE

8010
04:55:41,736 --> 04:55:43,537
A GOOD TARGET, YOU HAVE THIS

8011
04:55:43,537 --> 04:55:45,806
ARRAY OF WAYS TO INTERFERE WITH

8012
04:55:45,806 --> 04:55:48,643
IT, WITH SMALL MOLECULES OF

8013
04:55:48,643 --> 04:55:51,012
COURSE, WITH PROTEINS, WITH

8014
04:55:51,012 --> 04:55:53,781
GENOME ENGINEERING, WITH THESE

8015
04:55:53,781 --> 04:55:55,850
DEGRADERS, AND IT'S JUST

8016
04:55:55,850 --> 04:55:57,618
EXPLODING AND SO, THE ARSENAL, I

8017
04:55:57,618 --> 04:56:01,622
DON'T LIKE TO USE A WAR -- SORRY

8018
04:56:01,622 --> 04:56:06,460
I TO STOP MY ALARM.

8019
04:56:06,460 --> 04:56:07,161
ONE SECOND.

8020
04:56:07,161 --> 04:56:11,365
YES, BUT THIS ALARM -- I

8021
04:56:11,365 --> 04:56:13,968
APOLOGIZE, ALEXIA STOP.

8022
04:56:13,968 --> 04:56:17,571
SORRY ABOUT THAT, IT WAS TO

8023
04:56:17,571 --> 04:56:19,373
REMIND ME TO BE ON THIS PANEL,

8024
04:56:19,373 --> 04:56:22,610
OBVIOUSLY NOT THE RIGHT TIME.

8025
04:56:22,610 --> 04:56:24,278
SO THE AGERINAL OF TOOLS WILL

8026
04:56:24,278 --> 04:56:26,180
MAKE IT VERY EXCITING BUT ALSO

8027
04:56:26,180 --> 04:56:27,048
CONNIFIESING ABOUT WHAT'S THE

8028
04:56:27,048 --> 04:56:31,519
BEST WAY TO BUILD AN INTERVEKS

8029
04:56:31,519 --> 04:56:34,388
AND IT MIGHT CHANGE LIKE LOTS OF

8030
04:56:34,388 --> 04:56:35,456
THINGS, THE ECONOMICS AND THE

8031
04:56:35,456 --> 04:56:37,725
LEADERS IN THE FIELD, THE WAYS,

8032
04:56:37,725 --> 04:56:38,793
WHAT DISEASES ARE EASY TO TREAT

8033
04:56:38,793 --> 04:56:40,094
AND WHAT DISEASES ARE HARD TO

8034
04:56:40,094 --> 04:56:41,395
TREAT, SO I THINK THAT THERE'S

8035
04:56:41,395 --> 04:56:43,531
GOING -- I DON'T WANT TO CALL IT

8036
04:56:43,531 --> 04:56:45,566
CONFUSION BUT IT MIGHT LOOK LIKE

8037
04:56:45,566 --> 04:56:46,500
A VERY DIFFERENT WORLD TO THOSE

8038
04:56:46,500 --> 04:56:50,504
OF US IF WE WERE JUST

8039
04:56:50,504 --> 04:56:51,505
TRANSPORTED AHEAD, IT WILL LOOK

8040
04:56:51,505 --> 04:56:52,907
VERY DIFFERENT, TO ANSWER YOUR

8041
04:56:52,907 --> 04:56:54,308
QUESTION, IT WILL LOOK VERY

8042
04:56:54,308 --> 04:56:56,711
DIFFERENT AND IT WILL BE A MUCH

8043
04:56:56,711 --> 04:57:06,520
MORE MIXTURE OF APPROACHES

8044
04:57:06,520 --> 04:57:07,755
TOWARDS APPROACHES FORWARD.

8045
04:57:07,755 --> 04:57:08,289
>> THEOR PUTTIC.

8046
04:57:08,289 --> 04:57:10,491
>> SO I THINK IT'S ALWAYS
SAFE

8047
04:57:10,491 --> 04:57:14,261
TO MAKE A PREDICTION THAT THE

8048
04:57:14,261 --> 04:57:16,230
FUTURE WILL LOOK LIKE NOW ONLY

8049
04:57:16,230 --> 04:57:18,432
MORE SO, RIGHT IN AND SO, IT

8050
04:57:18,432 --> 04:57:22,603
WILL ALWAYS, WE WILL ALWAYS BE

8051
04:57:22,603 --> 04:57:24,004
REQUIRED TO FIND MEDICINES THAT

8052
04:57:24,004 --> 04:57:25,940
ARE EFFICACIOUS FOR THE THING,

8053
04:57:25,940 --> 04:57:28,509
THE DEC THAT WE NEED TO TREAT

8054
04:57:28,509 --> 04:57:30,444
THAT ARE SAFE, THAT HAVE ALL OF

8055
04:57:30,444 --> 04:57:32,046
THESE PROPERTIES TO ALLOW US TO

8056
04:57:32,046 --> 04:57:34,882
MAKE THEM, THAT'S NOT GOING TO

8057
04:57:34,882 --> 04:57:35,116
GO AWAY.

8058
04:57:35,116 --> 04:57:36,984
I AGREE WITH RUSS, THAT THERE'S

8059
04:57:36,984 --> 04:57:40,888
HUGE OPPORTUNITIES ON THE

8060
04:57:40,888 --> 04:57:44,325
HORIZON, I LOVE HAVING COME TO A

8061
04:57:44,325 --> 04:57:46,427
COMPANY THAT FOCUSES ON

8062
04:57:46,427 --> 04:57:49,563
BIOLOGICS AND SMALL MOLECULES,

8063
04:57:49,563 --> 04:57:52,166
THAT HAS BOTH OF THOSE TOOL

8064
04:57:52,166 --> 04:57:54,702
KITS, PLUS MANY COMBINATIONS

8065
04:57:54,702 --> 04:57:55,770
WITHIN IT INSIDE.

8066
04:57:55,770 --> 04:58:00,841
AND SO, THAT'S EXCITING AND

8067
04:58:00,841 --> 04:58:03,444
OPENS UP THE SET OF TOOLS THAT

8068
04:58:03,444 --> 04:58:08,082
WE HAVE TO PLAY WITH TO

8069
04:58:08,082 --> 04:58:08,883
INFLUENCE HUMAN BIOLOGY, BUT

8070
04:58:08,883 --> 04:58:10,651
APPROXIMATE WE DID, WE STILL

8071
04:58:10,651 --> 04:58:12,553
HAVE TO INFLUENCE HUMAN BIOLOGY

8072
04:58:12,553 --> 04:58:14,188
IN WAYS THAT ARE EFFECTIVE AND

8073
04:58:14,188 --> 04:58:17,691
WE HAVE TO PROVE IT.

8074
04:58:17,691 --> 04:58:21,362
I ALSO TRULY BELIEVE THAT THE

8075
04:58:21,362 --> 04:58:24,965
FUTURE WILL ALWAYS BELONG TO

8076
04:58:24,965 --> 04:58:27,768
THOSE PEOPLE WHO CAN -- THOSE

8077
04:58:27,768 --> 04:58:30,771
TEAMS, THOSE INDIVIDUALS, THOSE

8078
04:58:30,771 --> 04:58:34,608
COMPANIES THAT CAN HARNESS BOTH

8079
04:58:34,608 --> 04:58:37,311
THE HUMAN EXPERTISE AND SHE'S

8080
04:58:37,311 --> 04:58:41,515
NEW AUTOMATION COMPUTATIONAL

8081
04:58:41,515 --> 04:58:41,816
APPROACHES.

8082
04:58:41,816 --> 04:58:48,656
THE SIGNIFY  -- CYBORG IS GOING,

8083
04:58:48,656 --> 04:58:49,824
NOT THE HUMAN ALONE.

8084
04:58:49,824 --> 04:58:52,526
>> I THINK I HAVE A
SLIGHTLY

8085
04:58:52,526 --> 04:58:57,264
DIFFERENT SPIN, ON WHAT RUSS

8086
04:58:57,264 --> 04:58:57,731
SAID BECAUSE I AGREE.

8087
04:58:57,731 --> 04:58:59,834
I THINK WE WILL HAVE NEW MODEL

8088
04:58:59,834 --> 04:59:00,467
CITIZEN DALLASCOWBOYS.COM IITYS

8089
04:59:00,467 --> 04:59:03,404
BUT IF YOU LOOK AT THESE NEW

8090
04:59:03,404 --> 04:59:04,371
EMERGING MODALITIES, THERE'S

8091
04:59:04,371 --> 04:59:06,006
STILL A SMALL MOLECULE COMPONENT

8092
04:59:06,006 --> 04:59:08,676
WHETHER YOU'RE LOOKING AT RADIO

8093
04:59:08,676 --> 04:59:10,511
THERAPY OR ANTIBODY DRUG

8094
04:59:10,511 --> 04:59:12,780
CONJUGATES OR TARGETED PROTEIN

8095
04:59:12,780 --> 04:59:14,048
DEGRADATION, YOU STILL HAVE TO

8096
04:59:14,048 --> 04:59:16,116
BE ABLE TO DESIGN A SMALL

8097
04:59:16,116 --> 04:59:21,455
MOLECULE AND MY HOPE AT LEAST

8098
04:59:21,455 --> 04:59:22,356
FOR THE NEXT 10 YEARS FROM NOW

8099
04:59:22,356 --> 04:59:24,625
IS THAT WE HAVE A BETTER

8100
04:59:24,625 --> 04:59:26,126
UNDERSTANDING OF THE THINGS THAT

8101
04:59:26,126 --> 04:59:27,361
ARE STUMPING US RIGHT NOW

8102
04:59:27,361 --> 04:59:29,630
BECAUSE I THINK A LOT OF WHAT'S

8103
04:59:29,630 --> 04:59:32,766
STUMPS US NOW IS WE DON'T HAVE

8104
04:59:32,766 --> 04:59:34,535
ENOUGH OF AN UNDERSTANDING OF

8105
04:59:34,535 --> 04:59:37,304
ALL THE OFFTARGETS THAT WE'RE

8106
04:59:37,304 --> 04:59:37,872
TRYING TO AVOID.

8107
04:59:37,872 --> 04:59:41,442
WE DON'T HAVE ENOUGH OF AN

8108
04:59:41,442 --> 04:59:43,644
UNDERSTANDING OF DRUG

8109
04:59:43,644 --> 04:59:45,279
METABOLIZING ENZYMES AND HOW TO

8110
04:59:45,279 --> 04:59:50,384
DESIGN MOLECULES TO AVOID THOSE

8111
04:59:50,384 --> 04:59:52,052
LIABILITIES AND THROUGH A

8112
04:59:52,052 --> 04:59:53,821
COMBINATION OF DATA GENERATION

8113
04:59:53,821 --> 04:59:55,856
AND ADVANCES IN STRUCTURAL

8114
04:59:55,856 --> 04:59:58,092
BIOLOGY, WE CAN GIVE A LOT MORE

8115
04:59:58,092 --> 05:00:00,194
OF AN UNDERSTANDING AND THEN

8116
05:00:00,194 --> 05:00:02,496
MAYBE, THE THINGS BEYOND POTENCY

8117
05:00:02,496 --> 05:00:04,899
WHICH ARE REALLY, THE TIME

8118
05:00:04,899 --> 05:00:07,067
CONSUMING PARTS OF DRUG

8119
05:00:07,067 --> 05:00:13,474
DISCOVERY NOW WILL BECOME MORE

8120
05:00:13,474 --> 05:00:18,312
STREAMLINED. 

8121
05:00:18,312 --> 05:00:18,579
>> YEAH.

8122
05:00:18,579 --> 05:00:19,046
OKAY.

8123
05:00:19,046 --> 05:00:19,513
JOEL?

8124
05:00:19,513 --> 05:00:21,448
>> I THINK BIOLOGY WILL
ALWAYS

8125
05:00:21,448 --> 05:00:22,583
SURPRISE US.

8126
05:00:22,583 --> 05:00:24,919
THERE WILL BE SOMETHING

8127
05:00:24,919 --> 05:00:25,686
FUNDAMENTAL ANOTHER RNA SPECIES

8128
05:00:25,686 --> 05:00:27,121
OR SOMETHING, WHO KNOWS THAT WE

8129
05:00:27,121 --> 05:00:28,422
WON'T BE ABLE TO TAKE INTO,A

8130
05:00:28,422 --> 05:00:30,190
COUNT BUT BECAUSE OF THAT AND IT

8131
05:00:30,190 --> 05:00:32,026
WILL PROBABLY TAKE A WHILE JUST

8132
05:00:32,026 --> 05:00:35,029
A MUCH BIGGER INVESTMENT IN

8133
05:00:35,029 --> 05:00:35,663
STANDARDIZING CLINICAL DATA,

8134
05:00:35,663 --> 05:00:38,799
REALLY MAKING THE USE OF

8135
05:00:38,799 --> 05:00:40,834
CLINICAL TRIALS, PATIENT SAMPLES

8136
05:00:40,834 --> 05:00:41,635
MAKE AN IMPACT ON THE DATA SETS

8137
05:00:41,635 --> 05:00:43,137
WE HAVE TO BETTER UNDERSTAND

8138
05:00:43,137 --> 05:00:50,444
DECS AND IN BIGGER FOCUS THAN

8139
05:00:50,444 --> 05:00:52,146
PREVENTION OF DISEASES, AS

8140
05:00:52,146 --> 05:00:53,948
OPPOSE TO ACTUALLY TREATING

8141
05:00:53,948 --> 05:00:59,553
THEM, SO JUST GENERAL BASIC

8142
05:00:59,553 --> 05:01:00,387
UNDERSTANDING.

8143
05:01:00,387 --> 05:01:04,325
>> YEAH, THOSE WERE ALL
TERRIFIC

8144
05:01:04,325 --> 05:01:04,591
THOUGHTS.

8145
05:01:04,591 --> 05:01:06,560
I WAS TOLD -- I THINK, IT SEEMS

8146
05:01:06,560 --> 05:01:08,228
I HOPE IT'S CLEAR THAT 10 YEARS

8147
05:01:08,228 --> 05:01:10,230
FROM NOW OUR DATA LANDSCAPE

8148
05:01:10,230 --> 05:01:14,268
HOPEFULLY IN BIOLOGY WILL BE

8149
05:01:14,268 --> 05:01:16,070
SIGNIFICANTLY IMPROVED FROM

8150
05:01:16,070 --> 05:01:17,938
WHERE WE ARE I KNOW THERE'S LOTS

8151
05:01:17,938 --> 05:01:19,506
OF FOLKS WORKING ON THAT, IF

8152
05:01:19,506 --> 05:01:21,008
PEOPLE ARE SORT OF TAKING HOLD

8153
05:01:21,008 --> 05:01:23,577
AND THINKING ABOUT HOW WE CAN DO

8154
05:01:23,577 --> 05:01:24,845
THAT EFFECTIVELY, SO AT LEAST I

8155
05:01:24,845 --> 05:01:33,554
THINK WE HOPE TO HAVE THAT TO

8156
05:01:33,554 --> 05:01:35,556
LOOK FORWARD WELL, GOODNESS, WE

8157
05:01:35,556 --> 05:01:37,291
COULD DO 1 MORE, WE COULD WRAP

8158
05:01:37,291 --> 05:01:37,524
IT UP.

8159
05:01:37,524 --> 05:01:39,293
WE'VE BEEN GOING FOR A WHILE.

8160
05:01:39,293 --> 05:01:41,895
ARE THERE ANY FINAL THOUGHTS ON

8161
05:01:41,895 --> 05:01:44,298
THE PANEL ANYONE WANTS TO SHARE.

8162
05:01:44,298 --> 05:01:45,733
IS THERE SOME BURNING QUESTION

8163
05:01:45,733 --> 05:01:47,968
YOU HOPED THAT I WOULD ASK THAT

8164
05:01:47,968 --> 05:01:54,174
I DIDN'T?

8165
05:01:54,174 --> 05:01:55,042
CLOSING THOUGHTS?

8166
05:01:55,042 --> 05:01:58,078
WE'RE QUIET, WE TALKED A LOT.

8167
05:01:58,078 --> 05:01:58,912
SHAYAM, ANYTHING FROM YOU,

8168
05:01:58,912 --> 05:02:01,849
ANYTHING YOU WANT TO ASK THESE

8169
05:02:01,849 --> 05:02:03,617
FOLKS WHILE WE HAVE THESE

8170
05:02:03,617 --> 05:02:04,618
MARVELOUS PEOPLE ON THE HOOK.

8171
05:02:04,618 --> 05:02:06,754
>> I MEAN WE COVERED A LOT
OF,

8172
05:02:06,754 --> 05:02:08,722
YOU KNOW I LOT OF DIFFERENT

8173
05:02:08,722 --> 05:02:11,225
TOPICS, ALL I CAN SAY IS THIS

8174
05:02:11,225 --> 05:02:12,459
WAS EXTREMELY ENLIGHTENING AND I

8175
05:02:12,459 --> 05:02:15,295
CAN HEAR YOU ALL TALK FOR HOURS.

8176
05:02:15,295 --> 05:02:17,398
BUT I'M SURE WE'RE ALL LIKE,

8177
05:02:17,398 --> 05:02:18,899
GETTING TIRED AND YOU KNOW

8178
05:02:18,899 --> 05:02:20,134
PROBABLY WE SHOULD MOVE FORWARD,

8179
05:02:20,134 --> 05:02:23,637
SO THANK YOU SO MUCH FOR

8180
05:02:23,637 --> 05:02:27,741
EVERYONE, THAT WAS SAB LIEWT --

8181
05:02:27,741 --> 05:02:31,078
ABSOLUTELY AMAZING AND THANK YOU

8182
05:02:31,078 --> 05:02:33,213
ROMMIE FOR MODERATING THIS.

8183
05:02:33,213 --> 05:02:34,948
NTHANK YOU FOR PARTICIPATING.

8184
05:02:34,948 --> 05:02:35,849
I HAD A GOOD TIME.

8185
05:02:35,849 --> 05:02:37,918
>> THANK YOU FOR THE
PANELISTS

8186
05:02:37,918 --> 05:02:39,486
WHAT I WILL DO IS I WILL SHARE

8187
05:02:39,486 --> 05:02:44,024
MY SLIDE HERE XI WILL CALL FOR

8188
05:02:44,024 --> 05:02:45,926
THE ORGANIZERS TO TURN THEIR

8189
05:02:45,926 --> 05:02:47,661
CAMERAS AND MICs ON AND JUST

8190
05:02:47,661 --> 05:02:49,430
ASK EVERYONE TO COME UP WITH A

8191
05:02:49,430 --> 05:02:50,764
QUICK SUMMARY OF KEY TAKE AWAYS

8192
05:02:50,764 --> 05:02:54,234
AND REFLECT ON THE NEXT, THE

8193
05:02:54,234 --> 05:03:04,778
PAST 2 DAYS BEFORE WE WRAP THIS

8194
05:03:12,920 --> 05:03:13,087
UP.

8195
05:03:13,087 --> 05:03:16,790
>> SO WE HAVE, OKAY, THE
SLIDE

8196
05:03:16,790 --> 05:03:17,491
IS UP.

8197
05:03:17,491 --> 05:03:20,894
SO MAYBE ALEX YOU CAN START WITH

8198
05:03:20,894 --> 05:03:23,697
YOUR THOUGHTS ABOUT THESE 2 DAYS

8199
05:03:23,697 --> 05:03:25,599
AND WHAT YOU THINK WERE THE BEST

8200
05:03:25,599 --> 05:03:30,604
KEY TAKE AWAYS HERE.

8201
05:03:30,604 --> 05:03:30,971
>> THANK YOU.

8202
05:03:30,971 --> 05:03:41,448
SO I THINK REALLY HAD SUPER

8203
05:03:45,552 --> 05:03:45,919
EXCITING EVENT.

8204
05:03:45,919 --> 05:03:47,788
WE TARTED WITH A LOT OF --

8205
05:03:47,788 --> 05:03:50,023
STARTED WITH A LOT OF

8206
05:03:50,023 --> 05:03:51,625
ENTHUSIASM, BUT I WOULD SAY

8207
05:03:51,625 --> 05:03:53,393
ENTHUSIASMOT PART OF SPEAKERS

8208
05:03:53,393 --> 05:03:57,998
AND LISTENERS, THE WORKSHOP AS

8209
05:03:57,998 --> 05:03:58,599
BEEN OVERSUBSCRIBED SO THERE

8210
05:03:58,599 --> 05:04:01,301
WERE A LOT OF TREMENDOUSLY

8211
05:04:01,301 --> 05:04:04,204
INTERESTED PEOPLE IN THE

8212
05:04:04,204 --> 05:04:04,805
ACCOMPLISHMENTS.

8213
05:04:04,805 --> 05:04:06,206
IN NO SMALLER NUMBER OF IDEAS

8214
05:04:06,206 --> 05:04:09,243
AND THEIR ACTIONS FOR FUTURE

8215
05:04:09,243 --> 05:04:16,650
RESEARCH, SO I THINK IF ANYTHING

8216
05:04:16,650 --> 05:04:23,657
IN THIS WORKSHOP THAT IF PEOPLE

8217
05:04:23,657 --> 05:04:25,159
THINK ABOUT WHAT TO DO AND WHAT

8218
05:04:25,159 --> 05:04:27,127
TO CHOOSE, I THINK IT HAS BEEN

8219
05:04:27,127 --> 05:04:29,263
PROFESSIONAL AND STIMULATING

8220
05:04:29,263 --> 05:04:30,931
EMPLOY SO, 1 BIG TAKE A WAY THAT

8221
05:04:30,931 --> 05:04:35,169
I WOULD LIKE TO SHARE.

8222
05:04:35,169 --> 05:04:36,470
>> I WANT TO --
DISCUSSION,

8223
05:04:36,470 --> 05:04:46,413
YEAH, I THINK THIS IS IT'S A

8224
05:04:46,413 --> 05:04:47,781
GOOD AMOUNT OF THE PRESENTATION

8225
05:04:47,781 --> 05:04:49,550
THAT WE LOOK AT AND I THINK IT'S

8226
05:04:49,550 --> 05:04:51,752
GREAT DISCUSSION ABOUT THE

8227
05:04:51,752 --> 05:04:53,487
INTEGRATION OF COMMUTATIONAL

8228
05:04:53,487 --> 05:04:54,254
MEDICINE AND EXPERIMENTAL

8229
05:04:54,254 --> 05:04:56,290
VALDATION OF THOSE EMPLOY I

8230
05:04:56,290 --> 05:04:57,291
THINK THIS IS REALLY IMPORTANT

8231
05:04:57,291 --> 05:05:01,461
PART AND IT WAS A LOT OF TALKS,

8232
05:05:01,461 --> 05:05:02,496
ACTUALLY SHOWN THAT THE MACHINE

8233
05:05:02,496 --> 05:05:04,498
LEARN XG IT WAS A PART OF THE

8234
05:05:04,498 --> 05:05:05,098
GOAL, YOU REMEMBER 1 YESTERDAY

8235
05:05:05,098 --> 05:05:15,609
AND IN THE MORNING WE TRIED TO

8236
05:05:17,744 --> 05:05:19,813
OPEN IT'S ICSIGHTING TO SEE

8237
05:05:19,813 --> 05:05:21,215
BIOTECH INVOLVED IN THE

8238
05:05:21,215 --> 05:05:22,983
INFORMATION AND AS WELL AS

8239
05:05:22,983 --> 05:05:24,318
ACADEMIA SHOWING THE NEW

8240
05:05:24,318 --> 05:05:29,056
APPROACH AND SO NEW TRENDS AND

8241
05:05:29,056 --> 05:05:31,758
MULTIOMICS DATA FROM RECURSION,

8242
05:05:31,758 --> 05:05:33,460
GREAT TALK, ALL THE DIFFERENT

8243
05:05:33,460 --> 05:05:35,929
SLIDES IN WHICH THEY INCORPORATE

8244
05:05:35,929 --> 05:05:36,897
AND CAPITULATIZE TO BUILD

8245
05:05:36,897 --> 05:05:38,031
MISSION LEARNING, I THINK THIS

8246
05:05:38,031 --> 05:05:48,342
IS REALLY EXCITING TO SEE AND OF

8247
05:05:48,342 --> 05:05:49,743
COURSE BINDING IMPRESSIONS AS

8248
05:05:49,743 --> 05:05:51,044
WELL, DAVID SHOWED US THAT WORK

8249
05:05:51,044 --> 05:05:53,814
AS WELL, GREAT PANEL DISCUSSION.

8250
05:05:53,814 --> 05:05:57,517
IT WAS REALLY GREAT

8251
05:05:57,517 --> 05:05:57,851
CONVERSATION.

8252
05:05:57,851 --> 05:06:08,362
FACILITATE TODAY, YEAH, IT WAS

8253
05:06:13,500 --> 05:06:14,034
AWESOME.

8254
05:06:14,034 --> 05:06:15,535
>> THANKS, SHAYAM.

8255
05:06:15,535 --> 05:06:17,771
>> I AGREE WITH EVERYTHING
SAID

8256
05:06:17,771 --> 05:06:18,739
EARLIER, GREAT PRESENTATIONS,

8257
05:06:18,739 --> 05:06:21,675
GREAT WORKSHOP, I LEARNED A LOT

8258
05:06:21,675 --> 05:06:23,543
GIVEN MY OLD SCHOOL TRAINING OF

8259
05:06:23,543 --> 05:06:26,079
LAB CHEMISTRY SO I HAVE GENERIC

8260
05:06:26,079 --> 05:06:27,281
POINTS I THOUGHT WERE VERY

8261
05:06:27,281 --> 05:06:28,649
IMPORTANT IN THIS PARTICULAR

8262
05:06:28,649 --> 05:06:32,486
WORKSHOP, 1 OF THE POINTS AGAIN,

8263
05:06:32,486 --> 05:06:35,922
IS THE TALK ABOUT BIOLOGICS AND

8264
05:06:35,922 --> 05:06:36,923
ANTIGENS WAS REALLY FASCINATING

8265
05:06:36,923 --> 05:06:39,593
AND I THINK NOW, TAKING THAT

8266
05:06:39,593 --> 05:06:40,494
VERTICAL MODEL AND EXPRESSING

8267
05:06:40,494 --> 05:06:46,366
THEM IN THE RIGHT CELL LINES AND

8268
05:06:46,366 --> 05:06:47,634
EXPRESSING THEM, THOSE ARE THE

8269
05:06:47,634 --> 05:06:52,873
NEXT QUESTIONS, THE HAPPY MEDIUM

8270
05:06:52,873 --> 05:06:54,207
OF OF EXPRESSING THESE MOLECULES

8271
05:06:54,207 --> 05:06:55,642
WOULD BE A--BITS IMPORTANT

8272
05:06:55,642 --> 05:06:58,178
ASPECT, I THINK, I THOUGHT

8273
05:06:58,178 --> 05:06:58,712
REPURPOSESSING AND

8274
05:06:58,712 --> 05:07:00,914
IDENTIFICATION OF THESE TARGETS

8275
05:07:00,914 --> 05:07:02,115
AND REPOSITIONING OF THE DRUGS

8276
05:07:02,115 --> 05:07:03,684
FINDING THAT INFORMATION USING

8277
05:07:03,684 --> 05:07:07,387
AI MACHINE LEARNING TOOLS WAS

8278
05:07:07,387 --> 05:07:09,956
PRETTY FASCINATING. 

8279
05:07:09,956 --> 05:07:12,059
AGAIN THAT IS SOMETHING THAT I

8280
05:07:12,059 --> 05:07:14,494
ASSUME WOULD BE USED MORE

8281
05:07:14,494 --> 05:07:16,129
BROADLY BECAUSE OF THE BROAD

8282
05:07:16,129 --> 05:07:20,834
INFORMATION THAT'S ALREADY THERE

8283
05:07:20,834 --> 05:07:21,968
AND REPOSITIONING OF THE OLDER

8284
05:07:21,968 --> 05:07:24,838
DRUGS WILL BE A FASCINATING

8285
05:07:24,838 --> 05:07:25,038
AREA.

8286
05:07:25,038 --> 05:07:27,074
AS A DEVELOPER MYSELF, I HAVE

8287
05:07:27,074 --> 05:07:29,476
ALWAYS LOOKED AT PROCESSES, THE

8288
05:07:29,476 --> 05:07:31,044
PRODUCT KIND OF A THING AND

8289
05:07:31,044 --> 05:07:34,448
THEREFORE BRINGING THIS MACHINE

8290
05:07:34,448 --> 05:07:36,850
LEARNING AND AI MODELS, HOW DO

8291
05:07:36,850 --> 05:07:39,186
YOU DEVELOP A PRODUCT BECOMES AN

8292
05:07:39,186 --> 05:07:40,921
IMPORTANT ASSPEC, CAN YOU SCALE

8293
05:07:40,921 --> 05:07:43,056
UP IT GOES THROUGH THE MODELING

8294
05:07:43,056 --> 05:07:44,858
FOR THE ASPECT LAB, AND THE

8295
05:07:44,858 --> 05:07:45,959
MANUFACTURING SITE, I THINK

8296
05:07:45,959 --> 05:07:47,494
THOSE ARE IMPORTANT AREAS TO

8297
05:07:47,494 --> 05:07:50,030
LOOK INTO HOW MUCH IS THAT

8298
05:07:50,030 --> 05:07:53,367
TRANSLATION ARE THESE MOLECULES

8299
05:07:53,367 --> 05:07:55,035
ABLE TO SCALE, AS IS BEING SAID,

8300
05:07:55,035 --> 05:07:58,472
OR ARE THERE ISSUES IN SCALING

8301
05:07:58,472 --> 05:08:02,876
UP AND MANUFACTURABILITY OF THIS

8302
05:08:02,876 --> 05:08:03,677
MOLECULES, YOU KNOW?

8303
05:08:03,677 --> 05:08:04,945
THAT IS SOMETHING TO WATCH OUT

8304
05:08:04,945 --> 05:08:07,247
FOR AND I'M SURE PEOPLE ARE

8305
05:08:07,247 --> 05:08:09,683
LOOKING AT THIS AND BRIDGING

8306
05:08:09,683 --> 05:08:10,350
THAT GAP.

8307
05:08:10,350 --> 05:08:18,058
THE OTHER POINT IS, YOU KNOW, IT

8308
05:08:18,058 --> 05:08:19,726
SEEMS LIKE THE LEARNING ASPECT

8309
05:08:19,726 --> 05:08:23,230
OF IT, OR DEEP LEARNING ASPECT,

8310
05:08:23,230 --> 05:08:25,966
IS REALLY OUTPACING THE CLINICAL

8311
05:08:25,966 --> 05:08:29,302
DEVELOPMENT AND I WORRY THERE

8312
05:08:29,302 --> 05:08:34,374
MIGHT BE A REGULATORY LEARNING

8313
05:08:34,374 --> 05:08:35,542
THAT NEEDS TO HAPPEN, YOU KNOW

8314
05:08:35,542 --> 05:08:37,010
WHICH I THINK IS PROBABLY --

8315
05:08:37,010 --> 05:08:42,682
THERE'S A LOT OF LEARNING THAT'S

8316
05:08:42,682 --> 05:08:44,818
HAPPENING UP FRONT BUT MORE THAT

8317
05:08:44,818 --> 05:08:46,086
GAP WILL INCREASE WITH THE

8318
05:08:46,086 --> 05:08:47,421
MODELS COMING UP AND TRYING TO

8319
05:08:47,421 --> 05:08:48,822
BRIDGE THAT GAP OR NARROW THAT

8320
05:08:48,822 --> 05:08:51,324
GAP WOULD BE AN IMPORTANT

8321
05:08:51,324 --> 05:08:56,830
ASPECT, I AND THEN PREDICTION OF

8322
05:08:56,830 --> 05:08:59,833
TOXICITY, IMPORTANT, PKPD, AND

8323
05:08:59,833 --> 05:09:02,202
OTHER DRUG TARGETS WE TALKED OFF

8324
05:09:02,202 --> 05:09:05,372
TARGET EEIVETS AND UNDERSTANDING

8325
05:09:05,372 --> 05:09:07,340
THE CONTRIBUTES OF THESE VARIOUS

8326
05:09:07,340 --> 05:09:09,743
MOLECULES THAT ARE BEING MODELED

8327
05:09:09,743 --> 05:09:15,248
WOULD BE AN IMPORTANT FACET I

8328
05:09:15,248 --> 05:09:15,449
THINK.

8329
05:09:15,449 --> 05:09:16,817
THOSE WOULD BE IMPORTANT TO

8330
05:09:16,817 --> 05:09:19,152
MONITOR SO I LOOK AT COMPLETELY

8331
05:09:19,152 --> 05:09:21,054
FROM A PRODUCT DEVELOPMENT LAB

8332
05:09:21,054 --> 05:09:22,556
CHEMIST RADIOY OR A BIOLOGICAL

8333
05:09:22,556 --> 05:09:24,357
POINT OF VIEW AND I THINK

8334
05:09:24,357 --> 05:09:26,760
BRIDGING THE GAP WILL BE AN

8335
05:09:26,760 --> 05:09:28,094
IMPORTANT ASPECT, 1 OTHER POINT

8336
05:09:28,094 --> 05:09:29,262
I QUICKLY WANT TO MAKE IS THAT

8337
05:09:29,262 --> 05:09:31,698
THIS IS NOT GOING TO JUST BE AN

8338
05:09:31,698 --> 05:09:32,933
ISSUE FOR VACCINES AND

8339
05:09:32,933 --> 05:09:35,135
THERAPEUTICS BUT EVEN FROM A

8340
05:09:35,135 --> 05:09:37,170
DIAGNOSTIC'S POINT OF VIEW, A

8341
05:09:37,170 --> 05:09:39,272
LOT OF VARIABLES AND OTHER

8342
05:09:39,272 --> 05:09:41,842
DEVICES DO USE ALGORITHMS AND

8343
05:09:41,842 --> 05:09:44,177
LEARNING TOOLS AND REGULATORY

8344
05:09:44,177 --> 05:09:46,480
PATH FORWARD FOR SOME OF THESE

8345
05:09:46,480 --> 05:09:48,748
NEW TECHNOLOGIES WILL BE REALLY

8346
05:09:48,748 --> 05:09:56,623
CRITICAL, I THINK TO OBSERVE.

8347
05:09:56,623 --> 05:09:59,192
THOSE WERE MY ONLY HIGH LEVEL

8348
05:09:59,192 --> 05:10:01,194
POINTS BUT REALLY FASCINATING

8349
05:10:01,194 --> 05:10:01,595
MEETING.

8350
05:10:01,595 --> 05:10:05,265
I TOOK A LOT OUT OF THIS MEETING

8351
05:10:05,265 --> 05:10:08,401
AND THANK YOU TO YOU AND NCATS

8352
05:10:08,401 --> 05:10:09,336
FOR ORGANIZING THIS.

8353
05:10:09,336 --> 05:10:10,704
IT THIS WAS FANTASTIC.

8354
05:10:10,704 --> 05:10:12,272
>> THANK YOU SO MUCH, THAT
WAS

8355
05:10:12,272 --> 05:10:14,774
AWESOME TO HEAR ALL OF YOUR

8356
05:10:14,774 --> 05:10:15,809
LEARNINGS AND YOU'RE BEING PART

8357
05:10:15,809 --> 05:10:17,010
OF THIS.

8358
05:10:17,010 --> 05:10:18,678
ROMMIE, I WILL GIVE YOU SOME

8359
05:10:18,678 --> 05:10:21,314
TIME TO ALSO SAY A WORD OR 2 AND

8360
05:10:21,314 --> 05:10:27,854
THEN WE CAN WRAP UP.

8361
05:10:27,854 --> 05:10:31,424
SURE, I MEAN JUST I SORT OF

8362
05:10:31,424 --> 05:10:32,726
ENJOYED JUST SEEING THE BREDTH

8363
05:10:32,726 --> 05:10:35,128
OF TOPICS IN THIS SPACE, JUST TO

8364
05:10:35,128 --> 05:10:37,597
ME IT'S ASTONISHING HOW MUCH IS

8365
05:10:37,597 --> 05:10:40,500
GOING ON, AND YOU KNOW JUST THE

8366
05:10:40,500 --> 05:10:41,801
WHOLE PIPELINE AT THE MOMENT.

8367
05:10:41,801 --> 05:10:44,638
SO IT WAS REALLY, IT WAS REALLY,

8368
05:10:44,638 --> 05:10:47,807
I THINK BENEFICIAL TO HEAR FROM

8369
05:10:47,807 --> 05:10:49,910
FOLKS, I'M REALLY EXCITED ABOUT

8370
05:10:49,910 --> 05:10:51,778
WHAT WE HEARD, ESPECIALLY THE

8371
05:10:51,778 --> 05:10:52,612
OPEN SCIENCE INITIATIVES, I

8372
05:10:52,612 --> 05:10:54,080
THINK THAT'S HUGE, YOU KNOW WE

8373
05:10:54,080 --> 05:10:58,418
HAVE TO REALLY GET SUCH A BETTER

8374
05:10:58,418 --> 05:11:00,020
GRIP ON THE DATA AND ESPECIALLY,

8375
05:11:00,020 --> 05:11:01,421
THERE'S A LOT OF MOVEMENT AND A

8376
05:11:01,421 --> 05:11:05,058
LOT OF MOVEMENT AND THINGS

8377
05:11:05,058 --> 05:11:06,126
HAPPENING THERE, SO I'M

8378
05:11:06,126 --> 05:11:06,960
PARTICULARLY, YOU KNOW I'M HAPPY

8379
05:11:06,960 --> 05:11:08,361
TO HEAR IT FROM ALL THESE

8380
05:11:08,361 --> 05:11:10,530
DIFFERENT ANGLES AND I HAD A LOT

8381
05:11:10,530 --> 05:11:11,364
OF FUN.

8382
05:11:11,364 --> 05:11:13,333
SO THANKS FOR ALSO ORGANIZING

8383
05:11:13,333 --> 05:11:15,702
THIS AND FOR THE PARTICIPANTS

8384
05:11:15,702 --> 05:11:17,637
AND SPEAKERS.

8385
05:11:17,637 --> 05:11:19,773
>> AWESOME THANK YOU
ROMMIE.

8386
05:11:19,773 --> 05:11:21,975
I WILL ECHO THAT, THIS WAS

8387
05:11:21,975 --> 05:11:22,275
EXCELLENT.

8388
05:11:22,275 --> 05:11:26,379
TWO DAYS OF LEARNING FOR ME AS

8389
05:11:26,379 --> 05:11:28,148
WELL, SO WONDERFUL TALKS.

8390
05:11:28,148 --> 05:11:29,716
THANKS FOR ALL OF OUR SPEAKERS

8391
05:11:29,716 --> 05:11:31,985
AND ALSO THANK YOU ALL, IT WAS

8392
05:11:31,985 --> 05:11:33,653
WONDERFUL WORKING WITH YOU AND

8393
05:11:33,653 --> 05:11:35,155
OUR ATTENDEES AS WELL, SO I WILL

8394
05:11:35,155 --> 05:11:37,490
WRAP UP WITH 1 MORE SLIDE HERE.

8395
05:11:37,490 --> 05:11:39,893
SO IF YOU'RE STILL HERE THANK

8396
05:11:39,893 --> 05:11:41,661
YOU FOR HANGING UNTIL THE LAST

8397
05:11:41,661 --> 05:11:45,031
STRETCH, WE APPRECIATE YOU SO

8398
05:11:45,031 --> 05:11:45,231
MUCH.

8399
05:11:45,231 --> 05:11:48,768
WE WOULD LOVE TO HEAR YOUR

8400
05:11:48,768 --> 05:11:50,670
FEEDBACK, OKAY?

8401
05:11:50,670 --> 05:11:51,705
DID YOU ENJOY THIS EVENT.

8402
05:11:51,705 --> 05:11:53,239
HOW COULD WE HAVE DONE THIS

8403
05:11:53,239 --> 05:11:53,607
BETTER FOR YOU?

8404
05:11:53,607 --> 05:11:55,742
SO THIS IS A QR CODE HERE, I

8405
05:11:55,742 --> 05:11:58,111
THINK THE BURDEN ON THIS IS 2 OR

8406
05:11:58,111 --> 05:11:58,979
3 MINUTES.

8407
05:11:58,979 --> 05:12:00,313
SHOULDN'T BE TOO LONG SO IF YOU

8408
05:12:00,313 --> 05:12:02,415
COULD JUST TAKE THIS FEEDBACK

8409
05:12:02,415 --> 05:12:05,752
SURVEY, AND LET US KNOW THAT

8410
05:12:05,752 --> 05:12:08,154
WILL BE FANTASTIC WE WILL ALSO

8411
05:12:08,154 --> 05:12:10,123
E-MAIL YOU THE LINKS HERE, SO IF

8412
05:12:10,123 --> 05:12:12,292
YOU CAN'T CATCH IT RIGHT NOW,

8413
05:12:12,292 --> 05:12:15,462
YOU COULD DO IT LATER.

8414
05:12:15,462 --> 05:12:17,197
THIS WORKSHOP IS BEING RECORDED,

8415
05:12:17,197 --> 05:12:20,266
SO YOU WILL ALL RECEIVE THE

8416
05:12:20,266 --> 05:12:24,037
LINKS TO VIEW THIS, ON DEMAND

8417
05:12:24,037 --> 05:12:28,908
WHENEVER YOU WOULD LIKE AND 1

8418
05:12:28,908 --> 05:12:31,645
MORE FINAL WORD, I LEAD THE

8419
05:12:31,645 --> 05:12:33,346
[INDISCERNIBLE] PROGRAM, THIS

8420
05:12:33,346 --> 05:12:33,947
PROGRAM HELPED ORGANIZE THIS

8421
05:12:33,947 --> 05:12:35,348
EVENT, THIS IS MULTIPLE WAY YOU

8422
05:12:35,348 --> 05:12:37,984
CAN CONNECT WITH US, WE DO

8423
05:12:37,984 --> 05:12:39,953
MULTIPLE OTHER WORKSHOPS, AS

8424
05:12:39,953 --> 05:12:44,157
WELL ON DIFFERENT TOPICS ON

8425
05:12:44,157 --> 05:12:45,158
DISCOVERY AND SO CONNECT WITH US

8426
05:12:45,158 --> 05:12:46,660
AND YOU CAN LEARN MORE ABOUT OUR

8427
05:12:46,660 --> 05:12:47,193
PROGRAM.

8428
05:12:47,193 --> 05:12:47,861
ALL RIGHT, EVERYONE, I'M GOING

8429
05:12:47,861 --> 05:12:49,763
TO CLOSE IT UP AND ADJOURN.

8430
05:12:49,763 --> 05:12:52,866
I AGAIN, THANK YOU SO MUCH

8431
05:12:52,866 --> 05:12:53,900
EVERYONE, SPEAKERS, ORGANIZERS,

8432
05:12:53,900 --> 05:12:56,469
EVERYONE, HAVE A GOOD 1! 

8433
05:12:56,469 --> 05:13:06,846
>> ALL RIGHT, BYE.