1
00:00:04,910 --> 00:00:08,113
THANKS YOU SO MUCH FOR EVERYONE

2
00:00:08,113 --> 00:00:11,216
WHO CAME IN PERSON AND OTHERS

3
00:00:11,216 --> 00:00:14,052
WHO ARE ON THE ZOOM.

4
00:00:14,052 --> 00:00:17,556
IT'S REALLY A PLEASURE TO HAVE

5
00:00:17,556 --> 00:00:21,026
DR. PENG JIANG AS OUR CCR GRAND

6
00:00:21,026 --> 00:00:23,628
ROUNDS SPEAKER TODAY.

7
00:00:23,628 --> 00:00:28,333
SO, HE JOINED OUR NEW BRANCH IN

8
00:00:28,333 --> 00:00:30,502
JULY 2019, SO HE'S AROUND FOUR

9
00:00:30,502 --> 00:00:31,636
YEARS HERE.

10
00:00:31,636 --> 00:00:35,307
AND HE WILL SHOW YOU THE STELLAR

11
00:00:35,307 --> 00:00:37,609
WORK HE'S BEEN DOING IN THIS

12
00:00:37,609 --> 00:00:39,311
RELATIVELY SHORT TIME.

13
00:00:39,311 --> 00:00:43,415
SO HE AS YOU MAY KNOW WORKS ON

14
00:00:43,415 --> 00:00:46,985
ARTIFICIAL INTELLIGENCE AND

15
00:00:46,985 --> 00:00:49,221
COMPUTATIONAL BIOLOGY, AND

16
00:00:49,221 --> 00:00:50,922
IMMUNOTHERAPY FROM BASIC AND

17
00:00:50,922 --> 00:00:53,692
TRANSLATIONAL ASPECTS, AND HIS

18
00:00:53,692 --> 00:00:56,461
LAB INTEGRATES BOTH

19
00:00:56,461 --> 00:00:57,596
COMPUTATIONAL ANALYSIS AND

20
00:00:57,596 --> 00:01:06,938
EXPERIMENTAL ANIMAL WORK.

21
00:01:06,938 --> 00:01:08,006
SO, PENG COMPLETED UNDERGRAD IN

22
00:01:08,006 --> 00:01:11,276
ONE OF THE BEST UNIVERSITIES IN

23
00:01:11,276 --> 00:01:13,011
CHINA, IN THE WORLD, AND QUITE

24
00:01:13,011 --> 00:01:17,415
AMAZINGLY HE HAS DONE THAT WITH

25
00:01:17,415 --> 00:01:18,617
THE HIGHEST NATIONAL HONORS

26
00:01:18,617 --> 00:01:22,554
POSSIBLE, AND IN A SMALL COUNTRY

27
00:01:22,554 --> 00:01:25,357
LIKE CHINA IT'S NOT TRIVIAL

28
00:01:25,357 --> 00:01:26,324
FEAT.

29
00:01:26,324 --> 00:01:28,159
IT'S JUST AMAZING.

30
00:01:28,159 --> 00:01:29,661
AND THEN LESS IMPRESSIVE THINGS,

31
00:01:29,661 --> 00:01:36,935
HE WENT AND DID HIS Ph.D. IN

32
00:01:36,935 --> 00:01:37,869
PRINCETON, POSTDOC AT

33
00:01:37,869 --> 00:01:39,671
DANA-FARBER HARVARD, NOT AS

34
00:01:39,671 --> 00:01:45,176
IMPRESSIVE AS, YOU KNOW, GETTING

35
00:01:45,176 --> 00:01:49,814
THE NATIONAL HONOR IN CHINA, I

36
00:01:49,814 --> 00:01:50,882
WOULD THINK.

37
00:01:50,882 --> 00:01:52,784
I DIDN'T EVEN GET A NATIONAL

38
00:01:52,784 --> 00:01:55,086
HONOR IN A SMALL REGIONAL PLACE

39
00:01:55,086 --> 00:01:57,689
IN ISRAEL, SO, YOU KNOW, WHICH

40
00:01:57,689 --> 00:02:05,096
IS INHERENTLY A VERY SMALL

41
00:02:05,096 --> 00:02:05,297
PLACE.

42
00:02:05,297 --> 00:02:10,402
AND OUR AND PATHS MET IN 2018,

43
00:02:10,402 --> 00:02:13,572
BOTH OUR LABS, WE PUBLISHED

44
00:02:13,572 --> 00:02:15,640
PAPERS IN NATURE MEDICINE, I

45
00:02:15,640 --> 00:02:17,075
THINK SIDE BY SIDE.

46
00:02:17,075 --> 00:02:17,442
>> YES.

47
00:02:17,442 --> 00:02:21,012
>> SIDE BY SIDE EVEN.

48
00:02:21,012 --> 00:02:23,114
AND ABOUT PREDICTING SIGNATURES

49
00:02:23,114 --> 00:02:25,216
OF EXPRESSION FOR IMMUNOTHERAPY,

50
00:02:25,216 --> 00:02:30,655
BUT THE SMALL DETAIL IS THAT

51
00:02:30,655 --> 00:02:31,856
PENG'S WORK IS CITED FIVE TIMES

52
00:02:31,856 --> 00:02:33,992
MORE THAN OUR WORK.

53
00:02:33,992 --> 00:02:38,430
BY NOW I JUST CHECKED YESTERDAY,

54
00:02:38,430 --> 00:02:40,432
2500 TIMES, IT WAS PUBLISHED

55
00:02:40,432 --> 00:02:44,269
JUST IN 2018.

56
00:02:44,269 --> 00:02:47,172
SO, JUST TO GIVE YOU AN IDEA.

57
00:02:47,172 --> 00:02:52,944
SINCE THEN, SINCE HE JOINED HIS 

58
00:02:52,944 --> 00:02:56,948
OWN LAB, PUBLISHED PAPERS IN

59
00:02:56,948 --> 00:02:57,549
NATURE MEDICINE, NATURE

60
00:02:57,549 --> 00:03:01,119
COMPUTATION, DOING STELLAR WORK.

61
00:03:01,119 --> 00:03:07,225
HE RECEIVED THE NCI K99 PATHWAY

62
00:03:07,225 --> 00:03:09,527
TO INDEPENDENCE AWARD,

63
00:03:09,527 --> 00:03:13,198
SCHOLAR-IN-TRAINING AWARD,

64
00:03:13,198 --> 00:03:16,368
TECHNOLOGY AND INNOVATION AWARD,

65
00:03:16,368 --> 00:03:23,008
CRI, AND MOSTLY SIMPLY NCI AWARD

66
00:03:23,008 --> 00:03:24,609
FOR DATA SCIENCE.

67
00:03:24,609 --> 00:03:27,746
WE'RE SO LUCKY TO HAVE HIM,

68
00:03:27,746 --> 00:03:30,281
BEYOND BEING AN AMAZING

69
00:03:30,281 --> 00:03:31,416
SCIENTIST HE'S A WONDERFUL

70
00:03:31,416 --> 00:03:34,285
PERSON AND VERY KIND AND

71
00:03:34,285 --> 00:03:35,487
GENEROUS COLLEAGUE, AS THOSE OF

72
00:03:35,487 --> 00:03:38,289
YOU WHO WORK WITH HIM KNOW, AND

73
00:03:38,289 --> 00:03:41,760
ANYTIME I TALK TO HIM I LEARN,

74
00:03:41,760 --> 00:03:43,561
YOU KNOW, SO WHAT CAN SOMEONE

75
00:03:43,561 --> 00:03:43,995
ASK FOR MORE?

76
00:03:43,995 --> 00:03:52,971
THANK YOU SO MUCH FOR COMING TO

77
00:03:52,971 --> 00:03:55,206
TALK TO US.

78
00:03:55,206 --> 00:03:59,544
>> SO LUCKY FOR ME, LIKE
DATA

79
00:03:59,544 --> 00:04:04,449
SCIENCE BRANCH, FIRST YOUNG 

80
00:04:04,449 --> 00:04:05,150
FACULTY HAS RECRUITED.

81
00:04:05,150 --> 00:04:08,186
TODAY I WOULD LIKE TO SHARE SOME

82
00:04:08,186 --> 00:04:10,689
OF OUR RECENT RESEARCH WORK,

83
00:04:10,689 --> 00:04:16,561
UTILIZING BIG DATA APPROACH TO

84
00:04:16,561 --> 00:04:17,328
STUDY INTRACELLULAR SIGNATURE.

85
00:04:17,328 --> 00:04:22,200
BEFORE STARTING I WOULD LIKE TO 

86
00:04:22,200 --> 00:04:24,035
RECOMMEND REVIEW OF PAPER THAT

87
00:04:24,035 --> 00:04:26,738
WE AND OTHER COLLEAGUES WROTE

88
00:04:26,738 --> 00:04:35,213
ABOUT APPLICATION DATA AND A.I.

89
00:04:35,213 --> 00:04:41,953
IN TRANSLATIONAL RESEARCH. 

90
00:04:41,953 --> 00:04:42,954
CANCER RESEARCHERS HERE, ALSO

91
00:04:42,954 --> 00:04:47,092
POTENTIAL PROBLEMS OF A.I.,

92
00:04:47,092 --> 00:04:48,126
LIMITATIONS.

93
00:04:48,126 --> 00:04:49,961
FOR EXAMPLE NOW WE SEE LOTS OF

94
00:04:49,961 --> 00:04:55,533
PAPER SHOWING A.I. IS DOING

95
00:04:55,533 --> 00:04:59,838
FANTASTIC JOB, SOME DIGITAL

96
00:04:59,838 --> 00:05:01,706
PATHOLOGY, CLASSIFICATION TASKS.

97
00:05:01,706 --> 00:05:07,912
REVIEW PAPER CITE LIKE REPORT.

98
00:05:07,912 --> 00:05:10,515
BUT ALL OF THE EVALUATIONS ARE

99
00:05:10,515 --> 00:05:13,118
BASED ON THE PERFORMANCE ON ONE

100
00:05:13,118 --> 00:05:16,521
SINGLE IMAGE, CLASSIFICATION.

101
00:05:16,521 --> 00:05:18,022
HOWEVER, CLINICAL DECISIONS FOR

102
00:05:18,022 --> 00:05:20,992
EXAMPLE IF PATHOLOGISTS ARE NOT

103
00:05:20,992 --> 00:05:25,697
SURE ABOUT THE CANCER OR

104
00:05:25,697 --> 00:05:28,299
NON-CANCER DECISION, USE MARKERS

105
00:05:28,299 --> 00:05:38,743
TO FIND DEFINITIVE ANSWER.

106
00:05:38,977 --> 00:05:40,111
(INDISCERNIBLE) TO SEE WHETHER

107
00:05:40,111 --> 00:05:42,080
IT'S TRUE AND NECESSARY.

108
00:05:42,080 --> 00:05:49,454
ALSO SOME PAPER FROM OUR NCI

109
00:05:49,454 --> 00:05:55,660
COLLEAGUES, PREDICTION PROBLEM 

110
00:05:55,660 --> 00:05:58,563
WHICH THEY CLAIM COLLEAGUES

111
00:05:58,563 --> 00:05:59,798
SOMEHOW -- SINGLE POINT

112
00:05:59,798 --> 00:06:04,469
MUTATIONS CAN STOP A LOT OF 

113
00:06:04,469 --> 00:06:05,770
STRUCTURES.

114
00:06:05,770 --> 00:06:07,739
FOR EXAMPLE, NOTATION DAMAGE

115
00:06:07,739 --> 00:06:11,075
HYDROPHOBIC POCKET IN THIS

116
00:06:11,075 --> 00:06:14,479
PROTEIN, LOSS OF FUNCTION, OTHER

117
00:06:14,479 --> 00:06:18,082
SIMPLY PREDICT SAME STRUCTURE,

118
00:06:18,082 --> 00:06:20,385
BASICALLY SIMILARITY IN

119
00:06:20,385 --> 00:06:21,352
MUTATION.

120
00:06:21,352 --> 00:06:23,721
IN GENERAL THIS REVIEW PAPER

121
00:06:23,721 --> 00:06:25,657
LOOKS TO BE CORRECT CITATION FOR

122
00:06:25,657 --> 00:06:31,162
A.I. HOW SHOULD YOU APPLY A.I. 

123
00:06:31,162 --> 00:06:37,735
IN CANCER TRANSLATIONAL

124
00:06:37,735 --> 00:06:38,336
RESEARCH.

125
00:06:38,336 --> 00:06:45,176
MAJOR TOPIC FOR TODAY IS

126
00:06:45,176 --> 00:06:46,044
INTRACELLULAR COMMUNICATION,

127
00:06:46,044 --> 00:06:53,718
RECEDING SIGNAL IS BASIC NOTE 

128
00:06:53,718 --> 00:06:54,619
ALMOST ALL IMMUNOLOGIC PRESENCE

129
00:06:54,619 --> 00:06:57,188
INCLUDING TUMOR EVASION.

130
00:06:57,188 --> 00:07:00,491
WE'VE BEEN FOCUSED ON DEVELOPING

131
00:07:00,491 --> 00:07:03,595
COMPUTATIONAL SOLUTIONS, 

132
00:07:03,595 --> 00:07:05,330
INTRACELLULAR COMMUNICATION WITH

133
00:07:05,330 --> 00:07:08,700
APPLICATION IN CANCER

134
00:07:08,700 --> 00:07:12,537
IMMUNOTHERAPIES. 

135
00:07:12,537 --> 00:07:16,908
FIRST WORK FOR STUDYING

136
00:07:16,908 --> 00:07:18,142
SIGNALING ACTIVITIES, DEVELOPED

137
00:07:18,142 --> 00:07:23,481
A COMPUTATIONAL MODEL IN TUMOR

138
00:07:23,481 --> 00:07:23,915
MICROENVIRONMENT.

139
00:07:23,915 --> 00:07:25,116
ALSO LIKE TO FIND

140
00:07:25,116 --> 00:07:29,320
(INDISCERNIBLE) BASED ON THIS

141
00:07:29,320 --> 00:07:30,588
MODEL.

142
00:07:30,588 --> 00:07:32,824
FINALLY INTRODUCING THE

143
00:07:32,824 --> 00:07:39,130
FRAMEWORK FOR ANALYZING DATA IN 

144
00:07:39,130 --> 00:07:43,468
CANCER RESEARCH.

145
00:07:43,468 --> 00:07:45,637
FIRST OFF, CYTOKINE ACTIVITY

146
00:07:45,637 --> 00:07:54,479
RESEARCH MOST COMMON APPROACH

147
00:07:54,479 --> 00:07:57,715
IMMUNOLOGIST APPLY, CYTOKINE

148
00:07:57,715 --> 00:07:59,384
RELEASE IS TRANSIENT PROCESS.

149
00:07:59,384 --> 00:08:05,790
IT WOULD MISS THE ACTIVITY.

150
00:08:05,790 --> 00:08:08,559
NOT EQUAL TO FUNCTION ACTIVITY.

151
00:08:08,559 --> 00:08:10,561
CYTOKINES COULD HAVE FUNCTION

152
00:08:10,561 --> 00:08:12,463
COMPLEXITY AND REDUNDANCY SO ALL

153
00:08:12,463 --> 00:08:15,667
OF THIS MEANS MORE EFFECTIVE

154
00:08:15,667 --> 00:08:20,838
APPROACH TO STUDY CYTOKINE

155
00:08:20,838 --> 00:08:21,739
SIGNALING ACTIVITIES.

156
00:08:21,739 --> 00:08:22,440
DEVELOPED THIS FRAMEWORK TO

157
00:08:22,440 --> 00:08:23,308
ANSWER TWO QUESTIONS.

158
00:08:23,308 --> 00:08:25,276
FIRST QUESTION IF YOU HAVE A

159
00:08:25,276 --> 00:08:30,214
POSITIVE FOCUS WHAT ARE THE

160
00:08:30,214 --> 00:08:31,849
CYTOKINES GENE FOCUSED, AND

161
00:08:31,849 --> 00:08:34,285
SECOND IF YOU HAVE SAMPLE WITH

162
00:08:34,285 --> 00:08:35,453
GENE EXPRESSION PROFILE

163
00:08:35,453 --> 00:08:37,155
AVAILABLE WHAT ARE CYTOKINE

164
00:08:37,155 --> 00:08:39,090
SIGNALING ACTIVITIES IN YOUR

165
00:08:39,090 --> 00:08:39,424
SAMPLE?

166
00:08:39,424 --> 00:08:42,827
SO THE FIRST STEP TO CREATE

167
00:08:42,827 --> 00:08:53,271
FRAMEWORK IS ALWAYS DATA 

168
00:08:58,743 --> 00:08:59,110
COLLECTION.

169
00:08:59,110 --> 00:09:01,446
WE HYPOTHESIZE THIS WAS

170
00:09:01,446 --> 00:09:03,848
UTILIZED, COULD SERVE AS

171
00:09:03,848 --> 00:09:05,516
KNOWLEDGE BASE TO MAKE KNOWLEDGE

172
00:09:05,516 --> 00:09:10,421
DISCOVERIES FOR NEW PROBLEMS.

173
00:09:10,421 --> 00:09:14,859
HOWEVER, KEY CHALLENGE IS THAT

174
00:09:14,859 --> 00:09:18,563
LIKE THE DATA DEPOSIT IN

175
00:09:18,563 --> 00:09:20,264
DIFFERENT DATABASE ARE

176
00:09:20,264 --> 00:09:20,798
HETEROGENEOUS, ESPECIALLY

177
00:09:20,798 --> 00:09:22,867
METADATA A FOR EXAMPLE LIKE THEY

178
00:09:22,867 --> 00:09:24,836
GIVE YOU A FREE TOUCH TO

179
00:09:24,836 --> 00:09:26,237
DESCRIBE EACH TIME THEY DO,

180
00:09:26,237 --> 00:09:29,374
HOWEVER THIS FREE TOUCH COULD BE

181
00:09:29,374 --> 00:09:33,478
SO DIFFERENT ACROSS, THERE'S NO 

182
00:09:33,478 --> 00:09:36,014
WAY YOU CAN EASILY KNOW

183
00:09:36,014 --> 00:09:37,081
CONDITION, DURATION, CELL MODEL.

184
00:09:37,081 --> 00:09:41,452
YOU HAVE TO READ THEM AND EXACT

185
00:09:41,452 --> 00:09:43,488
THIS KEY INFORMATION.

186
00:09:43,488 --> 00:09:45,089
DATA INTEGRATION PROBLEM, FIRST

187
00:09:45,089 --> 00:09:47,725
DEVELOP A FRAMEWORK CALLED

188
00:09:47,725 --> 00:09:55,633
FRAMEWORK FOR DATA CURATION, 

189
00:09:55,633 --> 00:09:57,902
AVAILABLE HERE.

190
00:09:57,902 --> 00:10:04,042
SO UTILIZE FUNCTIONS TO HELP YOU

191
00:10:04,042 --> 00:10:05,410
TO REANNOTATE DIFFERENT EDITIONS

192
00:10:05,410 --> 00:10:10,415
TO MAKE THEM POSSIBLE FOR

193
00:10:10,415 --> 00:10:12,417
ALGORITHMIC AUTOMATIC ANALYSIS.

194
00:10:12,417 --> 00:10:15,386
WITH THE HELP OF FRAMEWORK

195
00:10:15,386 --> 00:10:17,955
COLLECT MORE THAN 20,000 KITE

196
00:10:17,955 --> 00:10:18,790
SIGN TREATMENT RESPONSE

197
00:10:18,790 --> 00:10:21,459
PROFILES, EACH PROFILE HAVE AT

198
00:10:21,459 --> 00:10:28,399
LEAST TWO BIOLOGICAL REPLICASE,

199
00:10:28,399 --> 00:10:31,269
SO DATA PROFILE IS A LOT MORE

200
00:10:31,269 --> 00:10:40,745
THAN THIS 20,000 NUMBER, SOME

201
00:10:40,745 --> 00:10:42,780
CYTOKINES INTERFERENCE HAVE

202
00:10:42,780 --> 00:10:47,452
LARGE DATASET COLLECTED, FOR

203
00:10:47,452 --> 00:10:49,253
MOST CYTOKINES, THERE'S EITHER

204
00:10:49,253 --> 00:10:51,823
ZERO OR VERY FEW DATASETS

205
00:10:51,823 --> 00:10:52,356
COLLECTED.

206
00:10:52,356 --> 00:10:56,260
AND HERE ARE SOME EXAMPLES OF

207
00:10:56,260 --> 00:10:56,894
DATASET.

208
00:10:56,894 --> 00:10:58,196
EACH POINT MEANS ONE TREATMENT

209
00:10:58,196 --> 00:11:04,635
RESPONSE EXPERIMENT IN SOME CELL

210
00:11:04,635 --> 00:11:05,103
MODEL.

211
00:11:05,103 --> 00:11:09,073
AND Y-AXIS IS TARGETED GENE.

212
00:11:09,073 --> 00:11:11,209
X-AXIS IS TREATMENT CYTOKINE.

213
00:11:11,209 --> 00:11:16,013
YOU CAN SEE DATASET CAPTURED THE

214
00:11:16,013 --> 00:11:17,148
WELL KNOWN LIKE CYTOKINE

215
00:11:17,148 --> 00:11:21,352
RESPONSE CASCADE.

216
00:11:21,352 --> 00:11:23,488
FOR EXAMPLE, IL-12 INDUCE

217
00:11:23,488 --> 00:11:25,022
INTERFERON GAMMA, INDUCE BY

218
00:11:25,022 --> 00:11:27,925
IL-12, MOST OF THIS EXPERIMENT,

219
00:11:27,925 --> 00:11:35,399
AND THEN INTERFERON GAMMA WILL

220
00:11:35,399 --> 00:11:37,368
INDUCE (INDISCERNIBLE) DATA

221
00:11:37,368 --> 00:11:37,935
COLLECTION.

222
00:11:37,935 --> 00:11:44,175
YOU CAN ALSO SEE OVERALL WHAT IS

223
00:11:44,175 --> 00:11:49,580
THE CYTOKINE REGULATORY CASCADE.

224
00:11:49,580 --> 00:11:50,314
X-AXIS.

225
00:11:50,314 --> 00:11:52,383
Y-AXIS IS LEVEL TWO CYTOKINE.

226
00:11:52,383 --> 00:11:59,323
WE CAN SEE SOME TNF ALPHA WORKS

227
00:11:59,323 --> 00:12:04,795
THROUGH, INDUCE A BUNCH OF

228
00:12:04,795 --> 00:12:06,964
CHEMOKINES, INTERFERENCE THROUGH

229
00:12:06,964 --> 00:12:08,799
THE STAT1 PATHWAY INTRODUCE A

230
00:12:08,799 --> 00:12:11,369
DIFFERENT SET OF CYTOKINES.

231
00:12:11,369 --> 00:12:13,771
WITH THIS DATASET CAN CHECK

232
00:12:13,771 --> 00:12:14,705
CYTOKINES CAN WORK TOGETHER OR

233
00:12:14,705 --> 00:12:17,475
FIGHT AGAINST EACH OTHER.

234
00:12:17,475 --> 00:12:23,981
MOST OF THE TIME CYTOKINES

235
00:12:23,981 --> 00:12:24,348
(INDISCERNIBLE).

236
00:12:24,348 --> 00:12:28,586
FOR EXAMPLE, INDUCE THE PATHWAY,

237
00:12:28,586 --> 00:12:31,856
WE CAN SEE THE CO-INDUCE OR

238
00:12:31,856 --> 00:12:35,893
CO-REPRESS A LOT OF TARGET.

239
00:12:35,893 --> 00:12:37,228
CYTOKINES WILL FIGHT AGAINST

240
00:12:37,228 --> 00:12:38,763
EACH OTHER.

241
00:12:38,763 --> 00:12:41,866
HERE IS GOOD EXAMPLE.

242
00:12:41,866 --> 00:12:49,807
IL-1 BETA INDUCE SOME

243
00:12:49,807 --> 00:12:53,778
CHEMOKINES, THEY ALL SHARE THE

244
00:12:53,778 --> 00:12:57,248
SAME RECEPTOR.

245
00:12:57,248 --> 00:12:58,449
BUT CONSIDERED GROWTH FACTOR

246
00:12:58,449 --> 00:12:59,750
BEFORE, TENDS TO DOWNREGULATE

247
00:12:59,750 --> 00:13:00,051
THIS TARGET.

248
00:13:00,051 --> 00:13:02,553
TO THE BEST OF OUR KNOWLEDGE

249
00:13:02,553 --> 00:13:06,991
BEFORE THIS THERE WAS NO REPORT

250
00:13:06,991 --> 00:13:08,659
ABOUT THE (INDISCERNIBLE) OF BMP

251
00:13:08,659 --> 00:13:16,234
6 GROWTH FACTOR AGAINST IL-1

252
00:13:16,234 --> 00:13:16,434
BETA.

253
00:13:16,434 --> 00:13:17,735
WE PERFORM EXPERIMENTAL DATA.

254
00:13:17,735 --> 00:13:20,404
COVID WAS HOT TOPIC DEALING WITH

255
00:13:20,404 --> 00:13:22,306
THIS PRODUCT.

256
00:13:22,306 --> 00:13:26,510
WE TREAT LUNG EPITHELIAL CELL 

257
00:13:26,510 --> 00:13:30,781
LINE WITH IL-1 BETA, AND YOU CAN

258
00:13:30,781 --> 00:13:32,950
SEE INTRACELLULAR PROTEIN LEVEL

259
00:13:32,950 --> 00:13:38,089
BY FLOW CYTOMETRY AND SECRETION

260
00:13:38,089 --> 00:13:44,362
LEVEL BY ELISA, BMP6

261
00:13:44,362 --> 00:13:50,735
SIGNIFICANTLY CONTROL INDUCTION.

262
00:13:50,735 --> 00:13:53,537
SO THIS MOLECULE, AND KEY GOAL

263
00:13:53,537 --> 00:13:56,841
OF THE FRAMEWORK TO ACHIEVE

264
00:13:56,841 --> 00:13:59,010
CLINICAL PREDICTION OF FUNCTION.

265
00:13:59,010 --> 00:14:01,545
SO, IN ORDER TO PREDICT CYTOKINE

266
00:14:01,545 --> 00:14:04,181
ACTIVITY FROM CLINICAL SAMPLE WE

267
00:14:04,181 --> 00:14:05,850
USE VERY SIMPLE LINEAR

268
00:14:05,850 --> 00:14:07,418
REGRESSION MODEL.

269
00:14:07,418 --> 00:14:10,354
THE MATRIX IS SIMPLY CYTOKINE

270
00:14:10,354 --> 00:14:16,927
TREATMENT RESPONSE PROFILES WE

271
00:14:16,927 --> 00:14:22,266
COLLECT, ON THE RESPONSE 

272
00:14:22,266 --> 00:14:25,403
EXPRESSION PROFILE, REGRESSION

273
00:14:25,403 --> 00:14:27,471
COEFFICIENT, IF IT'S LOW

274
00:14:27,471 --> 00:14:27,838
(INDISCERNIBLE).

275
00:14:27,838 --> 00:14:29,473
HERE IS EXAMPLE.

276
00:14:29,473 --> 00:14:34,178
THIS DATA IS FROM CLINICAL TRIAL

277
00:14:34,178 --> 00:14:37,548
OF A PATIENT TREATED WITH IL-1

278
00:14:37,548 --> 00:14:38,749
BETA NEUTRALIZING ANTIBODY.

279
00:14:38,749 --> 00:14:42,520
IN THIS CLINICAL TRIAL THEY HAVE

280
00:14:42,520 --> 00:14:44,021
PLACEBO GROUP.

281
00:14:44,021 --> 00:14:45,756
THEY ALSO DEFINE DIFFERENT LEVEL

282
00:14:45,756 --> 00:14:49,260
OF PATIENT RESPONSE.

283
00:14:49,260 --> 00:14:53,898
DAY 15 AFTER IL-1 NEUTRALIZING

284
00:14:53,898 --> 00:14:54,131
THERAPY.

285
00:14:54,131 --> 00:14:58,669
THEY ALSO GENE EXPRESSION

286
00:14:58,669 --> 00:15:02,940
PROFILE PATIENT SAMPLES, YOU CAN

287
00:15:02,940 --> 00:15:06,344
SEE PREDICTION OF IL-1 BETA

288
00:15:06,344 --> 00:15:09,814
ACTIVITY, HINGE ON THIS BECAUSE

289
00:15:09,814 --> 00:15:11,515
IT'S Y-AXIS, IT'S REALLY

290
00:15:11,515 --> 00:15:15,853
ACTUALLY CLINICAL RESPONSE ON

291
00:15:15,853 --> 00:15:16,987
DAY 15 WHICH MEANS ALMOST

292
00:15:16,987 --> 00:15:20,157
PREDICT WHAT WILL BE THE TRUE

293
00:15:20,157 --> 00:15:22,993
RESPONDERS TO THIS NEUTRALIZING

294
00:15:22,993 --> 00:15:27,331
ANTIBODY. 

295
00:15:27,331 --> 00:15:31,535
HERE ANOTHER CLINICAL STUDY,

296
00:15:31,535 --> 00:15:32,770
CLINICAL ENDPOINT IS INTERFERON

297
00:15:32,770 --> 00:15:38,109
ALPHA IN THE PATIENT BLOOD.

298
00:15:38,109 --> 00:15:46,417
PBMC EXPRESSION PROFILE FOR THIS

299
00:15:46,417 --> 00:15:51,889
PATIENT, PREDICTION, CLINICAL 

300
00:15:51,889 --> 00:15:52,456
ENDPOINT.

301
00:15:52,456 --> 00:15:55,726
IN CANCER THE MOST STUDY IS EGF

302
00:15:55,726 --> 00:16:05,002
THERAPY, SO TAKE TWO CLINICAL

303
00:16:05,002 --> 00:16:13,310
STUDIES, KINASE INHIBITOR, AND

304
00:16:13,310 --> 00:16:13,778
BEVACIZUMAB.

305
00:16:13,778 --> 00:16:17,848
IF THEY HAVE HIGHER ACTIVITY

306
00:16:17,848 --> 00:16:21,018
PREDICTED, YOU CAN SEE BETTER

307
00:16:21,018 --> 00:16:26,023
CLINICAL OUTCOME.

308
00:16:26,023 --> 00:16:28,659
REMARKABLY NCI COLLEAGUE

309
00:16:28,659 --> 00:16:30,761
PERFORMED THIS LARGE SCALE

310
00:16:30,761 --> 00:16:34,031
CLINICAL TRIAL OF BEVACIZUMAB

311
00:16:34,031 --> 00:16:35,666
SENSE PLACEBO IN GLIOBLASTOMA,

312
00:16:35,666 --> 00:16:39,370
AND CLINICAL TRIAL RESULT IS

313
00:16:39,370 --> 00:16:43,040
SHOWN NO BENEFIT OF BEVACIZUMAB

314
00:16:43,040 --> 00:16:45,576
COMPARED TO PLACEBO.

315
00:16:45,576 --> 00:16:47,478
HOWEVER, RETROSPECTIVE ANALYSIS

316
00:16:47,478 --> 00:16:48,913
OF CLINICAL TRIAL DATA WHICH

317
00:16:48,913 --> 00:16:51,015
THEY HAVE PRE-TREATMENT

318
00:16:51,015 --> 00:16:51,682
EXPRESSION PROFILES MEANS THERE

319
00:16:51,682 --> 00:16:52,850
ARE SOME EFFICACY.

320
00:16:52,850 --> 00:16:59,123
YOU HAVE TO FOCUS ON THE PATIENT

321
00:16:59,123 --> 00:17:01,525
WITH HIGHER VEGF THERAPY, NEED

322
00:17:01,525 --> 00:17:07,264
TO TARGET IN THE TUMOR FIRST.

323
00:17:07,264 --> 00:17:09,066
POTENTIALLY MAY RESCUE CLINICAL

324
00:17:09,066 --> 00:17:11,702
TRIAL IF YOU HAVE PROTEIN

325
00:17:11,702 --> 00:17:14,205
EXPRESSION PROFILE AVAILABLE.

326
00:17:14,205 --> 00:17:17,942
WE COLLECT MANY DATA CYCLE OF

327
00:17:17,942 --> 00:17:21,812
CYTOKINE NEUTRALIZING THERAPY

328
00:17:21,812 --> 00:17:25,549
FROM NCBIG, YOU CAN SEE

329
00:17:25,549 --> 00:17:28,152
PREDICTION WHICH MEANS IF CELL

330
00:17:28,152 --> 00:17:29,487
SIGNAL PREDICT REDUCTION OF

331
00:17:29,487 --> 00:17:37,261
CYTOKINE ACTIVITY, THE PATIENT 

332
00:17:37,261 --> 00:17:42,333
WILL BE THE RESPONDER.

333
00:17:42,333 --> 00:17:47,104
HERE IS OUR FRAMEWORK.

334
00:17:47,104 --> 00:17:48,272
INTERACTIVE MODELS, RESEARCH

335
00:17:48,272 --> 00:17:49,573
DATA, DEMONSTRATE DATA

336
00:17:49,573 --> 00:17:52,209
COLLECTION CAN REVEAL NEW

337
00:17:52,209 --> 00:17:54,078
KNOWLEDGE SUCH AS

338
00:17:54,078 --> 00:17:57,081
ANTI-INFLAMMATORY MOLECULE 

339
00:17:57,081 --> 00:17:59,016
FIGHTING AGAINST IL-1 BETA.

340
00:17:59,016 --> 00:18:00,818
FURTHER CREATE ANOTHER

341
00:18:00,818 --> 00:18:03,354
COMPUTATIONAL MODEL T RAS TO

342
00:18:03,354 --> 00:18:13,864
ADDRESS IMPORTANT QUESTION IN

343
00:18:16,066 --> 00:18:16,467
IMMUNOLOGY.

344
00:18:16,467 --> 00:18:20,137
(INDISCERNIBLE) THE OTHER ARE

345
00:18:20,137 --> 00:18:20,704
NON-RESPONDERS.

346
00:18:20,704 --> 00:18:27,645
WORKS IN ONE PART OF THE MODEL,

347
00:18:27,645 --> 00:18:29,179
NO TREATMENT.

348
00:18:29,179 --> 00:18:34,985
THIS ADOPTIVE TIL TRANSFER DOES

349
00:18:34,985 --> 00:18:41,458
NOT WORK ANYMORE, CAR-T WORKS IN

350
00:18:41,458 --> 00:18:41,926
SOME.

351
00:18:41,926 --> 00:18:43,594
ONE REASON FOR LIMITED

352
00:18:43,594 --> 00:18:45,563
THERAPEUTIC EFFICACY FOR SOLID

353
00:18:45,563 --> 00:18:50,267
TUMORS IS IT'S SUCH AN

354
00:18:50,267 --> 00:18:52,536
IMMUNOSUPPRESSIVE ENVIRONMENT.

355
00:18:52,536 --> 00:18:54,371
CANCER CELLS, TUMOR CONSIDER TO

356
00:18:54,371 --> 00:19:04,815
FIGHT BACK, ONLY ONE OF MY

357
00:19:05,082 --> 00:19:06,050
REASONS.

358
00:19:06,050 --> 00:19:10,120
IT'S SUCH A SUPPRESSIVE

359
00:19:10,120 --> 00:19:10,421
ENVIRONMENT.

360
00:19:10,421 --> 00:19:13,023
SO, TO THINK OUT A SOLUTION

361
00:19:13,023 --> 00:19:20,264
STILL FOCUS ON CYTOKINE, LOOK AT

362
00:19:20,264 --> 00:19:22,199
T CELL.

363
00:19:22,199 --> 00:19:25,169
WE ANALYZE WHICH T CELL CAN DO

364
00:19:25,169 --> 00:19:26,503
SO WELL IN SUPPRESSIVE

365
00:19:26,503 --> 00:19:27,171
ENVIRONMENT.

366
00:19:27,171 --> 00:19:29,139
THIS IS OUR MODEL.

367
00:19:29,139 --> 00:19:32,209
IN THIS EXAMPLE UTILIZE SINGLE

368
00:19:32,209 --> 00:19:36,447
CELL STUDY OF T CELL FROM

369
00:19:36,447 --> 00:19:37,982
MELANOMA TUMOR.

370
00:19:37,982 --> 00:19:45,456
THIS WAS PUBLISHED PREVIOUS

371
00:19:45,456 --> 00:19:46,223
GRADUATE STUDENT.

372
00:19:46,223 --> 00:19:52,129
T CELLS, X-AXIS IS THE TGF-BETA

373
00:19:52,129 --> 00:19:54,398
ACTIVITY PREDICTED.

374
00:19:54,398 --> 00:19:55,566
INDICATOR OF IMMUNOSUPPRESSIVE

375
00:19:55,566 --> 00:19:57,501
IS RECEDING.

376
00:19:57,501 --> 00:20:03,140
ON THE Y-AXIS THE CELL CYCLE

377
00:20:03,140 --> 00:20:09,713
ACTIVITY PREDICTED USING KEGG

378
00:20:09,713 --> 00:20:11,915
CELL CYCLE PATHWAY.

379
00:20:11,915 --> 00:20:14,785
THE CORRELATION IS NOT TRUE FOR

380
00:20:14,785 --> 00:20:18,589
ALL THE T CELLS. 

381
00:20:18,589 --> 00:20:29,133
ACCORDING TO EXPRESSION MARKER, 

382
00:20:32,569 --> 00:20:35,239
OBSERVE DIFFERENT LOCALIZATIONS.

383
00:20:35,239 --> 00:20:40,511
THIS GRAPH SHOW BASIC IDEA OF

384
00:20:40,511 --> 00:20:42,446
MODEL SEARCHING FOR MOLECULAR

385
00:20:42,446 --> 00:20:43,480
MARKERS, MODULATE

386
00:20:43,480 --> 00:20:46,950
IMMUNOSUPPRESSION AND T CELL

387
00:20:46,950 --> 00:20:48,018
PROLIFERATION, UTILIZE TWO, TO

388
00:20:48,018 --> 00:20:50,954
IDENTIFY FROM MARKER GENES OF

389
00:20:50,954 --> 00:20:54,858
TUMOR RESILIENT T CELLS, FIRST

390
00:20:54,858 --> 00:20:56,393
STAGE, QUANTIFIES SIGNAL EACH T

391
00:20:56,393 --> 00:21:02,466
CELL RECEIVE FROM TUMOR

392
00:21:02,466 --> 00:21:04,234
MICROENVIRONMENT USING SINGLE

393
00:21:04,234 --> 00:21:05,502
CELL STUDY.

394
00:21:05,502 --> 00:21:07,838
IN SECOND STATE YOU COULD HAVE T

395
00:21:07,838 --> 00:21:09,606
CELL WHO RECEIVED LOTS OF

396
00:21:09,606 --> 00:21:14,845
SUPPRESSIVE SIGNAL BUT ARE STILL

397
00:21:14,845 --> 00:21:15,479
PROLIFERATIVE.

398
00:21:15,479 --> 00:21:17,114
TUMOR RESILIENT T CELLS, THE WAY

399
00:21:17,114 --> 00:21:19,950
TO FIND MARKER IS THROUGH VERY

400
00:21:19,950 --> 00:21:24,521
SIMPLE INTERACTION REGRESSION.

401
00:21:24,521 --> 00:21:29,359
YOU PUT LIKE IMMUNOSUPPRESSION 

402
00:21:29,359 --> 00:21:31,862
AND MOLECULAR MARKER

403
00:21:31,862 --> 00:21:33,464
INTERACTION, COEFFICIENT, THAT

404
00:21:33,464 --> 00:21:35,933
IS SCORE FOR TUMOR RESILIENT T

405
00:21:35,933 --> 00:21:36,600
CELL MARKERS.

406
00:21:36,600 --> 00:21:37,568
THEIR ARE EXAMPLES.

407
00:21:37,568 --> 00:21:40,237
THIS IS A PAPER FROM OUR NCI

408
00:21:40,237 --> 00:21:45,476
COLLEAGUES, TRYING TO FIND SOME

409
00:21:45,476 --> 00:21:49,246
MARKERS OF POSITION T CELLS.

410
00:21:49,246 --> 00:21:53,150
AND HERE ARE GENE SCORE.

411
00:21:53,150 --> 00:21:53,383
X-AXIS.

412
00:21:53,383 --> 00:21:57,454
ON Y-AXIS IS TRES SCORE.

413
00:21:57,454 --> 00:22:01,625
PREDICTION FROM OUR NCI

414
00:22:01,625 --> 00:22:03,727
COLLEAGUES PAPER, FROM MODEL,

415
00:22:03,727 --> 00:22:05,229
PATIENT SAMPLE.

416
00:22:05,229 --> 00:22:09,066
AND THEN WE TEST WHETHER SIGNAL

417
00:22:09,066 --> 00:22:13,437
COULD ACHIEVE SOME PREDICTION

418
00:22:13,437 --> 00:22:16,440
PERFORMANCE, HERE IS OUR

419
00:22:16,440 --> 00:22:17,241
PREDICTION.

420
00:22:17,241 --> 00:22:21,478
FIRST COLLECT SINGLE CELL

421
00:22:21,478 --> 00:22:28,185
RNAseq DATASET FROM 168 

422
00:22:28,185 --> 00:22:33,524
TUMORS, 19 CANCER TYPES.

423
00:22:33,524 --> 00:22:38,262
WE HAVE PROFILE FOR EACH

424
00:22:38,262 --> 00:22:42,466
SIGNATURE, MEDIA SIGNATURE.

425
00:22:42,466 --> 00:22:50,874
AND THEN VALIDATION IS TOTALLY 

426
00:22:50,874 --> 00:22:53,544
INDEPENDENT, T CELL PROFILE,

427
00:22:53,544 --> 00:22:57,447
INHIBITORS, INFUSION FOR CAR

428
00:22:57,447 --> 00:22:58,782
Ts, PRE-MANUFACTURE SAMPLE FOR

429
00:22:58,782 --> 00:23:05,656
CAR T OR TIL TRANSFER THERAPIES.

430
00:23:05,656 --> 00:23:07,758
MANUFACTURE SAMPLES PBMC BEFORE

431
00:23:07,758 --> 00:23:12,963
THE CONSTRUCTION FOR CAR T,

432
00:23:12,963 --> 00:23:15,999
TUMOR SAMPLES BEFORE T CELLS.

433
00:23:15,999 --> 00:23:19,937
AND THEN WE LOOK AT CORRELATION.

434
00:23:19,937 --> 00:23:23,073
IT'S POSITIVE, PREDICT T CELL TO

435
00:23:23,073 --> 00:23:27,177
BE AFFECTED T CELL.

436
00:23:27,177 --> 00:23:29,780
OTHERWISE PREDICT WITH 95,

437
00:23:29,780 --> 00:23:34,451
EXAMPLES OF PREDICTION.

438
00:23:34,451 --> 00:23:35,819
COHORT HERE, Y-AXIS IS

439
00:23:35,819 --> 00:23:38,589
CORRELATION SCORE I SHOW HERE.

440
00:23:38,589 --> 00:23:41,058
THE COLOR IS ACTUAL CLINICAL

441
00:23:41,058 --> 00:23:42,326
OUTCOME, IN EACH CLINICAL STUDY.

442
00:23:42,326 --> 00:23:44,828
ARE YOU CAN SEE OUR PREDICTION

443
00:23:44,828 --> 00:23:48,632
PRETTY WELL WITH THE REAL

444
00:23:48,632 --> 00:23:49,867
CLINICAL OUTCOME LABEL.

445
00:23:49,867 --> 00:23:55,172
AND SO MANY T CELL EFFICACY

446
00:23:55,172 --> 00:23:58,842
SIGNATURE, FOR EXAMPLE SIGNATURE

447
00:23:58,842 --> 00:24:03,847
FROM OUR NCI COLLEAGUES, T CELL

448
00:24:03,847 --> 00:24:05,816
PRODUCE SIGNATURE, PD-1

449
00:24:05,816 --> 00:24:07,117
SIGNATURE, CAR T DYSFUNCTION

450
00:24:07,117 --> 00:24:11,655
SIGNATURE, WE DID PERFORMANCE

451
00:24:11,655 --> 00:24:15,158
COMPARISON AND OVERALL TRES BEST

452
00:24:15,158 --> 00:24:16,226
PREDICTION PERFORMANCE ACROSS

453
00:24:16,226 --> 00:24:17,461
SIGNATURE COMPARED.

454
00:24:17,461 --> 00:24:21,565
AND THEN FURTHER LOOK AT SOME

455
00:24:21,565 --> 00:24:25,502
BULK RNA SEQUENCING DATA FOR

456
00:24:25,502 --> 00:24:27,137
PRE-MANUFACTURE SAMPLES.

457
00:24:27,137 --> 00:24:29,840
IT'S VERY EXPENSIVE, AND IF WE

458
00:24:29,840 --> 00:24:31,942
CAN PREDICT BASED ON

459
00:24:31,942 --> 00:24:32,609
PRE-MANUFACTURED SAMPLES COULD

460
00:24:32,609 --> 00:24:36,780
BE LIKE STOP THE EXPENSIVE

461
00:24:36,780 --> 00:24:43,186
PROCESS FOR OBVIOUS

462
00:24:43,186 --> 00:24:45,022
NON-RESPONDERS.

463
00:24:45,022 --> 00:24:47,557
FOR BOTH, THE EXPRESSION WITH

464
00:24:47,557 --> 00:24:48,525
SIGNATURES IS QUITE PREDICTIVE

465
00:24:48,525 --> 00:24:51,461
OF THE OUTCOME IN THE PATIENT.

466
00:24:51,461 --> 00:24:53,830
PATIENT IS POTENTIALLY FOR CELL

467
00:24:53,830 --> 00:24:55,365
THERAPY, CAN FIND OBVIOUS

468
00:24:55,365 --> 00:24:57,301
NON-RESPONDERS AND STOP THE

469
00:24:57,301 --> 00:25:01,471
EXPENSIVE PROCESS AT THE VERY

470
00:25:01,471 --> 00:25:04,074
BEGINNING.

471
00:25:04,074 --> 00:25:06,944
AND ALSO COMPARISON PREDICTION 

472
00:25:06,944 --> 00:25:09,680
COMPARISON, ONLY TWO SIGNATURES

473
00:25:09,680 --> 00:25:12,149
PREDICTION OF PRE-MANUFACTURED

474
00:25:12,149 --> 00:25:15,953
SAMPLES TRES SIGNATURE ON THE

475
00:25:15,953 --> 00:25:18,588
SIGNATURE FROM OUR COLLEAGUES,

476
00:25:18,588 --> 00:25:21,091
TRES FOR THE BEST.

477
00:25:21,091 --> 00:25:26,063
STEM CELL SIGNATURE ALL OTHER

478
00:25:26,063 --> 00:25:30,033
SIGNATURES DIRECTION PREDICTION.

479
00:25:30,033 --> 00:25:31,635
SO, ENCOURAGED BY RELIABLE

480
00:25:31,635 --> 00:25:32,302
PREDICTION PERFORMANCE ON

481
00:25:32,302 --> 00:25:33,470
CLINICAL OUTCOME FURTHER LOOK AT

482
00:25:33,470 --> 00:25:37,641
OUR SIGNATURE TO SEE WHETHER WE

483
00:25:37,641 --> 00:25:41,445
CAN FIND MOLECULAR REGULATORS OF

484
00:25:41,445 --> 00:25:41,712
APPROACH.

485
00:25:41,712 --> 00:25:45,182
SO, WE FOCUS ON THE MODELS FROM

486
00:25:45,182 --> 00:25:48,819
OUR TR ERICS MODEL.

487
00:25:48,819 --> 00:25:50,821
MOST THERAPIES CURRENTLY BY

488
00:25:50,821 --> 00:25:54,291
SMALL MOLECULE COMPOUND, BY

489
00:25:54,291 --> 00:25:57,327
NEUTRALIZING, MOSTLY INHIBIT A

490
00:25:57,327 --> 00:25:59,029
TARGET.

491
00:25:59,029 --> 00:26:01,365
WE FOCUS ON MARKERS.

492
00:26:01,365 --> 00:26:09,172
THE NUMBER ONE MARKER IS GENE

493
00:26:09,172 --> 00:26:10,040
CALLED FIBP.

494
00:26:10,040 --> 00:26:12,642
THE FULL NAME IS

495
00:26:12,642 --> 00:26:13,176
(INDISCERNIBLE).

496
00:26:13,176 --> 00:26:14,511
VERY FEW LITERATURE ABOUT THIS

497
00:26:14,511 --> 00:26:24,955
GENE EXCEPT THIS GENE INTERACT.

498
00:26:24,955 --> 00:26:26,523
CANCER CELL LINES CAN STOP

499
00:26:26,523 --> 00:26:29,893
CANCER CELL LINE GROWTH HOWEVER

500
00:26:29,893 --> 00:26:31,361
CANNOT REPRODUCE THE RESULT,

501
00:26:31,361 --> 00:26:33,530
PRETTY MUCH UNKNOWN GENE.

502
00:26:33,530 --> 00:26:36,533
NUMBER ONE MARKER IN MODEL.

503
00:26:36,533 --> 00:26:45,308
LET'S DO SOME VALIDATION.

504
00:26:45,308 --> 00:26:47,477
FIRST INITIAL COHORT OF MOUSE,

505
00:26:47,477 --> 00:26:55,218
CANCER CELL LINES, YOU CAN SEE

506
00:26:55,218 --> 00:26:58,055
IN THIS HUMAN, TARGET CANCER

507
00:26:58,055 --> 00:27:02,659
CELL LINES. 

508
00:27:02,659 --> 00:27:05,262
T CELLS WITH KNOCKOUT,

509
00:27:05,262 --> 00:27:08,231
SIGNIFICANT CANCER CELLS, RED

510
00:27:08,231 --> 00:27:11,001
FLUORESCENT MARKERS, COMPARED TO

511
00:27:11,001 --> 00:27:15,072
T CELLS WITH KNOCKOUT.

512
00:27:15,072 --> 00:27:20,377
AND LIKE THE REPEAT EXPERIMENT,

513
00:27:20,377 --> 00:27:24,614
PRODUCE FROM TWO DONORS, TWO

514
00:27:24,614 --> 00:27:27,317
DIFFERENT CANCER CELL LINES.

515
00:27:27,317 --> 00:27:35,258
WE THOUGHT VERY CONVINCING.

516
00:27:35,258 --> 00:27:37,127
COMPARED TO AAVS KNOCKOUT,

517
00:27:37,127 --> 00:27:41,832
OVERALL SIMILAR TO OUR CONTROL,

518
00:27:41,832 --> 00:27:47,437
KNOCKOUT IS A MODIFICATION FOR 

519
00:27:47,437 --> 00:27:53,443
EFFICACY, AND THIS CBLB, FIBP IS

520
00:27:53,443 --> 00:27:55,879
AT THIS LEVEL SIMILAR.

521
00:27:55,879 --> 00:27:58,081
PERFORM IN VIVO MOUSE

522
00:27:58,081 --> 00:28:04,221
EXPERIMENT, ISOLATE THE MOUSE T

523
00:28:04,221 --> 00:28:07,157
CELLS, KNOCK OUT, CONTROL

524
00:28:07,157 --> 00:28:08,992
KNOCKOUT IS VERY IMPORTANT

525
00:28:08,992 --> 00:28:11,161
BECAUSE CRISPR GENE WILL CAUSE

526
00:28:11,161 --> 00:28:14,431
LOT OF ARTIFACT YOU WOULD NOT

527
00:28:14,431 --> 00:28:14,698
EXPECT.

528
00:28:14,698 --> 00:28:19,936
SO COMPARED TO THE CONTROL

529
00:28:19,936 --> 00:28:21,471
KNOCKOUT, REALLY ACHIEVE MUCH

530
00:28:21,471 --> 00:28:24,641
BETTER CONTROLLING OF THE TUMOR

531
00:28:24,641 --> 00:28:26,910
GROWTH IN THE EXPERIMENT.

532
00:28:26,910 --> 00:28:28,478
OVERALL THE TUMOR CONTROL

533
00:28:28,478 --> 00:28:32,149
EFFICACY IS SIMILAR TO OUR PATH

534
00:28:32,149 --> 00:28:36,486
OF CONTROL, GENE KNOCKOUT.

535
00:28:36,486 --> 00:28:39,523
SO, THIS SIGNIFICANT PATH FROM

536
00:28:39,523 --> 00:28:41,892
OUR STUDY WONDERING WHY KNOCKOUT

537
00:28:41,892 --> 00:28:45,295
CAN ENHANCE T CELL EFFICACY ON

538
00:28:45,295 --> 00:28:45,662
TUMORS.

539
00:28:45,662 --> 00:28:52,869
SO PERFORMED SOME RNA SEQUENCING

540
00:28:52,869 --> 00:28:55,005
EXPERIMENT, FIBP KNOCKOUT, SEE

541
00:28:55,005 --> 00:28:58,074
WHAT THE EXPRESSED GENES.

542
00:28:58,074 --> 00:29:02,846
SURPRISINGLY AFTER WE GOT

543
00:29:02,846 --> 00:29:04,981
RNAseq PERFORMED, GENE

544
00:29:04,981 --> 00:29:06,883
ANALYSIS, ALL OF THE PATHWAY IS

545
00:29:06,883 --> 00:29:17,394
VERY SIMPLE, ALL OF THEM ARE 

546
00:29:17,994 --> 00:29:20,864
DOWNREGULATION OF CHOLESTEROL

547
00:29:20,864 --> 00:29:23,800
METABOLISM, LIKE ENZYME FOR

548
00:29:23,800 --> 00:29:25,569
HUMAN CHOLESTEROL METABOLISM

549
00:29:25,569 --> 00:29:29,906
PATHWAYS, ALL DOWNREGULATION

550
00:29:29,906 --> 00:29:35,812
AFTER KNOCKOUT, TO REMOVE 

551
00:29:35,812 --> 00:29:40,217
CHOLESTEROL FROM THE CELL, IL-6

552
00:29:40,217 --> 00:29:43,253
GIVE US SIMPLE CHOLESTEROL

553
00:29:43,253 --> 00:29:45,255
METABOLISM DOWNREGULATION,

554
00:29:45,255 --> 00:29:47,657
FURTHER VALIDATE USING

555
00:29:47,657 --> 00:29:49,659
MULTI-PANEL FLOW ANALYSIS, T

556
00:29:49,659 --> 00:29:50,760
CELLS WITH KNOCKOUT ISOLATE FROM

557
00:29:50,760 --> 00:29:51,828
MOUSE TUMOR.

558
00:29:51,828 --> 00:30:02,372
YOU CAN SEE CHOLESTEROL LEVEL IS

559
00:30:03,073 --> 00:30:03,673
SIGNIFICANTLY LOWER THAN

560
00:30:03,673 --> 00:30:05,175
POSITIVE CONTROL.

561
00:30:05,175 --> 00:30:06,810
IF YOU OVEREXPRESS THE

562
00:30:06,810 --> 00:30:12,616
CHOLESTEROL LEVEL IS GOING TO BE

563
00:30:12,616 --> 00:30:16,786
ELEVATED. 

564
00:30:16,786 --> 00:30:18,121
REVERSE DIRECTION.

565
00:30:18,121 --> 00:30:19,889
SO, KNOWN TO INDUCE T CELL

566
00:30:19,889 --> 00:30:23,793
STRESS AND T CELL EXHAUSTION.

567
00:30:23,793 --> 00:30:25,829
IF YOU EXPOSE T CELL, HIGH

568
00:30:25,829 --> 00:30:27,931
CHOLESTEROL MEDIUM FOR FIVE

569
00:30:27,931 --> 00:30:32,202
DAYS, YOU CAN SEE THAT T CELL ON

570
00:30:32,202 --> 00:30:35,005
TUMOR CELLS IS GOING TO BE

571
00:30:35,005 --> 00:30:36,873
SIGNIFICANTLY WORSE.

572
00:30:36,873 --> 00:30:37,741
HOWEVER, FOR KNOCKOUT T CELLS IF

573
00:30:37,741 --> 00:30:39,376
YOU SOAK THEM IN HIGH

574
00:30:39,376 --> 00:30:40,477
CHOLESTEROL MEDIUM FOR FIVE DAYS

575
00:30:40,477 --> 00:30:41,778
THEY DON'T CARE.

576
00:30:41,778 --> 00:30:44,981
THEY ACHIEVE THE SAME LEVEL OF

577
00:30:44,981 --> 00:30:45,849
CANCER TREATMENT EFFICACY

578
00:30:45,849 --> 00:30:48,485
COMPARED TO T CELL FROM

579
00:30:48,485 --> 00:30:50,086
CHOLESTEROL NORMAL MEDIUM.

580
00:30:50,086 --> 00:30:53,390
WHICH MEANS KNOCKOUT IS MAKING T

581
00:30:53,390 --> 00:30:55,358
CELLS LIKE REALLY RESILIENT TO

582
00:30:55,358 --> 00:30:57,560
HIGH CHOLESTEROL LEVEL, WHICH IS

583
00:30:57,560 --> 00:31:00,196
COMMON FEATURE OF SOLID TUMOR

584
00:31:00,196 --> 00:31:00,964
MICROENVIRONMENT BECAUSE TUMOR

585
00:31:00,964 --> 00:31:05,869
TAKE CHOLESTEROL THEIR FOOD TO

586
00:31:05,869 --> 00:31:06,169
GROW.

587
00:31:06,169 --> 00:31:06,770
PRESENTLY, STARTING TREATMENT

588
00:31:06,770 --> 00:31:08,438
DOES NOT BRING DOWN CHOLESTEROL

589
00:31:08,438 --> 00:31:11,074
LEVELS IN T CELLS.

590
00:31:11,074 --> 00:31:12,876
FOR T CELLS BEFORE, STARTING

591
00:31:12,876 --> 00:31:14,477
TREATMENT, BRING DOWN THE

592
00:31:14,477 --> 00:31:16,880
CHOLESTEROL LEVELS, AS YOU

593
00:31:16,880 --> 00:31:17,881
EXPECT.

594
00:31:17,881 --> 00:31:21,785
BUT REPEAT THIS PATTERN MANY

595
00:31:21,785 --> 00:31:23,153
TIMES.

596
00:31:23,153 --> 00:31:26,389
YOU WILL SEE A SIGNIFICANTLY

597
00:31:26,389 --> 00:31:27,324
ELEVATED LEVEL OF CHOLESTEROL.

598
00:31:27,324 --> 00:31:30,527
WE DON'T KNOW THE REASON WHY.

599
00:31:30,527 --> 00:31:33,463
THIS IS SO CONSISTENT.

600
00:31:33,463 --> 00:31:35,532
MEASUREMENT FOR EFFECTOR T CELL

601
00:31:35,532 --> 00:31:37,767
STATIN IS NOT GOOD

602
00:31:37,767 --> 00:31:38,868
ANTI-CHOLESTEROL THERAPY.

603
00:31:38,868 --> 00:31:45,342
AND THEN WE HAVE CREATED MANY

604
00:31:45,342 --> 00:31:47,177
FUNCTION MODULES, CAN EVALUATE T

605
00:31:47,177 --> 00:31:48,678
CELL MARKERS, BASED ON

606
00:31:48,678 --> 00:31:52,782
PREDICTION ON MARKERS YOU WILL

607
00:31:52,782 --> 00:31:59,322
FIND USING THE PLATFORM, ALSO 

608
00:31:59,322 --> 00:32:01,024
DEMONSTRATE TRES MODEL APPROACH

609
00:32:01,024 --> 00:32:02,759
IN SOLID TUMORS.

610
00:32:02,759 --> 00:32:07,664
AFTER THIS STUDY WE FOUND THAT

611
00:32:07,664 --> 00:32:11,067
THE FOCUS ON MARKERS OF T CELL

612
00:32:11,067 --> 00:32:13,370
TRES MODELS IS WRONG BECAUSE

613
00:32:13,370 --> 00:32:16,673
MANY OF THE MARKERS EVEN TELL US

614
00:32:16,673 --> 00:32:17,841
SOMETHING SURPRISING.

615
00:32:17,841 --> 00:32:21,478
FOR EXAMPLE MY POSTDOC FOCUS ON

616
00:32:21,478 --> 00:32:23,546
MARKERS WHO ARE SECRETED

617
00:32:23,546 --> 00:32:24,447
PROTEINS.

618
00:32:24,447 --> 00:32:25,982
ONCE WE INJECT SECRETED PROTEINS

619
00:32:25,982 --> 00:32:31,855
INTO MOUSE TUMOR, TUMOR

620
00:32:31,855 --> 00:32:32,689
COMPLETELY GOT (INDISCERNIBLE),

621
00:32:32,689 --> 00:32:34,824
DRAMATIC CELL TYPE IN MOUSE

622
00:32:34,824 --> 00:32:37,227
PARTLY BECAUSE SO AGGRESSIVE.

623
00:32:37,227 --> 00:32:40,730
RIGHT NOW WE'RE FOLLOWING UP THE

624
00:32:40,730 --> 00:32:44,968
POSITIVE TRES MARKERS RIGHT NOW.

625
00:32:44,968 --> 00:32:47,604
AND THEN LIKE THE LAST TO

626
00:32:47,604 --> 00:32:50,039
INTRODUCE THE FRAMEWORK FOR

627
00:32:50,039 --> 00:32:58,581
SPECIAL ANALYSIS FOR TUMOR.

628
00:32:58,581 --> 00:33:02,519
SO, HERE IS OVERALL WORK FLOW,

629
00:33:02,519 --> 00:33:04,354
TYPICAL WORKFLOW ANALYZE DATA

630
00:33:04,354 --> 00:33:12,996
FROM THE PLATFORM FOR TUMOR. 

631
00:33:12,996 --> 00:33:17,901
TRANSCRIPTOMIC DATA, THE SIGNAL,

632
00:33:17,901 --> 00:33:19,536
THE FIRST STEP, KNOW THE

633
00:33:19,536 --> 00:33:21,137
DIFFERENT LOCATIONS.

634
00:33:21,137 --> 00:33:24,808
THEN YOU CAN DO ANALYSIS LIKE

635
00:33:24,808 --> 00:33:30,480
CELL-CELL INTERACTION ANALYSIS,

636
00:33:30,480 --> 00:33:32,615
RECEPTOR INTERACTION ANALYSIS,

637
00:33:32,615 --> 00:33:34,050
MANY COMPUTATIONAL PACKAGE DOING

638
00:33:34,050 --> 00:33:36,119
SIMILAR STUFF, DO WE WANT TO

639
00:33:36,119 --> 00:33:36,820
CREATE ANOTHER COMPUTATIONAL

640
00:33:36,820 --> 00:33:41,257
PACKAGE, HERE IS REASON.

641
00:33:41,257 --> 00:33:43,793
BECAUSE CURRENT ARE NOT --

642
00:33:43,793 --> 00:33:49,299
(INDISCERNIBLE) FOR SOME UNIQUE

643
00:33:49,299 --> 00:33:52,335
CHALLENGE.

644
00:33:52,335 --> 00:33:55,972
FIRST OF ALL, SINGLE CELL

645
00:33:55,972 --> 00:33:58,608
(INDISCERNIBLE) TYPICAL WAY USED

646
00:33:58,608 --> 00:34:05,148
TO FIND CANCER CELLS IS THROUGH

647
00:34:05,148 --> 00:34:06,149
OF (INDISCERNIBLE).

648
00:34:06,149 --> 00:34:08,051
THERE'S NO MOLECULAR MARKERS OF

649
00:34:08,051 --> 00:34:12,355
CANCER CELLS TYPICALLY LOOK AT 

650
00:34:12,355 --> 00:34:17,694
SEGMENTAL OPERATION PATTERNS IN

651
00:34:17,694 --> 00:34:19,362
SINGLE CELL TRANSCRIPTOMETRY

652
00:34:19,362 --> 00:34:19,562
FIRST.

653
00:34:19,562 --> 00:34:29,839
THIS IS WRONG. 

654
00:34:36,479 --> 00:34:37,380
HERE IS EXAMPLE.

655
00:34:37,380 --> 00:34:41,150
IF YOU PUT LOTS OF MICRO PHAGE

656
00:34:41,150 --> 00:34:44,120
CELL AT THE NORMAL REFERENCE,

657
00:34:44,120 --> 00:34:47,190
YOU'RE GOING TO GET DELETION THE

658
00:34:47,190 --> 00:34:49,759
CHROMOSOME 6 ARM, PROFILE AT

659
00:34:49,759 --> 00:34:50,293
CANCER CELL.

660
00:34:50,293 --> 00:34:52,762
OTHERWISE, IF YOU HAVE ENOUGH

661
00:34:52,762 --> 00:35:01,804
NORMAL CELLS, YOU'RE GOING TO

662
00:35:01,804 --> 00:35:05,108
SEE AMPLIFICATION, POTENTIAL

663
00:35:05,108 --> 00:35:06,409
CANCER CELLS.

664
00:35:06,409 --> 00:35:10,780
SECOND, MANY CANCER CELLS JUST

665
00:35:10,780 --> 00:35:14,183
DON'T HAVE ALTERATIONS BECAUSE

666
00:35:14,183 --> 00:35:19,155
IN THIS STUDY CANCER TUMORS WITH

667
00:35:19,155 --> 00:35:24,527
HIGH MUTATION LEVELS, RELATIVELY

668
00:35:24,527 --> 00:35:28,531
LOW COPY ALTERATIONS, NOT COPY

669
00:35:28,531 --> 00:35:32,802
ENOUGH ALTERATIONS.

670
00:35:32,802 --> 00:35:35,972
UNFORTUNATELY, WE HAVE STUDIED

671
00:35:35,972 --> 00:35:46,516
THIS PROBLEM, IN 2013 USED NIP 

672
00:35:46,983 --> 00:35:49,285
-- SNP ARRAY, PATTERNS ARE

673
00:35:49,285 --> 00:35:49,852
DISTANT.

674
00:35:49,852 --> 00:35:51,554
BREAST CANCER, SKIN CANCER,

675
00:35:51,554 --> 00:35:56,159
COLORECTAL, THEY ARE ALL ROUGHLY

676
00:35:56,159 --> 00:36:04,033
THE SAME PATTERNS.

677
00:36:04,033 --> 00:36:06,302
SO, WE FIRST TALKING ABOUT RATE

678
00:36:06,302 --> 00:36:09,405
FROM COHORT FOR DIFFERENT SOLID

679
00:36:09,405 --> 00:36:09,706
TUMOR TYPES.

680
00:36:09,706 --> 00:36:16,079
IF YOU SEE A SPOT SIMILAR,

681
00:36:16,079 --> 00:36:17,847
PROFILE AS CANCER.

682
00:36:17,847 --> 00:36:19,582
OTHERWISE WE GO TO SECOND TEST

683
00:36:19,582 --> 00:36:22,585
TO CHECK IF THE EXPRESSION

684
00:36:22,585 --> 00:36:28,992
PROFILE IS SIMILAR TO SOME TYPE

685
00:36:28,992 --> 00:36:30,960
OF TUMOR NORMAL.

686
00:36:30,960 --> 00:36:34,631
SIMILAR PROFILE HAVE CANCER

687
00:36:34,631 --> 00:36:34,831
SPOT.

688
00:36:34,831 --> 00:36:38,134
OTHERWISE WE CLASSIFY AS

689
00:36:38,134 --> 00:36:39,402
NON-CANCER DECOMPOSITION.

690
00:36:39,402 --> 00:36:47,577
AND THEN THE SECOND TYPE OF REAL

691
00:36:47,577 --> 00:36:50,613
CHALLENGE, IN THIS EXAMPLE YOU

692
00:36:50,613 --> 00:36:53,516
HAVE SOME REGIONS WITH A LOT OF

693
00:36:53,516 --> 00:36:53,883
CELLS.

694
00:36:53,883 --> 00:36:55,118
SOME STROMAL REGIONS WITH VERY

695
00:36:55,118 --> 00:36:56,019
FEW CELLS.

696
00:36:56,019 --> 00:37:01,024
IF YOU DO A CELL DECOMPOSITION

697
00:37:01,024 --> 00:37:02,759
ON MODEL, CELL FRACTION, ONE,

698
00:37:02,759 --> 00:37:03,626
FUNDAMENTALLY WRONG.

699
00:37:03,626 --> 00:37:05,128
CELL FRACTION IN HIGH DENSITY

700
00:37:05,128 --> 00:37:07,330
AND LOW DENSITY REGIONS ARE NOT

701
00:37:07,330 --> 00:37:10,299
COMPARATIVE WITH EACH OTHER.

702
00:37:10,299 --> 00:37:11,701
FORTUNATELY, WE FIND A SOLUTION,

703
00:37:11,701 --> 00:37:13,670
REGRESSION MODEL WE CAN SIMPLY

704
00:37:13,670 --> 00:37:18,574
PUT UNKNOWN COMPONENT IN THE

705
00:37:18,574 --> 00:37:22,078
REGRESSION MODEL, UNLESS

706
00:37:22,078 --> 00:37:24,447
REGRESSION PANELLIZE OVERALL

707
00:37:24,447 --> 00:37:26,315
TYPE, LOTS OF EXPRESSION INDUCED

708
00:37:26,315 --> 00:37:31,421
BY LOW DENSITY AND THEN THIS

709
00:37:31,421 --> 00:37:34,090
AUTOMATIC ADJUST TO OVERALL CELL

710
00:37:34,090 --> 00:37:38,261
FRACTION FOR LOW DENSITY TISSUE

711
00:37:38,261 --> 00:37:38,594
REGION.

712
00:37:38,594 --> 00:37:44,701
ALSO PERFORM CONVOLUTION TO MAKE

713
00:37:44,701 --> 00:37:49,238
SURE THE CO-LINEARITY WILL NOT

714
00:37:49,238 --> 00:37:50,106
AFFECT YOUR LIABILITY, TWO

715
00:37:50,106 --> 00:37:53,843
SIGNAL WITH EACH OTHER.

716
00:37:53,843 --> 00:38:02,185
WITH ALL OF THESE TRICKS,

717
00:38:02,185 --> 00:38:05,955
PERFORM (INDISCERNIBLE) OUR

718
00:38:05,955 --> 00:38:07,590
PATHOLOGY COLLABORATIVE,

719
00:38:07,590 --> 00:38:12,195
PATHOLOGY OF A PATIENT, TUMOR

720
00:38:12,195 --> 00:38:16,966
REGIONS, STROMAL REGIONS, 

721
00:38:16,966 --> 00:38:19,302
LYMPHOCYTE AND MACROPHAGE

722
00:38:19,302 --> 00:38:21,938
REGIONS. 

723
00:38:21,938 --> 00:38:25,308
FOUR REGIONS, ONE TRUTH, INDEED

724
00:38:25,308 --> 00:38:28,311
OUR STATED FRAMEWORK ACHIEVES

725
00:38:28,311 --> 00:38:30,179
BEST PREDICTION PERFORMANCE

726
00:38:30,179 --> 00:38:31,314
AMONG ALL METHOD COMPARED.

727
00:38:31,314 --> 00:38:34,751
LAST YOU CAN ALSO DO SOME

728
00:38:34,751 --> 00:38:35,351
CELL-CELL INTERACTION ANALYSIS

729
00:38:35,351 --> 00:38:40,123
TO SEE WHERE DOES THE CELL

730
00:38:40,123 --> 00:38:40,723
INTERACTION LIGAND RECEPTION

731
00:38:40,723 --> 00:38:43,059
FACTOR HAPPEN IN THE PROFILE.

732
00:38:43,059 --> 00:38:45,361
HERE IS EXAMPLE.

733
00:38:45,361 --> 00:38:51,768
IN THIS YOU CAN SEE THE CANCER

734
00:38:51,768 --> 00:38:54,804
CYTO BLASTS, TENDS TO REALLY

735
00:38:54,804 --> 00:38:56,205
CO-LOCALIZE WITH EACH OTHER,

736
00:38:56,205 --> 00:38:57,740
RECEPTION REGION.

737
00:38:57,740 --> 00:38:59,041
OVER HERE, REALLY MIXED WITH

738
00:38:59,041 --> 00:38:59,709
EACH OTHER.

739
00:38:59,709 --> 00:39:01,911
IN THE REGION THAT THEY INTERMIX

740
00:39:01,911 --> 00:39:05,181
WITH EACH OTHER, YOU SEE LOTS OF

741
00:39:05,181 --> 00:39:09,051
RECEPTOR PAIRS CO-EXPRESSED OVER

742
00:39:09,051 --> 00:39:10,219
THERE.

743
00:39:10,219 --> 00:39:11,954
AND THEN LIKE REVERSELY IF YOU

744
00:39:11,954 --> 00:39:16,592
LOOK AT THE TUMOR CELLS, THAT

745
00:39:16,592 --> 00:39:20,263
ARE CLOSE, REALLY FAR AWAY FROM

746
00:39:20,263 --> 00:39:21,631
THE CUFF ON MACROPHAGE

747
00:39:21,631 --> 00:39:22,965
INTERACTION LOCATIONS YOU CAN

748
00:39:22,965 --> 00:39:25,568
SEE WE HAVE A PRETTY DIFFERENT

749
00:39:25,568 --> 00:39:33,009
EXPRESSION PROFILE. 

750
00:39:33,009 --> 00:39:35,044
FOR TUMOR CELLS, CLOSE

751
00:39:35,044 --> 00:39:35,978
THEY GO SLOWER.

752
00:39:35,978 --> 00:39:39,048
HOWEVER, IF THEY HAVE HIGHER

753
00:39:39,048 --> 00:39:42,084
EXPRESSION LEVEL FOR THE

754
00:39:42,084 --> 00:39:43,886
MESENCHYMAL TRANSITION GENE

755
00:39:43,886 --> 00:39:45,254
SITE, WHICH MEANS TUMOR CELLS

756
00:39:45,254 --> 00:39:52,395
THAT  ARE FAR AWAY FROM THE

757
00:39:52,395 --> 00:39:56,399
STROMAL CELL REACTION ARE FASTER

758
00:39:56,399 --> 00:39:57,433
BUT LESS AGGRESSIVE.

759
00:39:57,433 --> 00:39:59,068
TUMOR CELL MUCH MORE AGGRESSIVE

760
00:39:59,068 --> 00:40:02,972
EVEN THROW THEY GO SLOWER.

761
00:40:02,972 --> 00:40:06,642
THIS IS EXAMPLE OF SPATIAL

762
00:40:06,642 --> 00:40:07,977
TRANSCRIPTOMICS WITH FRAMEWORK

763
00:40:07,977 --> 00:40:09,812
TIES BEHAVIOR DIFFERENCE OF

764
00:40:09,812 --> 00:40:13,182
TUMOR CELLS IN SPATIAL

765
00:40:13,182 --> 00:40:14,784
GEOGRAPHIC SITE.

766
00:40:14,784 --> 00:40:17,119
AND HERE IS OUR -- OUR PACKAGE

767
00:40:17,119 --> 00:40:20,122
HAS INTERACTIVE FUNCTION MODELS

768
00:40:20,122 --> 00:40:22,291
FOR YOU TO INTERACTIVELY FOLLOW

769
00:40:22,291 --> 00:40:25,027
DATASET, OUTCOME TO CELL

770
00:40:25,027 --> 00:40:25,261
PACKAGE.

771
00:40:25,261 --> 00:40:31,100
AND FINALLY THIS IS THE OVERALL

772
00:40:31,100 --> 00:40:35,504
FRAMEWORK HAVE DEVELOPED LIKE IN

773
00:40:35,504 --> 00:40:42,178
PAST FOUR YEARS TO STUDY THE

774
00:40:42,178 --> 00:40:43,112
INTRACELLULAR SIGNALINGS

775
00:40:43,112 --> 00:40:45,114
HOPEFULLY LIKE RESEPTEMBER

776
00:40:45,114 --> 00:40:46,115
GALORE YOUR PRODUCT.

777
00:40:46,115 --> 00:40:49,919
LAST THE SPECIAL THANKS TO MY

778
00:40:49,919 --> 00:40:52,154
BRILLIANT TEAM MEMBERS, I THINK

779
00:40:52,154 --> 00:40:58,361
I'M LIKELY TO GET SOME STELLAR

780
00:40:58,361 --> 00:41:00,830
(INDISCERNIBLE) OF MY CAREER,

781
00:41:00,830 --> 00:41:04,600
ESPECIALLY FIRST START AT NCI

782
00:41:04,600 --> 00:41:08,738
COVID PANDEMIC BREAK OUT, AND

783
00:41:08,738 --> 00:41:09,338
THEY WERE REMARKABLY RESILIENT

784
00:41:09,338 --> 00:41:11,908
AT THAT TIME.

785
00:41:11,908 --> 00:41:14,877
A SHOUT OUT, STILL MANAGE THE

786
00:41:14,877 --> 00:41:15,678
PRODUCT MOVING FORWARD DURING

787
00:41:15,678 --> 00:41:16,746
HARD TIME.

788
00:41:16,746 --> 00:41:25,354
SPECIAL THANKS TO MY NCI

789
00:41:25,354 --> 00:41:26,255
COLLEAGUES AND MENTORS.

790
00:41:26,255 --> 00:41:30,593
ALL OF THEM REALLY GIVING ME

791
00:41:30,593 --> 00:41:31,327
REMARKABLE HELP AND MENTORSHIP

792
00:41:31,327 --> 00:41:32,495
AS A YOUNG FACULTY.

793
00:41:32,495 --> 00:41:41,437
OTHERWISE I WOULD MAKE SO MANY

794
00:41:41,437 --> 00:41:41,737
MISTAKES.

795
00:41:41,737 --> 00:41:44,140
REALLY HELPED PRODUCT MOVING

796
00:41:44,140 --> 00:41:49,979
FORWARD DURING HARD TIME. 

797
00:41:49,979 --> 00:42:00,523
THANKS TO MY NCI COLLEAGUES AND

798
00:42:04,560 --> 00:42:04,727
MENTORS.

799
00:42:04,727 --> 00:42:04,994
[APPLAUSE]

800
00:42:04,994 --> 00:42:05,227
>> OKAY.

801
00:42:05,227 --> 00:42:07,063
THANK YOU SO MUCH.

802
00:42:07,063 --> 00:42:10,266
THIS IS SUCH A TOUR DU FORCE.

803
00:42:10,266 --> 00:42:12,969
I THINK YOU MANAGED TO PRESENT

804
00:42:12,969 --> 00:42:16,238
70 SLIDES IN 35 MINUTES.

805
00:42:16,238 --> 00:42:23,346
I THINK IT'S A NEW NCI RECORD. 

806
00:42:23,346 --> 00:42:27,883
MY HEAD STILL SPINS.

807
00:42:27,883 --> 00:42:28,851
IT'S TIME FOR DISCUSSION.

808
00:42:28,851 --> 00:42:29,852
>> YES, VERY INTERESTING.

809
00:42:29,852 --> 00:42:34,323
A QUESTION ABOUT THE TRES

810
00:42:34,323 --> 00:42:38,060
PIPELINE, HOW IT CAN BE EXTENDED

811
00:42:38,060 --> 00:42:40,229
TO CAPTURE LIKE SYNERGISTIC

812
00:42:40,229 --> 00:42:43,132
EFFECT BETWEEN MULTIPLE GENES.

813
00:42:43,132 --> 00:42:47,036
>> OH, SO LET ME
BACKTRACK,

814
00:42:47,036 --> 00:42:47,236
TRES?

815
00:42:47,236 --> 00:42:47,470
>> YES.

816
00:42:47,470 --> 00:42:49,939
>> YOU ARE ASKING WHETHER
TRES

817
00:42:49,939 --> 00:42:58,147
FRAMEWORK COULD BE REPURPOSED TO

818
00:42:58,147 --> 00:43:00,282
STUDY SYNERGIES BETWEEN

819
00:43:00,282 --> 00:43:01,017
PROTEINS.

820
00:43:01,017 --> 00:43:02,184
>> BETWEEN THE GENES
EXPRESSION

821
00:43:02,184 --> 00:43:02,385
RATIO.

822
00:43:02,385 --> 00:43:04,020
>> OH, I SEE.

823
00:43:04,020 --> 00:43:04,787
I SEE.

824
00:43:04,787 --> 00:43:08,791
YES, LIKE -- I SEE.

825
00:43:08,791 --> 00:43:13,429
I THINK OF COURSE, LIKE YOU

826
00:43:13,429 --> 00:43:15,731
CAN -- WITHOUT TRES BASICALLY

827
00:43:15,731 --> 00:43:16,499
INPUT, VECTORS CAUSE DIFFERENT

828
00:43:16,499 --> 00:43:18,434
SINGLE CELLS.

829
00:43:18,434 --> 00:43:22,038
YOU CAN SIMPLY CREATE DIFFERENT

830
00:43:22,038 --> 00:43:24,473
GENES AND INTO A DATA MATRIX,

831
00:43:24,473 --> 00:43:30,212
AND SEE WHETHER OR NOT YOU SEE

832
00:43:30,212 --> 00:43:31,213
SIGNIFICANT (INDISCERNIBLE).

833
00:43:31,213 --> 00:43:34,050
VERY GOOD POINT TO REPURPOSE THE

834
00:43:34,050 --> 00:43:34,316
FRAMEWORK.

835
00:43:34,316 --> 00:43:38,454
WHICH IS PROBABLY VERY RELEVANT

836
00:43:38,454 --> 00:43:41,557
TO YOUR CURRENT PROJECT.

837
00:43:41,557 --> 00:43:46,195
>> THANK YOU.

838
00:43:46,195 --> 00:43:47,063
 

839
00:43:47,063 --> 00:43:49,031
>> OTHERS QUESTIONS?

840
00:43:49,031 --> 00:43:50,499
SURELY YOU HAVE THOUGHTS TO

841
00:43:50,499 --> 00:43:50,933
SHARE.

842
00:43:50,933 --> 00:43:54,270
I SEE IT ON YOU.

843
00:43:54,270 --> 00:44:00,309
 

844
00:44:00,309 --> 00:44:01,844
 

845
00:44:01,844 --> 00:44:04,513
>> WE HAVE ALL THESE
PAPERS

846
00:44:04,513 --> 00:44:06,382
IDENTIFYING SIGNATURES, FROM ONE

847
00:44:06,382 --> 00:44:07,550
PAPER TO ANOTHER DON'T

848
00:44:07,550 --> 00:44:08,084
NECESSARILY OVERLAP.

849
00:44:08,084 --> 00:44:10,419
WHAT DO YOU THINK ABOUT THAT?

850
00:44:10,419 --> 00:44:14,690
WE'RE KIND OF LOOKING AT SAME

851
00:44:14,690 --> 00:44:16,859
DATASETS WITH VERY DIFFERENT

852
00:44:16,859 --> 00:44:18,127
SIGNATURE AT THE END.

853
00:44:18,127 --> 00:44:21,497
>> YOU RAISE A VERY
IMPORTANT

854
00:44:21,497 --> 00:44:25,835
POINT, THERE ARE SO MANY

855
00:44:25,835 --> 00:44:26,168
SIGNATURES.

856
00:44:26,168 --> 00:44:28,070
THEY OVERLAP WITH EACH OTHER.

857
00:44:28,070 --> 00:44:30,940
FIRST OF ALL, I CAN ONLY COMMENT

858
00:44:30,940 --> 00:44:32,475
ABOUT TRES SIGNATURES.

859
00:44:32,475 --> 00:44:37,279
THE TRES SIGNATURE HAS QUITE

860
00:44:37,279 --> 00:44:44,353
SOME OVERLAP WITH THE CENTRAL

861
00:44:44,353 --> 00:44:49,592
(INDISCERNIBLE) SIGNATURE. 

862
00:44:49,592 --> 00:44:51,293
MEANS QUIESCENT T CELLS, LIKE

863
00:44:51,293 --> 00:44:57,867
BOTH MARKERS ARE ALL VERY HIGH,

864
00:44:57,867 --> 00:44:58,968
TRES SIGNATURE.

865
00:44:58,968 --> 00:45:02,338
MEANWHILE TRES HAVE A FEW

866
00:45:02,338 --> 00:45:04,039
MARKERS, SOME SECRETION FACTORS

867
00:45:04,039 --> 00:45:09,345
NOT PROVEN IN THE SIGNATURES,

868
00:45:09,345 --> 00:45:10,846
AND (INDISCERNIBLE) ADDRESS THAT

869
00:45:10,846 --> 00:45:13,749
TO PROVE THAT.

870
00:45:13,749 --> 00:45:16,685
I GUESS LIKE -- AGAIN TO MAKE T

871
00:45:16,685 --> 00:45:19,955
CELL WORK BETTER, OR WORK WORSE.

872
00:45:19,955 --> 00:45:21,290
LIKE EACH SIGNATURES PROBABLY

873
00:45:21,290 --> 00:45:25,594
CAPTURE ONE PART OF THAT.

874
00:45:25,594 --> 00:45:27,930
AND TUMOR ARE SO HETEROGENEOUS.

875
00:45:27,930 --> 00:45:30,099
I GUESS DIFFERENT PATIENTS AND

876
00:45:30,099 --> 00:45:31,800
DIFFERENT TUMORS UTILIZE VERY

877
00:45:31,800 --> 00:45:33,402
DIFFERENT STRATEGIES TO MAKE T

878
00:45:33,402 --> 00:45:33,602
CELLS.

879
00:45:33,602 --> 00:45:36,305
AND YOU PROBABLY NEED VERY

880
00:45:36,305 --> 00:45:38,474
DIFFERENT STRATEGIES TO UTILIZE

881
00:45:38,474 --> 00:45:40,709
T CELLS ACCORDING TO THE

882
00:45:40,709 --> 00:45:42,111
SPECIFICITY, THE PROGRAM OF

883
00:45:42,111 --> 00:45:46,215
DIFFERENT T CELLS. 

884
00:45:46,215 --> 00:45:49,518
SIMPLY REFLECTION OF MOLECULAR

885
00:45:49,518 --> 00:45:53,422
MECHANISM THAT COULD HAVE LIKE

886
00:45:53,422 --> 00:45:58,194
OTHERWISE I DON'T THINK FROM ONE

887
00:45:58,194 --> 00:45:59,728
OR TWO UNIVERSAL FRAMEWORK FOR

888
00:45:59,728 --> 00:45:59,929
THIS.

889
00:45:59,929 --> 00:46:01,664
>> I GUESS WE DON'T DEAL
SO WELL

890
00:46:01,664 --> 00:46:03,232
WITH THE TUMOR SIDE, RIGHT?

891
00:46:03,232 --> 00:46:06,802
IF WE HAD MORE CONTEXT, MAYBE

892
00:46:06,802 --> 00:46:08,571
DIFFERENT SIGNATURE WOULD MAKE

893
00:46:08,571 --> 00:46:11,073
MORE SENSE, RIGHT?

894
00:46:11,073 --> 00:46:17,313
>> YEAH, IDEALLY, TUMOR
SPECIFIC

895
00:46:17,313 --> 00:46:17,880
CONTEXT, YEAH.

896
00:46:17,880 --> 00:46:18,948
>> HI.

897
00:46:18,948 --> 00:46:20,149
VERY NICE WORK.

898
00:46:20,149 --> 00:46:21,016
THANKS SO MUCH.

899
00:46:21,016 --> 00:46:25,754
A WHILE AGO A LOT OF PEOPLE

900
00:46:25,754 --> 00:46:27,656
TRIED TO ANALYZE ASSOCIATION

901
00:46:27,656 --> 00:46:32,828
THAT FOLLOWED JUST HISTOLOGY

902
00:46:32,828 --> 00:46:38,100
ANALYSIS JUST LIKE WITH GENE

903
00:46:38,100 --> 00:46:38,434
EXPRESSION.

904
00:46:38,434 --> 00:46:41,437
COULD YOU COMMENT ON THAT?

905
00:46:41,437 --> 00:46:46,108
>> OH, OKAY.

906
00:46:46,108 --> 00:46:46,709
TRICKY.

907
00:46:46,709 --> 00:46:51,547
HIV GENE EXPRESSION FIRST OF ALL

908
00:46:51,547 --> 00:46:55,784
LIKE DEVELOPMENT FRAMEWORK BY

909
00:46:55,784 --> 00:46:56,986
PREDICTION, LIKE MY

910
00:46:56,986 --> 00:46:58,354
UNDERSTANDING CAN ONLY PREDICT

911
00:46:58,354 --> 00:47:02,992
CERTAIN GENES WITH GOOD

912
00:47:02,992 --> 00:47:03,292
PERFORMANCE.

913
00:47:03,292 --> 00:47:04,893
EYTAN, DO YOU HAVE SOMETHING TO

914
00:47:04,893 --> 00:47:05,728
ADD UP HERE?

915
00:47:05,728 --> 00:47:09,365
>> YEAH, SO WE NOW BELIEVE
THAT

916
00:47:09,365 --> 00:47:10,899
WITH THE RECENT ADVANCES IN DEEP

917
00:47:10,899 --> 00:47:14,370
LEARNING A.I., ALL THESE

918
00:47:14,370 --> 00:47:18,340
BUZZWORDS, YOU CAN PREDICT QUITE

919
00:47:18,340 --> 00:47:18,807
WELL.

920
00:47:18,807 --> 00:47:19,575
IT'S REALLY SURPRISING

921
00:47:19,575 --> 00:47:22,044
EXPRESSION OF GENES AND

922
00:47:22,044 --> 00:47:24,079
METHYLATION OF GENES DIRECTLY.

923
00:47:24,079 --> 00:47:26,582
AND PEOPLE HAVE SHOWN IT BEFORE

924
00:47:26,582 --> 00:47:30,386
US, AND WE ALSO STUDY THAT.

925
00:47:30,386 --> 00:47:37,126
ACTUALLY, OUR NEXT BIG CHALLENGE

926
00:47:37,126 --> 00:47:38,727
IS TO PREDICT SPATIAL

927
00:47:38,727 --> 00:47:41,363
TRANSCRIPTOMICS, REMAINS TO BE

928
00:47:41,363 --> 00:47:41,563
SEEN.

929
00:47:41,563 --> 00:47:43,332
THERE HAVE BEEN, FOR EXAMPLE, ZU

930
00:47:43,332 --> 00:47:44,900
PUBLISHED A PAPER A FEW YEARS

931
00:47:44,900 --> 00:47:48,904
AGO, AND OTHERS, AND SO ON, FOR

932
00:47:48,904 --> 00:47:49,672
A FEW HUNDRED GENES.

933
00:47:49,672 --> 00:47:53,442
WE BELIEVE WE CAN DO IT FOR 

934
00:47:53,442 --> 00:47:54,810
THOUSANDS OF GENES.

935
00:47:54,810 --> 00:47:57,313
SO IS ANYONE IS INTERESTED, WE

936
00:47:57,313 --> 00:47:59,315
HAVE A bioRxiv PAPER NOW

937
00:47:59,315 --> 00:48:03,752
UNDER REVIEW WHETHER WE SHOW THE

938
00:48:03,752 --> 00:48:06,588
PREDICTIVE ABILITY OF 

939
00:48:06,588 --> 00:48:07,723
TRANSCRIPTOMICS FROM THE SLIDES.

940
00:48:07,723 --> 00:48:08,657
VERY INTERESTING.

941
00:48:08,657 --> 00:48:11,593
I WOULD LIKE TO ASK THE

942
00:48:11,593 --> 00:48:12,928
AUDIENCE, WHAT PATHWAYS DO YOU

943
00:48:12,928 --> 00:48:15,431
THINK ARE THE MOST PREDICTABLE,

944
00:48:15,431 --> 00:48:18,734
THE EXPRESSION OF WHICH PATHWAYS

945
00:48:18,734 --> 00:48:20,336
ARE OF MOST PREDICTABLE FROM THE

946
00:48:20,336 --> 00:48:28,711
HME SLIDES AT LEAST IN OUR

947
00:48:28,711 --> 00:48:28,911
HANDS?

948
00:48:28,911 --> 00:48:29,111
YEAH?

949
00:48:29,111 --> 00:48:29,311
WHAT?

950
00:48:29,311 --> 00:48:29,545
>> EMT.

951
00:48:29,545 --> 00:48:30,879
>> EMT.

952
00:48:30,879 --> 00:48:34,450
WELL, NOT ACCORDING TO OUR

953
00:48:34,450 --> 00:48:34,717
ANALYSIS.

954
00:48:34,717 --> 00:48:37,019
WE FIND CELL PROLIFERATION AND

955
00:48:37,019 --> 00:48:38,987
SOME IMMUNE INTERACTION PATHWAYS

956
00:48:38,987 --> 00:48:42,591
ARE PREDICTED ACROSS ALMOST ALL

957
00:48:42,591 --> 00:48:49,531
CELL TYPES IN TCGA FROM HNE

958
00:48:49,531 --> 00:48:49,832
IMAGES.

959
00:48:49,832 --> 00:48:51,333
PREDICTIVE ABILITY IS FAR FROM

960
00:48:51,333 --> 00:48:53,435
PERFECT, AND SO ON.

961
00:48:53,435 --> 00:49:01,944
BUT I HAVE TO SAY THE TOOLS 

962
00:49:01,944 --> 00:49:02,444
UNDERLYING ARCHITECTURES

963
00:49:02,444 --> 00:49:04,980
EVOLVING SO FAST.

964
00:49:04,980 --> 00:49:06,782
I'M SURE PEOPLE CAN DO BETTER

965
00:49:06,782 --> 00:49:08,283
THAN WE DID.

966
00:49:08,283 --> 00:49:09,485
WE'RE ALSO TRYING TO DO BETTER

967
00:49:09,485 --> 00:49:11,920
THAN WHAT CAN'T BE PUBLISHED.

968
00:49:11,920 --> 00:49:13,722
I'M STRUCK BY SO MUCH

969
00:49:13,722 --> 00:49:14,022
INFORMATION.

970
00:49:14,022 --> 00:49:17,593
AS YOU KNOW, A LOT OF YOU KNOW

971
00:49:17,593 --> 00:49:20,295
THERE'S MULTISPECTRAL IMAGING

972
00:49:20,295 --> 00:49:23,298
COMING UP, BIG PHARMA IS BIG

973
00:49:23,298 --> 00:49:25,334
INTO MULTISPECTRAL IMAGING, AND

974
00:49:25,334 --> 00:49:28,637
I JUST IMAGINE WITH DEEP

975
00:49:28,637 --> 00:49:31,707
LEARNING, MULTISPECTRAL  IMAGING

976
00:49:31,707 --> 00:49:33,142
YOU'LL BE ABLE TO DO WAY BETTER

977
00:49:33,142 --> 00:49:38,180
THAN WHAT WE DO FROM THE H AND 

978
00:49:38,180 --> 00:49:48,323
THE E. 

979
00:49:49,458 --> 00:49:50,058
RIGHT?

980
00:49:50,058 --> 00:49:51,260
>> YOU SHOWED KIND OF THE
BENCH

981
00:49:51,260 --> 00:49:55,097
TO BEDSIDE AND BACK, STARTING 

982
00:49:55,097 --> 00:49:57,499
WITH THE HUMAN DATASETS AND THEN

983
00:49:57,499 --> 00:49:58,934
T CELL RESILIENT STUDIES LOOKING

984
00:49:58,934 --> 00:50:01,036
AT SPECIFICALLY AT MOUSE MODELS,

985
00:50:01,036 --> 00:50:03,071
BACK TO HUMAN STUDIES.

986
00:50:03,071 --> 00:50:03,839
WONDERING FOR THOSE WHO HAVE

987
00:50:03,839 --> 00:50:07,443
MOUSE MODELS THAT WE HAVE TO USE

988
00:50:07,443 --> 00:50:08,544
BECAUSE OF THE DIFFICULTY AND

989
00:50:08,544 --> 00:50:11,180
RARITY OF SEEING SOME OF THE

990
00:50:11,180 --> 00:50:11,847
THINGS WE'RE STUDYING

991
00:50:11,847 --> 00:50:12,948
CLINICALLY, TO WHAT EXTENT ARE

992
00:50:12,948 --> 00:50:17,419
THE TOOLS THAT YOU DEVELOPED

993
00:50:17,419 --> 00:50:19,822
APPLICABLE YET POOR MOUSE

994
00:50:19,822 --> 00:50:21,223
STUDIES WHERE PERHAPS MOUSE DATA

995
00:50:21,223 --> 00:50:23,559
JUST DOESN'T EXIST IN THIS

996
00:50:23,559 --> 00:50:23,792
CONTEXT?

997
00:50:23,792 --> 00:50:26,161
>> EXCUSE ME, I WANT TO
MAKE

998
00:50:26,161 --> 00:50:28,764
SURE I CAPTURE THE QUESTION.

999
00:50:28,764 --> 00:50:31,567
ASKING FOR THE TOOLS AND MOUSE

1000
00:50:31,567 --> 00:50:33,435
MODEL RESULT, WHAT ARE THE

1001
00:50:33,435 --> 00:50:35,404
FRACTION COULD BE SOMEHOW

1002
00:50:35,404 --> 00:50:36,371
REFLECTING CLINICAL SETTINGS, IS

1003
00:50:36,371 --> 00:50:37,406
YOUR QUESTION?

1004
00:50:37,406 --> 00:50:39,975
DO YOU MIND --

1005
00:50:39,975 --> 00:50:44,847
>> FOR EXAMPLE, TRES, AND
YOUR

1006
00:50:44,847 --> 00:50:45,814
SPATIAL TRANSCRIPTOMICS

1007
00:50:45,814 --> 00:50:49,017
APPROACHES, ALL USE BIG DATA. 

1008
00:50:49,017 --> 00:50:50,219
PRIMARILY FROM HUMAN DATA

1009
00:50:50,219 --> 00:50:50,452
SOURCES.

1010
00:50:50,452 --> 00:50:52,087
IN THE CASES WHERE YOU DID LOOK

1011
00:50:52,087 --> 00:50:56,191
BACK AT THE MOUSE MODELS, THOSE

1012
00:50:56,191 --> 00:50:57,659
WERE FOCUSED MOUSE EXPERIMENTS,

1013
00:50:57,659 --> 00:50:59,161
WONDERING ARE WE THERE YET WITH

1014
00:50:59,161 --> 00:51:02,931
MOUSE DATA TO BE ABLE TO APPLY

1015
00:51:02,931 --> 00:51:04,533
SIMILAR TOOLS FOR THOSE TOOLS

1016
00:51:04,533 --> 00:51:05,567
SPECIFICALLY YOU DEVELOPED TO

1017
00:51:05,567 --> 00:51:09,505
MOUSE DATASETS OR DO WE NOT HAVE

1018
00:51:09,505 --> 00:51:10,772
THE DATA THERE?

1019
00:51:10,772 --> 00:51:12,641
>> I SEE.

1020
00:51:12,641 --> 00:51:17,713
FIRST I WANT TO CLARIFY FOR

1021
00:51:17,713 --> 00:51:20,382
PREDICTION EVALUATION HUMAN

1022
00:51:20,382 --> 00:51:22,584
CLINICAL DATA, FOR EXAMPLE DATA

1023
00:51:22,584 --> 00:51:25,354
FROM CLINICAL TRIALS, SOME

1024
00:51:25,354 --> 00:51:27,623
SMALLER SCALE CLINICAL STUDY

1025
00:51:27,623 --> 00:51:29,057
FROM PRE-MANUFACTURED SAMPLES OF

1026
00:51:29,057 --> 00:51:33,595
CELL THERAPIES, SO LIKE AT LEAST

1027
00:51:33,595 --> 00:51:36,498
ON TRES USE HUMAN CLINICAL DATA

1028
00:51:36,498 --> 00:51:38,800
TO DEMONSTRATE THE POSSIBILITY

1029
00:51:38,800 --> 00:51:42,838
TO FIND SOME PATIENT FROM THE

1030
00:51:42,838 --> 00:51:43,105
THERAPIES.

1031
00:51:43,105 --> 00:51:50,913
ON THE MOUSE DATA I TALK ABOUT,

1032
00:51:50,913 --> 00:51:53,382
VALIDATION PART.

1033
00:51:53,382 --> 00:51:54,049
>> RIGHT.

1034
00:51:54,049 --> 00:51:57,252
>> THE TARGET.

1035
00:51:57,252 --> 00:52:00,022
SO CURRENTLY TRES NOT APPLY FOR

1036
00:52:00,022 --> 00:52:02,491
MOUSE DATA LIKE CLINICAL DATA IS

1037
00:52:02,491 --> 00:52:03,959
MORE IMPORTANT.

1038
00:52:03,959 --> 00:52:07,863
AND THE MOUSE, ALL OF OUR IS

1039
00:52:07,863 --> 00:52:10,632
MORE FROM THERAPEUTIC VALIDATION

1040
00:52:10,632 --> 00:52:11,066
PART, YEAH.

1041
00:52:11,066 --> 00:52:11,300
>> OKAY.

1042
00:52:11,300 --> 00:52:12,834
>> THE QUESTION IS CAN
YOU?

1043
00:52:12,834 --> 00:52:16,972
>> IF WE HAVE MOUSE MODELS
FOR

1044
00:52:16,972 --> 00:52:17,406
DISCOVERY.

1045
00:52:17,406 --> 00:52:19,775
>> TO GET TO THE HUMAN.

1046
00:52:19,775 --> 00:52:26,114
>> I THINK THE ANSWER IS
YOU 

1047
00:52:26,114 --> 00:52:26,381
COULD.

1048
00:52:26,381 --> 00:52:32,120
LIKE REPURPOSE FOR SOME MOUSE

1049
00:52:32,120 --> 00:52:35,958
DATA, SOME CONFUSION, RIGHT?

1050
00:52:35,958 --> 00:52:38,927
(INDISCERNIBLE) PEOPLE USE FOR

1051
00:52:38,927 --> 00:52:39,127
MOUSE.

1052
00:52:39,127 --> 00:52:39,795
>> RIGHT.

1053
00:52:39,795 --> 00:52:44,666
BRIEFLY TO RESPOND TO THAT, SO,

1054
00:52:44,666 --> 00:52:49,104
THE MOUSE MODEL DESIGNED TO

1055
00:52:49,104 --> 00:52:51,006
MIMIC DATA PATHOGENESIS OF A

1056
00:52:51,006 --> 00:52:54,242
HUMAN, HUMAN DISEASE, IN THAT

1057
00:52:54,242 --> 00:52:58,013
WAY YOU CAN -- SO BY COMPARATIVE

1058
00:52:58,013 --> 00:53:00,983
STUDY OF GENOMIC STUDY, OTHER

1059
00:53:00,983 --> 00:53:06,054
STUDY, SHOW YOU CAN DEFINE WHAT 

1060
00:53:06,054 --> 00:53:08,590
EXACTLY THE HUMAN DISEASE

1061
00:53:08,590 --> 00:53:11,793
PROCESS MOUSE MODEL FOR SO THE

1062
00:53:11,793 --> 00:53:13,762
EXAMPLE SO WE INDUCE IN OUR

1063
00:53:13,762 --> 00:53:18,133
MOUSE MODEL WE INDUCE BY UV, SO

1064
00:53:18,133 --> 00:53:21,803
WE MIMIC THIS PROCESS.

1065
00:53:21,803 --> 00:53:25,474
SO MORE CONFIDENCE TO USE TO

1066
00:53:25,474 --> 00:53:28,377
STUDY THE RESPONSE OF THE UV

1067
00:53:28,377 --> 00:53:30,679
INDUCE HUMAN, THAT'S A QUICK

1068
00:53:30,679 --> 00:53:35,917
ANSWER, SO THAT'S -- BUT BY

1069
00:53:35,917 --> 00:53:41,223
DEFINE MODELING I THINK THIS CAN

1070
00:53:41,223 --> 00:53:42,257
BE ACHIEVED.

1071
00:53:42,257 --> 00:53:44,359
>> ALSO LIKE CURRENTLY
WHAT

1072
00:53:44,359 --> 00:53:46,795
WE'RE ALSO DOING IS USING MOUSE

1073
00:53:46,795 --> 00:53:48,930
DATA TO SEE WHETHER THEY CAN

1074
00:53:48,930 --> 00:53:50,565
MODEL THE CLINICAL OUTCOME OF

1075
00:53:50,565 --> 00:53:54,936
HUMAN, FOR EXAMPLE, IN

1076
00:53:54,936 --> 00:53:56,705
COLLABORATION WITH WAKEFIELD,

1077
00:53:56,705 --> 00:54:00,108
HISTOPATHOLOGY FEATURES FROM 

1078
00:54:00,108 --> 00:54:02,044
MOUSE RESPONDERS AND

1079
00:54:02,044 --> 00:54:03,879
NON-RESPONDERS, CERTAIN THOSE

1080
00:54:03,879 --> 00:54:05,480
HUMAN CLINICAL DATA, MORE WHAT

1081
00:54:05,480 --> 00:54:08,517
WE'RE DOING IS STILL FROM MOUSE

1082
00:54:08,517 --> 00:54:12,587
TO HUMAN, YOU KNOW, PRAGMATIC

1083
00:54:12,587 --> 00:54:15,424
WAY LIKE THIS DIRECTION BUT IF

1084
00:54:15,424 --> 00:54:21,263
YOU COULD REVERSE DIRECTION,

1085
00:54:21,263 --> 00:54:21,797
CRITICISM (INDISCERNIBLE).

1086
00:54:21,797 --> 00:54:21,997
>> HI.

1087
00:54:21,997 --> 00:54:23,465
THANK YOU FOR A VERY INTERESTING

1088
00:54:23,465 --> 00:54:24,032
TALK.

1089
00:54:24,032 --> 00:54:25,567
SO I THOUGHT I WOULD TAKE THE

1090
00:54:25,567 --> 00:54:29,805
OPPORTUNITY TO ASK A SELFISH

1091
00:54:29,805 --> 00:54:31,206
QUESTION THAT'S RELATED.

1092
00:54:31,206 --> 00:54:33,942
IN THE ERA OF MULTIOMICS WHERE

1093
00:54:33,942 --> 00:54:37,546
WE LOOK AT PROTEOMICS AND

1094
00:54:37,546 --> 00:54:38,714
TRANSCRIPTOMICS, THE TWO

1095
00:54:38,714 --> 00:54:39,347
APPROACHES DON'T ALWAYS ALIGN.

1096
00:54:39,347 --> 00:54:42,984
WE MIGHT SEE, YOU KNOW, SOME

1097
00:54:42,984 --> 00:54:45,721
FINDINGS FROM OMIC APPROACHES WE

1098
00:54:45,721 --> 00:54:47,522
DON'T SEE FROM TRANSCRIPTOMIC

1099
00:54:47,522 --> 00:54:49,157
APPROACHES, SORT OF DO YOU

1100
00:54:49,157 --> 00:54:50,092
CONSIDER THAT THERE ARE

1101
00:54:50,092 --> 00:54:52,494
LIMITATIONS TO SOME OF THESE

1102
00:54:52,494 --> 00:54:53,095
TRANSCRIPTOMIC APPROACHES AT

1103
00:54:53,095 --> 00:54:55,764
LEAST IN MY DATA I MIGHT FIND

1104
00:54:55,764 --> 00:54:56,998
SOMETHING THAT'S REALLY RELEVANT

1105
00:54:56,998 --> 00:54:59,234
FROM FLOW OR IMAGING DATA THAT I

1106
00:54:59,234 --> 00:55:01,803
JUST CAN'T FIND IN MY

1107
00:55:01,803 --> 00:55:02,838
TRANSCRIPTOMIC DATA, HOW DO

1108
00:55:02,838 --> 00:55:05,073
YOU -- WHAT DO YOU THINK ABOUT

1109
00:55:05,073 --> 00:55:05,373
THAT?

1110
00:55:05,373 --> 00:55:15,684
>> SO THIS IS ACTUALLY A
GREAT

1111
00:55:15,684 --> 00:55:16,485
QUESTION.

1112
00:55:16,485 --> 00:55:17,686
EXAMPLE, THE ANTI-INFLAMMATORY

1113
00:55:17,686 --> 00:55:26,394
FUNCTION VALIDATE FLOW CYTOMETRY

1114
00:55:26,394 --> 00:55:28,697
AND ELISA.

1115
00:55:28,697 --> 00:55:30,699
(INDISCERNIBLE) EVEN UNTIL NOW

1116
00:55:30,699 --> 00:55:34,636
DO NOT -- LIKE WITH REPEAT THE

1117
00:55:34,636 --> 00:55:37,806
FLOW, WE SEE PROTEIN CHANGE

1118
00:55:37,806 --> 00:55:39,641
COLOR ON TRANSCRIPTOMIC LEVEL,

1119
00:55:39,641 --> 00:55:41,042
OTHER CHANGE BY PCR, STILL DON'T

1120
00:55:41,042 --> 00:55:42,410
UNDERSTAND WHY.

1121
00:55:42,410 --> 00:55:46,014
ALSO FOR SOME OF MY ONGOING

1122
00:55:46,014 --> 00:55:49,918
PROJECT WE LOOK AT MULTIPLY

1123
00:55:49,918 --> 00:55:53,321
PROTEOMICS IMAGING DATA.

1124
00:55:53,321 --> 00:55:55,724
LIKE DIFFERENT PATTERNS,

1125
00:55:55,724 --> 00:55:59,127
COMPARED TO TRANSCRIPTOMIC

1126
00:55:59,127 --> 00:55:59,361
STUDIES.

1127
00:55:59,361 --> 00:56:03,832
ONE REASON IS THE QUALITY.

1128
00:56:03,832 --> 00:56:06,701
FOR MANY PROTEOMICS DATA BASED

1129
00:56:06,701 --> 00:56:12,908
ON EITHER RPTA ARRAYS OR

1130
00:56:12,908 --> 00:56:15,610
MULTIPLEXING LIKE COVAX, NOT

1131
00:56:15,610 --> 00:56:16,044
VALIDATION.

1132
00:56:16,044 --> 00:56:18,947
IF YOU RUN A TRIAL, A LOT OF

1133
00:56:18,947 --> 00:56:22,951
SMEARS, GERM KNOCKOUT YOU DON'T 

1134
00:56:22,951 --> 00:56:24,386
SEE TARGET OR DOWNREGULATION.

1135
00:56:24,386 --> 00:56:27,989
SO I THINK THE QUALITY OF

1136
00:56:27,989 --> 00:56:28,723
MULTIPLEX PROTEOMICS IMAGING,

1137
00:56:28,723 --> 00:56:34,129
FOR EXAMPLE, IS QUITE LIMITED

1138
00:56:34,129 --> 00:56:35,897
UNTIL NOW.

1139
00:56:35,897 --> 00:56:37,499
AND HARDER WORK LIKE

1140
00:56:37,499 --> 00:56:39,201
TRANSCRIPTOMIC DATA DOES NOT

1141
00:56:39,201 --> 00:56:41,469
REFLECT PROTEIN CHANGE WHICH IS

1142
00:56:41,469 --> 00:56:41,903
ANOTHER STEP.

1143
00:56:41,903 --> 00:56:45,440
I THINK THAT THE ANSWER IS EACH

1144
00:56:45,440 --> 00:56:48,076
TECHNOLOGY HAVE LIMITATIONS.

1145
00:56:48,076 --> 00:56:49,678
IDEALLY SHOULD USE BOTH TO DRAW

1146
00:56:49,678 --> 00:56:50,412
CONCLUSION.

1147
00:56:50,412 --> 00:56:52,147
THERE'S NO WAY YOU CAN DRAW

1148
00:56:52,147 --> 00:56:54,015
RELIABLE CONCLUSION BASED ON

1149
00:56:54,015 --> 00:56:57,118
ONLY ONE TECHNOLOGY, RIGHT?

1150
00:56:57,118 --> 00:57:01,389
 

1151
00:57:01,389 --> 00:57:03,225
>> HI.

1152
00:57:03,225 --> 00:57:05,694
SO, MY QUESTION IS ABOUT DATA

1153
00:57:05,694 --> 00:57:06,761
INTEGRATION AND CYTO Seq.

1154
00:57:06,761 --> 00:57:09,297
YOU TRAIN ON A DATASET THAT

1155
00:57:09,297 --> 00:57:10,832
CONTAINS A LOT OF TUMOR TYPES.

1156
00:57:10,832 --> 00:57:14,069
AND THEN YOU USE IT TO PREDICT A

1157
00:57:14,069 --> 00:57:15,270
SINGLE CLINICAL TRIAL.

1158
00:57:15,270 --> 00:57:18,039
HOW DOES THIS MODEL PERFORM

1159
00:57:18,039 --> 00:57:19,774
ACROSS DIFFERENT CLINICAL TRIALS

1160
00:57:19,774 --> 00:57:22,077
OR TUMOR TYPES FOR LIKE OTHER

1161
00:57:22,077 --> 00:57:23,778
ONES THAT IT'S BETTER AT

1162
00:57:23,778 --> 00:57:26,615
PREDICTING AND ONES THAT ARE

1163
00:57:26,615 --> 00:57:27,849
WORSE?

1164
00:57:27,849 --> 00:57:29,618
>> YEAH, SO I GUESS
PROBABLY CAN

1165
00:57:29,618 --> 00:57:33,121
SPLIT YOUR QUESTION IN TWO

1166
00:57:33,121 --> 00:57:35,457
DIFFERENT PART.

1167
00:57:35,457 --> 00:57:38,460
FIRST DATA COLLECTION FROM MANY

1168
00:57:38,460 --> 00:57:41,363
DIFFERENT MODELS, RIGHT?

1169
00:57:41,363 --> 00:57:43,932
FOR EXAMPLE TGF-BETA, WE FIND

1170
00:57:43,932 --> 00:57:46,835
TREATMENT RESPONSE SIGNATURE SO

1171
00:57:46,835 --> 00:57:49,871
DIFFERENT BETWEEN FIBROBLASTS ON

1172
00:57:49,871 --> 00:57:51,773
LYMPHOCYTE, THEY ALMOST INDUCE A

1173
00:57:51,773 --> 00:57:56,645
DISTINCT SET OF DOWNSTREAM

1174
00:57:56,645 --> 00:57:57,112
TARGET.

1175
00:57:57,112 --> 00:57:57,946
CYTO Seq USE SIGNATURE IS

1176
00:57:57,946 --> 00:58:00,348
WRONG IDEA.

1177
00:58:00,348 --> 00:58:02,951
HOWEVER WE CANNOT ACHIEVE

1178
00:58:02,951 --> 00:58:04,552
SPECIFIC PREDICTION BECAUSE MANY

1179
00:58:04,552 --> 00:58:06,488
OTHER CYTOKINES ONLY HAVE THREE

1180
00:58:06,488 --> 00:58:09,457
OR TWO PROFILES AVAILABLE.

1181
00:58:09,457 --> 00:58:13,628
THERE'S NO LUXURY TO DO THE

1182
00:58:13,628 --> 00:58:15,563
LINUX PREDICTION IN GENERAL,

1183
00:58:15,563 --> 00:58:18,300
WHICH MEANS CYTO Seq BY DOING

1184
00:58:18,300 --> 00:58:19,234
LINEAGE PREDICTION IS WRONG

1185
00:58:19,234 --> 00:58:22,504
IDEA, THE ONLY WAY WE COULD DO

1186
00:58:22,504 --> 00:58:22,771
RIGHT NOW.

1187
00:58:22,771 --> 00:58:25,974
AND FOR THE CLINICAL TRIAL

1188
00:58:25,974 --> 00:58:26,474
PREDICTION PERFORMANCE,

1189
00:58:26,474 --> 00:58:29,911
ACCORDING TO OUR SYSTEMATIC

1190
00:58:29,911 --> 00:58:33,014
ANALYSIS IF THE CYTOKINE

1191
00:58:33,014 --> 00:58:35,250
PREDICTION IS ACCURATE, PB 4 IS

1192
00:58:35,250 --> 00:58:36,484
ACCURATE IN CLINICAL TRIALS

1193
00:58:36,484 --> 00:58:37,585
WE'RE EVALUATING.

1194
00:58:37,585 --> 00:58:41,089
THE PROBLEM IS NOT REALLY GOOD

1195
00:58:41,089 --> 00:58:43,925
FOR CERTAIN FAMILY OF CYTOKINES.

1196
00:58:43,925 --> 00:58:47,696
FOR EXAMPLE IL-4 FAMILY DOES NOT

1197
00:58:47,696 --> 00:58:51,132
WORK WELL.

1198
00:58:51,132 --> 00:58:57,906
IN TERMS OF CLINICAL DIVERSITY,

1199
00:58:57,906 --> 00:58:59,541
(INDISCERNIBLE) IN GENERAL, BUT

1200
00:58:59,541 --> 00:59:06,348
MAJOR PROBLEM IS LIMITATION OF 

1201
00:59:06,348 --> 00:59:07,248
IL-4 FAMILIES.

1202
00:59:07,248 --> 00:59:09,184
DOES THAT ANSWER THE QUESTION?

1203
00:59:09,184 --> 00:59:11,753
>> YEAH, THE PROBLEM IS
LACK OF

1204
00:59:11,753 --> 00:59:11,987
DATA?

1205
00:59:11,987 --> 00:59:13,488
>> NO, PLENTY OF DATA.

1206
00:59:13,488 --> 00:59:18,860
YOU SEE LOTS OF THEM.

1207
00:59:18,860 --> 00:59:22,998
REALLY CAPTURING THE -- LIKE

1208
00:59:22,998 --> 00:59:25,567
YESTERDAY, YOU KNOW, CLINICAL

1209
00:59:25,567 --> 00:59:28,503
FUNCTION SYMPOSIUM, LIKE CRUCIAL

1210
00:59:28,503 --> 00:59:31,106
TALK ABOUT STUDY OF CELL

1211
00:59:31,106 --> 00:59:38,213
CYTOKINES INDUCE IN NEW GENES, I

1212
00:59:38,213 --> 00:59:41,649
THINK (INDISCERNIBLE) INSTEAD OF

1213
00:59:41,649 --> 00:59:42,484
DOING LOTS OF DIFFERENT

1214
00:59:42,484 --> 00:59:45,887
EXPERIMENT WHICH MEANS LOTS OF

1215
00:59:45,887 --> 00:59:48,189
DATA FOR IL-14.

1216
00:59:48,189 --> 00:59:50,825
CELLS CAPTURE THE KNOWLEDGE.

1217
00:59:50,825 --> 00:59:52,894
HOWEVER, DON'T KNOW WHY IF YOU

1218
00:59:52,894 --> 00:59:55,230
APPLY THE PREDICTION MODEL TO DO

1219
00:59:55,230 --> 00:59:57,532
A VERY SIMPLE LINEAR REGRESSION

1220
00:59:57,532 --> 01:00:07,242
IT JUST DOES NOT PREDICT IL-4

1221
01:00:07,242 --> 01:00:08,410
BLOCKINGOUT OF OUTCOME.

1222
01:00:08,410 --> 01:00:11,579
LOGICALLY THERE SHOULDN'T BE. 

1223
01:00:11,579 --> 01:00:12,914
THIS IS SOMETHING THAT'S STILL

1224
01:00:12,914 --> 01:00:15,750
(INDISCERNIBLE) US BUT INTEREST

1225
01:00:15,750 --> 01:00:17,185
IN PASSING FORWARD, WILLING TO

1226
01:00:17,185 --> 01:00:21,156
COLLABORATE AND FIGURE OUT THE

1227
01:00:21,156 --> 01:00:21,389
REASON.

1228
01:00:21,389 --> 01:00:24,759
NOT BY LACK OF DATA, BY SOME

1229
01:00:24,759 --> 01:00:26,461
STILL UNKNOWN WHICH CANNOT BE

1230
01:00:26,461 --> 01:00:27,362
EXPLAINED RIGHT NOW.

1231
01:00:27,362 --> 01:00:29,898
>> THANK YOU.

1232
01:00:29,898 --> 01:00:30,231
 

1233
01:00:30,231 --> 01:00:32,700
>> IF THERE ARE NO MORE

1234
01:00:32,700 --> 01:00:34,069
QUESTIONS, WE'RE ALREADY BEYOND

1235
01:00:34,069 --> 01:00:34,836
TIME.

1236
01:00:34,836 --> 01:00:37,072
SO THANK YOU AGAIN SO MUCH FOR

1237
01:00:37,072 --> 01:00:41,443
THIS WONDERFUL TALK.

1238
01:00:41,443 --> 01:00:42,243
[APPLAUSE]

1239
01:00:42,243 --> 01:00:52,243
AND THANK YOU, EVERYONE,