>> I AM THOMAS FROM THE NIH AND I'M 1 OF THE CO CHAIRS OF THIS DAY TODAY. AND TODAY IS REALLY TO--THE NAMES ARE GONE SO I WANT TOED PAY HOMAGE TO ALL OF YOU WHO ARE HERE REPRESENTING PSYCHIATRIC GENETICS AND GENOMICS AND THE IDEA OF THIS DAY IS REALLY TO CELEBRATE YOU ALL, OF ALL THE PROGRESS THAT YOU AS A FIELD HAVE MADE OVER THE LAST 10-12 YEARS. 10-12 YEARS REALLY BUT ALSO TO POINT OUT SOME OF THE BOTTLENECKS AND COMPLEXITIES THAT STILL LAY AHEAD OF US. LET ME SEE IF I CAN DO THIS. NOPE. SO FIRST, I WANT TO THANK ALL THE PROGRAM COMMITTEE MEMBERS THAT HAVE PUT THE SCIENTIFIC PROGRAM TOGETHER FOR TODAY. HIGH CO CHAIR DR. STEVE HEIMAN, AND MY CO-CHAIR AND HE IS ALSO PAYING FOR ALLLET FOOD AND OF COURSE, YOU KNOW WE ARE NIH, AND WE CAN'T PAY FOR FOOD OR BEVERAGES OR THANK YOU TO THOSE, AND THEN PAMELA WHO COULD NOT MAKE IT TODAY BUT SHE IS ON THE WEB. THANKS ALL OF YOU FOR PUTTING TOGETHER A GREAT PROGRAM AND TAKING THE HEAT OFF ME. IF YOU DON'T LIKE SOMETHING, LET'S BLAME THEM. . SO, AND--YEAH, SHE'S HIDING, SHE'S SHY. LORI IS RIGHT HERE. [ APPLAUSE ] YOU KNOW THESE THEY THEY--THINGS ARE COMPLEX FROM THE GET-GO, IT'S DIFFICULT TO DO THIS AND I HAVE A TENDENCYY TO SAY LET'S DO THIS AND THIS AND THAT AND I HOPE LORI LEARNED IN THIS PROCESS TO HELP ME TO GO F-MYSELF AND IF NECESSARY, SHE'S BEEN VERY NICE AND I THANK YOU FOR ALL THE WORK YOU HAVE DONE AND IF THERE ARE ANY LOGISTICS ANNOUNCEMENTS, WE CAN MAKE THEM NOW. SHE DIDN'T WANT TO DO THAT BUT I INSISTED. >> HI, I WANT TO BRIEFLY INTRODUCE MYSELF, IF YOU NEED ANYTHING LOGISTICALLY, JUST COME SEE ME. JUST A FEW REMINDER, LUNCH IS AT 11:50, HOPEFULLY WE STAY ON TIME. IT'S ABOVE US WHERE BREAKFAST WAS SERVED AND IN YOUR PROGRAM, AFTER EACH SERIES OF TALKS THERE IS DISCUSSION, IF YOU CAN JUST LINE UP AT THE MICROPHONESOT MICROPHONESOT--MICROPHONES ON THE AISLES AND THAT'S ALL. OH YES, THERE WILL BE A TEAM SCIENCE PHOTO AT 11:50 WHEN WE BREAK FOR LUNCH. IF EVERYBODY COULD BE HERE. >> WE CAN'T TAKE PICTURES OF ALL OF YOU BECAUSE NOT ALL OF YOU SIGNED WAIVERS. >> THANK YOU. >> AND THEN I WANT TO THANK MY BOSS' HERE, DR. JOSH GORDON, DIRECTOR OF NIMH FOR SUBSCRIBING THIS MEETING AS WELL. SO, SO WE'RE CALLING THIS PSYCHIATRIC GENOMICS IN THE ERA OF TEAM SCIENCE AND IT'S TEAM SCIENCE THAT HAS BEEN AT THE CENTER OF THE ADVANCES AND WHAT YOU GUYS SEE HERE, I ASKED SOME OF YOU BEFORE, IF YOU COULD GUESS WHAT THIS IS, THIS IS A CO-PUBLICATION NETWORK IN 5 OF YOUR INTERVALS, OR 4 OF YOUR INTERVALS STARTING FROM 2000 THAT SHOWS HOW PEOPLE, SCIENTISTS IN PSYCHIATRIC GENOMICS HAVE PUBLISHED TOGETHER AND YOU CAN REALLY SEE THE INCREASE IN THE GROWTH OF YOUR COLLABORATIONS AS A METRIC. CO INSURE MARKET DENTALLY, THE LAST 1 LOOKS LIKE A SLIDE SHOW, AN INTERGENERATIONAL SLIDE SHOW. IF YOU SEE HERE--THIS IS THE SOLAR CELL. BUT THAT'S JUST AN ASIDE. SO, YOU KNOW TEAM SCIENCE HOWEVER, THE TEAM SCIENCE CONSORTIUM, MANY OF YOU MAY THINK, WE'RE ONLY TALKING ABOUT LARGE COLLABRATORIES WITH THOUSANDS OF INVESTIGATORS, THEY REALLY COME IN MANY DIFFERENT FLAVORS OR AS A FRIEND OF MINE, LIKES TO SAY, EACH FAMILY COMES IN ITS OWN, IN THIS CASE, HAPPY CIRCUMSTANCES SO I'M USING PTC AS AN EXAMPLE, SEQUENCING AND SOCIOLOGICAL GENOME AS ANOTHER BUT A SMALLER ASSOCIATE COME TOGETHER AND EXCHANGE DATA BEFORE PUBLICATION, OR EXAMPLE OF THE GENESIS CONSORTIUM, THE SMALL GROUP OF INVESTIGATORS CAME TOGETHER A COUPLE OF YEARS AGO, ACROSS THE COUNTRY, AND WORKED ON MAPPING SCHIZOPHRENIA, DE NOVO MUTATIONS, 1 OF THE FIRST PAPERS IN THAT SPACE. SO AS I SAID TEAM SCIENCE COMES IN MANY DIFFERENT FLAVORS AND IT SPEAKS TO A WAY OF COLLABORATE ING AND PERSPECTIVE ON DOING RESEARCH THAT IS NOVEL. AND REALLY HAS DEVELOPED AND SPRUNG UP OVER THE LAST 10-20 YEARS AND IT'S NOT THE ONLY FACTOR THAT HAS ADVANCED IN THE GENOMICS AND AS MUCH AS I WOULD LIKE TO TAKE ALL THE CREDIT FOR IT, I CAN'T BECAUSE I CAME TO NIH IN 2004. YOU KNOW REALLY AT A CONFLUENCE OF MANY FACTORS THAT HELPED US MOVE THE FIELD THE WAY WE ARE MOVING IT. IT WAS TEAM SCIENCE, IT WAS TECHNOLOGICAL AND CONCEPTUAL ADVANCES, AS YOU KNOW THE GENOME PROJECT WAS COMPLETED IN THE BEGINNING OF--I THINK IT WAS 2003, WITH IT CAME TECHNOLOGICAL ADVANCES IN PARALLEL ARRAY, SEQUENCING, GENOME TYPING AND NEXT GENERATION SEQUENCING AND OF COURSE, A DEVELOPMENT OF GENOMIC RESOURCES AND THE RECOGNITION THAT THEY'RE IMPORTANT AND TO COVER ALL THIS WAS A DEVELOPMENT OF A POLICY FRAMEWORK FOR SHARING AND ACCESS OF DATA AND SAMPLES. --THE GENOMICS YOU KNOW WE TRIED TO--ALL THE WAY--SIMPLY THE ONLY CONNECTIONS WE WERE ABLE TO MAKE WERE THE ONES TO THE VARIATIONS IN DNA AND BEHAVIOR SAYS MOLECULAR CELLS IN THE SCALE, ACROSS CIRCUITS AND BEHAVIORS, I THINK THESE ARE THE BIG CHALLENGES THAT THEY'RE STILL FACING AND AGAIN, THANK YOU TO PUT A PLUG IN FOR NIMH, RECENTLY RELEASED A NEW INITIATIVES THAT WE CALL NEUROSCIENCE APPROACHES FOR INTEGRATING LEVELS OF ANALYSIS, WE WILL ASK THE FIELD, YOU GUYS TO REALLY THINK ABOUT THESE PROBLEMS IN NOVEL WAYS OF CONNECTING DIFFERENT LEVELS OF ANALYSIS. AND BEFORE I HAND IT OVER TO STEVE, I WANT TO END THIS ON A MORE PERSONAL NOTE, AS YOU KNOW OVER THE LAST FEW YEARS WHEN YOU HEARD ME ON THE PHONE, I SOWBDED HIGH AND OFTEN I WOULD ATTEST TO THAT AND UNFORTUNATELY THIS THESE WERE ALL LEGAL HIGHS, SO AS I RECOVERED, OVER THE LAST YEAR, I'M SURE I SOUNDED DIFFERENTLY BUT THERE WERE OTHER NOISES IN THE BACKGROUND AND I WANT TO COME CLEAN AND TELL YOU WHAT THIS WAS NOW. AND I WANT TO INTRODUCE YOU TO MY ASSISTANT THAT COULDN'T MAKE IT TODAY, IT WILL BE ON THE INTERNET, BUT THEY HAVE TO SAVE THE HOUSE FROM MARINE ATTACKS AND THEY ULS WANT ME TO TELL YOU THAT THEY--UNLIKE THEIR DADDY, THEY REALLY LIKE MARINE MODELS AND THEY WANT YOU TO DO MORE. SO WITH THAT I WELCOME YOU TO A WONDERFUL DAY OF SCIENCE AND I WANT TO HAND IT OVER TO STEVE. [ APPLAUSE ] >> WELL, THANK YOU THOMAS. YOU'RE ACTUALLY VERY MODEST. BEFORE I COME BACK TO THAT, I DO WANT TO THANK LAURA BIGMAN WHO HAS THE UNUSUAL COMBINATION OF TRADE WHO IS ALSO POLITE AND RELENTLESS WHICH IS THE ONLY WAY THINGS GET DONE SO THANK YOU LAURA. SO THOMAS YOU SET ME UP FOR THIS. THERE WAS 1 PHONE CALL WE HAD WHERE THOMAS DID SOUND, WELL, HIGH, YOU KNOW STEROIDED UP AND I SAID WHAT'S UP THOMAS, HE SAID WELL I'M GETTING A BONE MARROW TRANSPLANT IN 2 HOURS. SO, THAT'S A GUY WHO NEVER SICKS OUT. THAT REALLY IS REMARKABLE AND WE'RE LUCKY TO HAVE YOU HERE LOOKING SO GOOD, THOMAS. OKAY, SO, PECULIARLY I HAVE AN INTRODUCTION AND THEN I GO INTO A HISTORICAL TALK, AND I THINK I GOT THE HISTORICAL TALK BECAUSE I'M SO OLD BUT NOT SO OLD THAT I KNEW CRAPELIN, BUT I THINK IT'S IMPORTANT TO REMIND US WHY WE'RE DO THANKSGIVING AND I THINK WE ALL KNOW WHEN WE LOOK AT THE MAJOR CLASSES OF DRUGS THAT WE USE TO TREAT MENTAL ILLNESS, THERE HAVE BEEN NO NEW THERAPEUTIC MECHANISMS SINCE THE 1950S. WE HAVE NO BIOMARKERS DESPITE THE DESPICABLE COMPANIES OUT THERE SELLING GENETIC TESTS AND SO FORLGT. AND WHAT'S REALLY CHALLENGING IS THERE HAVE BEEN NO ADVANCES IN EFFICACY SINCE CLOZEPINE, WHICH WAS DISCOVERED IN THE 1960S, OF COURSE IT'S SPECIAL EFFICACY WAS ONLY ESTABLISHED IN 1988 DELAYED BECAUSE OF SIDE EFFECTS BUT YOU KNOW DESPITE THE TOLERABILITY AND SAFETY OF CURRENT SSRIs AND SNRIs, NONE OF THEM IS MORE EFFICACIOUS THAN THE 2 VERY FIRST ANTIDEPRESSANTS WHICH DATE FROM 1957 AND THEN 1 OTHER EXAMPLE, THIS IS A SLIDE NOT FROM AN ANTISIGNIFY CHIATRY ADVOCATE BUT FROM THE FDA AND WHAT IT CHOSE IS THE AGGREGATE OF ALL CLINICAL TRIALS OF ANTIDEPRESSANTS THAT ULTIMATELY WERE APPROVED SO THESE WERE DRUGS THAT ALL GOT SOPHISTICATED TO THE MARKET AND WHAT YOU CAN SEE IS IF YOU AGGREGATE ALL OF THESE TRIALS, IT'S SLIGHTLY--THE DRUG PLACEBO DIFFERENCE IS ONLY 2-POINTS ON THE HAMILTON WHICH IS AWFUL, RIGHT? THAT'S ALMOST MINUTE TO MINUTE VARIATION DEPENDING ON WHEN YOU DO THE INTERVIEW. OF COURSE WE KNOW THAT A LOT OF THESE DRUGS ACTUALLY WORK, THAT'S WHY THEY WERE APPROVED BUT THE PROBLEM IS WITHOUT A BIOMARKER, THIS IS NOT PATIENT LEVEL DAT AYOU HAVE A SPREAD FOR EACH DRUG OF PEOPLE WHO REALLY BENEFIT AND PEOPLE WHO DON'T BENEFIT AT ALL BUT WE CAN'T TELL WHO THEY ARE WHICH LEADS TO LARGE AND INEFFICIENT CLINICAL TRIALS IN A LOT OF DIFFICULTY IF WE HAD BETTER MOLECULAR TARGETTINGS OF GETTING THESE DRUGS APPROVED. SO WHY THIS HALF CENTURY OF STATESIS, A LOT OF MY COLLEAGUES HAVE SEEN THIS 1. ALL-TOO-OFTEN. PARTLY WE KNOW THAT THE BRAIN IS EXTRAORDINARILY COMPLEX, YOU KNOW LOTS OF PEOPLE HAVE TRIED TO REVERSE ENGINEER THE BRAIN BY PUTTING--WITH PHARMACOLOGIC CHALLENGES BUT THAT'S A BIT OF AN INSANE PROPOSITION, THE BRAIN WITH A HUNDRED TRILLION SYNAPSES HAVE ALL THESE WORKING CAUSAL PATH. YOU PUT IN A DRUG AND A LOT OF STUFF HAPPENS AND BECAUSE THERE IS NO ANIMAL MODEL TO ULTIMATELY TELL YOU THAT HAVE YOU CURED OR TREATED BECAUSE OF BIPOLAR DISORDER YOU NEVER ACTUALLY GET AN ANSWER, ALTHOUGH YOU CAN WRITE PAPERS AND GET GRANTS. SO THERE'S THIS COMPLEXITY, THE LIVING HUMAN BRAIN IS INVALUABLE AS I'M FOND OF SAYING, IT'S A HARD PROBLEM, CANCER BIOLOGIST CAN ASK A SURGEON TO HAND OVER THE CELL AFTER AN EXCISIONAL BIOPSY WHEREAS, YOU KNOW THE INCAs HAD THIS PROCEDURE THAT LET OUT EVIL SPIRITS, NOT FDA APPROVED TO MY KNOWLEDGE BUT WE CAN'T GET LIVING TISSUE AND YOU KNOW EVEN IF WE COULD IT WOULD NOT HELP BECAUSE IT HAS TO DO WITH CIRCUITS. OUR IMAGES ARE WONDER BUT THEY DON'T HAVE THE RESOLUTION TO LOOK AT MOLECULAR AND CELLULAR MECHANISMS. IT MIELT IF WE EXPLOITED IT BETTER AND ALSO BECAUSE OF INTRINSIC HETEROGENEITY, WE SIMPLY CAN'T AFFORD TO DO ENOUGH STUDIES TO DO THEM, PURPOSELY TO STUDY THE BIOMARKERS, IT'S CHALLENGING AND THE SYMPTOM BASED DIAGNOSIS MAKE IT HARDER FOR THESE BIOLOGICAL INVESTIGATORS DOING IMAGINGLY AND OTHER INDIRECT STUDIES BECAUSE FAMOUSLY THE CRITERIA COULD BE 256 WAYS OF MAKING THE DIAGNOSIS AND SINCE IT'S 5 OF 9, 2 PEOPLE WANTING TO STUDY CAN DO THE SAME STUDY ON THE SINGLE CRITERION AND MY UNDERSTANDING IS THAT THE NEW CRITERION FOR POST-TRAUMATIC STRESS DISORDER MAKE 4 BILLION WAYS OF MAKING THE DIAGNOSIS AND FINALLY AS ALLUDED TO, OUR ANIMAL MODELS JUST DON'T CAPTURE HUMAN DISEASE BIOLOGY EVEN THOUGH I THINK IT'S REALLY IMPORTANT TO SAY, WE ABSOLUTELY NEED ANIMAL MODELS FOR BASIC SCIENCE AND BASICALLY--GETTING INITIAL UNDERSTANDING OF WHAT GENES ARE DOING AND SO FORTH BUT THEY DO NOT CAPTURE HUMAN DISEASE BIOLOGY. SO THE IMPORTANT THING IS THAT TODAY WE ARE HERE AND THIS IS SORT OF LIKE THE SHARED INVOCATION BECAUSE I THINK MOST PEOPLE--I HOPE EVERYONE IN THE AUDIENCE SHARES WITH YOU THAT OUR VERY BEST CHANCE EVER IN HISTORY TO ELUC DATE DISEASE MECHANISMS AND THEREFORE TO GET BIOMARKERS AND IMPROVE THERAPEUTICS WHICH IS SO IMPORTANT RESTSOT SUCCESSFUL PROSECUTION OF GLOBALLY CONDUCTED, RIGHT, ALL POPULATIONS UNBIASED LARGE SCALE STUDIES. THAT IS NOT ALL WE HAVE TO DO BUT THAT IS THE BASIC PLATFORM INFORMATION FROM WHICH WE WILL HAVE THE RIGHT TO HAVE NEW HYPOTHESIS AND STOP THINKING ABOUT, YOU KNOW THE SEROTONIN REUPTAKE TRANSPORTER. BY THE WAY, IN--I OPENED NEUROPSYCHOPHARMACOLOGY THIS MORNING EARLY ON E-MAIL AND THERE WAS A PAPER, UNDER POWER CANDIDATE GENE STUDY OF THE SEROTONIN REUPTAKE TRANSPORTER AND IT MAKES ME WANT TO CRY. OKAY, SO THE PROMISE OF GENETICS IF RIGOROUSLY CALCULATED AND APPLIED, IS SO IMPORTANT, EVERYBODY IN THIS AUDIENCE KNOWS BECAUSE IT'S SO IMPORTANT, THESE ARE FIX FIXED AND THEREFORE WE MAKE ASSOCIATIONS THEY ARE CAUSAL. ANY OTHER BIOLOGICAL FINDING INCLUDING THE EPIGENETICS, WE CAN GET THE HEADS BUT NOT THE RIGHT TISSUES COULD BE EFFECTS BUT NOT CAUSES. YOU ARE BORN WITH THESE SEQUENCES, NO SOWMATIC MUTATIONS& ARE IMPORTANT, THEY GET AN ADVANTAGE SO HEMI, MEG, AND CHRIS HAS DONE FABULOUS WORK SHOWING HOW SOMATIC CAN LEAD TO INTRACTABLE CONFOCAL EPILEPSYS BUT WHETHER OR NOT THESE CONTRIBUTE TO PATHOGENESIS OF THESE COMMON MENTAL DISORDERS. I MEAN SOME PEOPLE JUMP ON THESE BAND WAGONS TOO HARD AND THESE ARE HIGHLY INHERITABLE AND THEREFORE WE SHOULD NOT LOSE OUR FOCUS ON THIS INVESTIGATION, WHY WE'RE INVESTIGATING THESE RPGHT AREAS. ALSO, I THINK MANY OF US BELIEV IN A GENE'S FIRST VIEW, WHICH IS PSYCHIATRY HAS BEEN PLAGUED BY A SMALL NUMBER OF BIOLOGICAL HYPOTHESIS, MOSTLY REVERSE ENGINEERED FROM THE ACTIONS OF PHARMACOLOGY. DOPAMINE AND GLUTAMATE THEORIES OF SCHIZOPHRENIA. THEY'RE JUST FALLACIES THAT JOE WARNED US ABOUT IN 1960 AND EVERYBODY READ HIS PAPERS AND SAID WHAT A RIGOROUS MAN AND WENT ON TO DO THESE THINGS. WHAT WE NEED, AND AGAIN THIS IS WHAT THE GENETICS WILL PROVIDE ARE LARGE DATA SETS THAT GIVE US NEW OBSERVATIONS, NEW BIOLOGY BECAUSE THEY ARE UNBIASED, RIGHT? AND THEN WE WILL HAVE EARNED A WHOLE SET OF NEW HYPOTHESIS, THAT WE CAN EXPLOIT. THE OTHER THING, AGAIN, I THINK, NO NEED TO SAY IT IN THIS AUDIENCE BUT SO IMPORTANT AGAIN, LOOKING AT THE PSYCHIATRY JOURNALS, I THINK THE PEOPLE HERE HAVE TO GO OUT AND BE LEADERS AND MENTORS, THIS IS REALLY ON ALL OF US THAT IF YOU'RE GOING TO INVEST GRAD STUDENT AND POST DOC LIVES IN BIOLOGICAL FOLLOW UP STUDIES, THEN IT'S REALLY IMPORTANT THAT THE GENETIC FINDINGS ON WHICH THESE ARE BASED ARE ESTABLISHED WITH THE UTMOST RIGOR, NO P-HACKING, NO SMALL SAMPLES, NO CANDIDATE GENES BUT ARE REALLY RIGOROUSLY ESTABLISHED, APPROPRIATELY FINE MAPPED, NOT WISHFULLY FINE MAPPED BECAUSE AFTER ALL GRADUATE STUDENT AND POST DOC LIVES ARE VERY VALUABLE AND THERE WILL BE A LOLT TO DO BUT WE REALLY HAVE TO MAKE SURE THAT THAT MESSAGE GETS ACROSS. AND I'M EXCITED TO HAVE THIS GROUP HERE BECAUSE I THINK YOU ALL ARE THE LEADERS IN THIS AND I WANT TO THANK THOMAS AND THEN ALL OF HIS COLLEAGUES HERE, WHO HAVE WHO MADE THESE POSSIBLE. NOW 1 YOU WILL RECOGNIZE JIM WALTON. SO WHEN I WAS NIMH DIRECTOR, HE TOLD ME A NUMBER OF TIMES THAT THE REASON HE SUPPORTEDLET GENOME PROJECT AS HE OFTEN SAID HIS SON RUFUS HAS A SEVERE MENTAL ILLNESS AND HE THOUGHT THAT GENETICS WAS THE ONLY WAY THEY WOULD EVER UNDERSTAND WHAT WAS WRONG WITH HIS SON. AND I RATHER BELIEVED IT. HE SAID IT WITH REAL CONVICTION. AND THE REST, WE WILL FIGURE OUT WHO THEY ARE LATER. SO, SO THIS IS--I--LET ME--I THINK IF THEY'LL LOAD THE NEXT THE SECOND PART WHOEVER HAS THE--OR DO I DO THIS UP HERE WITH NO, IT'S UP. OKAY, SO I'VE BEEN ASKED TO TALK A BIT ABOUT THE HISTORY AND IN SOME WAYS THE ORIGIN OF MODERN PSYCHIATRY WITH CREPLIN, YOU'LL SEE THAT WAS THE PICTURE OF 1 OF THOMAS' COLLEAGUES ON THE LAST SLIDE. SO BASED ON CAREFUL OBSERVATION, FAMOUSLY, EVERYBODY KNOWS, CREPLIN, TRIED TO DISTINGUISH SCHIZOPHRENIA HAD WHICH HE CALLED DEMENTIA PREYCOX, WHICH WAS THE EARLY DECLINE, YOU WANT RETURN TO THE NAME FROM YOUR STUDIES AND WHAT HE CALLED MANIC DEPRESSIVE INSANITY AND HE FOCUSED ON SYMPTOMS CAREFULLY BUT HE ALSO FOCUSED ON THE COURSE OF THESE ILLNESSESS WHICH I THINK IS VERY IMPORTANT BECAUSE IT WAS A HOPEFUL WAY OF DISTINGUISHING DISORDERS BECAUSE IT HAS NOT PANNED OUT AND I WILL COME BACK TO THAT. HE ARGUED THAT THESE ARE BIOLOGICAL IN ETIOLOGY WHICH WAS A CONTENTIOUS POINT, AND RECOGNIZING THEIR FAMILIAR NATURE--WE COULD SEE THAT PATIENTS WITH SCHIZOPHRENIA AND BIPOLAR DISORDER WHILE SOME OF THEM READ HIS BOOK AND BEHAVED IN THAT WAY, MANY, MANY MANY HAVE SHARED SYMPTOMS OR CHANGED THEM OVER THE LIFETIME. THIS IS ALL--OBVIOUSLY IMPORTANT TO GENETICS BECAUSE THESE ARE THE PHENOTYPES THAT WE ARE STILL ASCERTAINING THAT WE'RE STILL ENTERING INTO OUR STUDIES, IN FACT MANY OF THE LARGE GENETIC STUDIES THAT WE STILL DO STILL HAVE THE LIMITATION THAT ALL WE KNOW ABOUT A PERSON IS THEIR CATEGORICAL DIAGNOSIS, THEIR BIOLOGICAL SEX AND THEIR COUNTRY OF ORIGIN AND SO A LOT RESTS ON THESE KINDS OF CATEGORIES. THE DSM SYSTEM REALLY IS WITH SOME REARRANGING OF DECK CHAIRS IS VERY MUCH LIKE THE DSM 3 WHICH HAD IT THROUGH CREPE LYNN'S OBSERVATIONS AND THROUGH WASHINGTON UNIVERSITY AND THE 60S AND 70S, AND ROBINSON AND GUZIERE IN A SHORT PAPER ABOUT SCHIZOPHRENIA SAID THAT VALID DIAGNOSIS WOULD EMERGE ULTIMATELY FROM,--THEY DIDN'T CLAIM TO HAVE IT, BUT THEY SAID IT WOULD EMERGE, LABORATORY STUDIESS WE STILL DON'T HAVE ANY BIOLOGICAL TESTS FOR THESE DISEASES. DELINEATION OF DISORDERS, THIS WENT UNFORTUNATELY TOWARD THE IDEA OF DISCONTINUOUS CATEGORIES. COURSE OF ILLNESS, VERY CREPELINIAN AND HISTORY. AND THEN THROUGH THE FINER CRITERIA AND RGC, THESE GROUPS CAME UP WITH THE IDEA OF VERY SPECIFIC OPERATIONALLIZED CRITERIA, SO THAT FOR EXAMPLE, INSTEAD OF SAYING SCHIZOPHRENIA IS CHRONIC WHICH CAN BE INTERPRETED IN MANY DIFFERENT WAYS, ULTIMATELY THE DSM SET A THRESHOLD OF 6 MONTHS. THE PROBLEM WITH THAT PSEUDOPRECISION A IS WE ACTUALLY USUALLY CAN AT ANY TIME WHEN SCHIZOPHRENIA IS STARTED BUT MORE THAN THAT, THEN ALL KINDS OF SILLY FILLER DIAGNOSIS WERE NEEDED LIKE SCHIZOPHRENIA FORM DISORDER WHICH WAS SOMEBODY WHO LOOKED PSYCHOTIC WHO WASN'T SICK FOR 6 MONTHS YET AND THEN THE OTHER DIAGNOSTIC SYSTEMS, LIKE THE WORLD HEALTH ORGANIZATIONS, SYSTEMS BY SPECIFYING THINGS WE DON'T KNOW AND JUST FROM THE VERY OUTSET AND THEN THE GROUP THOUGHT--IT APPEARED IN 1970 BUT IN 1971 COMPLETELY UNKNOWN PAPER JUST LIKE SCETS FRENNIA WHO END UP HAVING AFFECTIVE DISORDER WHICH ANYONE WHO IS A CLINICIAN IS COMMON ENOUGH TO SEE WAXING AND WANING OF AFFECTIVE SYSTEMS AND A REMIND THEY'RE OUR PATIENTS DON'T READ OUR TEXTBOOKS SO FOR GENETICISTS AND BIOLOGICAL PSYCHIATRISTS IN SOME WAYS THAT THE ORIGINAL RELIANCE ON DISFOR CONDITIONS THAT REALLY ARE BETTER CAPTURED AS CONTINUA AND THE GENETICS HAVE MADE THIS POINT VERY, VERY STRONGLY, IN TERMS OF CONTINUOUS BEING CONTINUOUS NORMALCY AND SPECTRA BECAUSE THEY SHARE BOTH GENETIC RISK FACTORS AND SYMPTOMS WITH EACH OTHER AND THIS IS THE AFFECTIVE DISORDER SHARING THAT WE SEE, THE OTHER THING IN THESE CATEGORIES DIAGNOSTIC THRESHOLDS ARE ARCH CONTRARY, AND THAT AGAIN THAT MAKE ITS CHALLENGING TO DO ANY KIND OF BIOLOGICAL STUDY AND DENIES TREATMENT TO CHILDREN OR TO PEOPLE WITH 4 OF 9, BUT BAD--ALTHOUGH A LOT OF DOCTORS ARE WISE ENOUGH TO IGNORE THAT, AND THEN THE OTHER PROBLEM IS THAT WITH THE GOAL OF ACHIEVING HOMOGENIUS GROUPS, THE DSM, THE PEOPLE WHO WROTE THE DSM, LET SPLITTING RUN AMUCK AND WITH SCANT EVIDENCE AND THAT RESULTS IN NARROW UNNATURAL DIAGNOSTIC SILOS AND THAT RESULTS IN ARTIFACTUAL CO MORBIDITY AND SHIFTING LIFETIME DIAGNOSIS, SO MANY PEOPLE AS YOU KNOW WITH ASDs, ALSO HAVE OCDs AND ADHDs AND CONDUCT ISSUES AND PEOPLE WITH MOOD AND ANXIETY DISORDERS HAVE MULTIPLE DIAGNOSIS. I GUESS THE ONLY BENEFIT FOR SOME PEOPLE HERE IS WE HAVE PGC GROUPS ACROSS THE DISORDER ANALYSIS AND PUT HUMPTY DUMPTY BACK TOGETHER AGAIN BUT ON BALANCE IT WOULD HAVE BEEN BETTER TO THE TO HAVE ENGAGED THAL ORGY OF UNNECESSARY SPLITTING. BUT NOW HERE, YOU KNOW HAVING MADE FUN OF THE DSM, IT'S EASY TO DO. I'VE BEEN ON DSM COMMITTEES, DO YOU IT ON THOSE COMMITTEES. THIS IS INTERESTING PARADOX WHICH IS THAT THE MAJOR HIOF HISTORICALLY SYNDROMES LIKE SCILTS FRENNIA, BIPOLAR DISORDER OR AFDs ARE INHERITABLE. I THINK THESE ARE IMPORTANT HISTORICALLY WELL-STUDIED DISORDERS AND THE DSM HAVE VERY LTLE TO DO WITH THE FACT THAT HE'S ARE HERITABLE. IN FACT THE RESULTS OF THE GENETICS AS WILL BE SEEN LATER REALLY UNDERMINE THE CATEGORICAL SEPARATENESS BUT THE IMPORTANT THING ABOUT UNBIASED LARGE SCALE GENETICS STUDY SYSTEM THAT ULTIMATELY POWER OVERWHELMS DIAGNOSTIC NOISE AND SO, THE FACT THAT THESE DISORDERS ARE ACTUALLY HETEROGENEOUS HETEROGENEOUS AND T HE DSM DOESN'T PICK OUT CATEGORICAL ENTITIES DOESN'T IMPEEZ PEDE ADEQUATELY POWERED STUDIES. OT OTHER HAND IF SOMEBODY DID A SMALL STUDY SAY OF MAJOR DEPRESSION, IT WOULD BE REALLY HOPELESS. YOU NEED FOR A VERY HETEROGENEOUS ROW GENIUS DISORDER LIKE THAT, ENORMOUS STATISTICAL POWER, BUT AGAIN, THESE ENTITIES AREN'T REAL, THEY'RE NOT--BUT THEY'RE CAPTURED IN NATURE BUT THEY ARE THESE COMPLEX DIMENSIONAL--WITH RESPECT TO NORMALCY AND SPECTRAL WITH RESPECT TO OTHER DIAGNOSIS CONDITIONS THAT TAKE ENORMOUS POPULATIONS TO FIND. NOW, ON THE OTHER HAND FOR OTHER AREAS LESS FORTUNATE THAN GENETICIST, IF THE DSM CRITERIA ARE LITERALLY TAKEN, LITERALLY BY REGULATORS AND THANKS TO THE R-DOC INITIATIVE, THEY'RE LEARNING THAT THAT'S NO LONGER OUR STUDY SECTIONS IN JOURNAL REFEREES, THEN THE DSM IS DISASTROUS FOR CLINICAL TRIALS AND NEUROIMAGING AND OTHER BIOLOGICAL STUDIES WHERE YOU CAN'T MUSTER THE STATISTICAL POWER TO OVERCOME THE DIAGNOSTIC NOISE AND HETEROGENEITY. ALL RIGHT, SO, THESE ARE THE ENTITIES WE HAD AND AGAIN THINKING NOT SO MUCH OF THE PRECISE DIAGNOSTIC SYSTEMS BUT OF THESE HISTORICAL ENTITIES LIKE SCHIZOPHRENIA, HISTORICALLY, YOU KNOW VERY FAMOUSLY, EVEN THOUGH IN SOME SENSE IT WAS ALREADY KNOWN THAT GENES PLAYED AN IMPORTANT ROLE IN SCHIZOPHRENIA AND THEY MADE A REALLY IMPORTANT HISTORICAL OR RHETORICAL POINT IN THEIR ADOPTION STUDIES, AND SHOWED THAT ADOPTEES TENDED TO LOOK MORE LIKE THEIR BIOLOGICAL PARENTS THAN THEIR ADOPTIVE PARENTS THAT GENES WERE NOT ENOUGH, THAT ENVIRONMENT OR SOME OTHER FACTORS ALSO MATTERED AND CONSISTENT WITH THAT ARE EARLIER STUDIES BY GOTS MAN AND SHIELDS, SHOWING THAT THE MORE--THIS IS FAMOUS--LOTS OF US USE THIS SLIDE IN THE AUDIENCE BUT THE& MORE YOU SHARE YOUR RISK OF SCHIZOPHRENIA, THE HIGHER THE RISK OF SCHIZOPHRENIA. GENES ARE NOT BY THEMSELVES, FATE AND THERE'S A CERTAIN AMOUNT OF FUNNY BUSINESS BECAUSE NOT ALL FIRST DEGREE RELATIVES SHARE THE SAME RISK WHICH RAISES QUESTIONS HERE ABOUT EPIGENETICS AND OTHER PROCESSES. TWIN STUDIES CONFIRMED OBVIOUSLY THE HERITABILITY OF THESE MENTAL ILLNESSES, SCHIZOPHRENIA IS BIPOLAR AND SO FORTH, BUT ALSO QUANTIFY THE AGGREGATE GENETIC EFFECTS AND DIDN'T TELL YOU ABOUT THE UNDERLYING COMPLEXITY AND ALREADY BY PRETTY EARLY IN THE HISTORY OF PSYCHITRY, WE KNEW THERE WERE NONMENDELIAN MDZD RATIOS AND GOTS--GOTTESMAN HAD ALREADY PUBLISHED A PAPER ON THIS, BUT THE PROBLEM OF COURSE ISA WE ALL LEARNED IN SCHOOL ABOUT MENDEL'S PEA PLANT HEIGHT, AND THESE ARE HIGHLY POLYGENIC AND IT'S MUCH EASIER TO LEARN MENDELIAN BIOLOGY. AT ANY RATE, I WOULD LIKE TO SAY THAT A POLYGENIC THEORY EMERGED FROM THESE NONMENDELLIAN RATIOS, THESE ARE SIMMINGLY OUT OF THE BLUE BUT JUST TO THE SHUFFLING OF THE DECK OF THE CASES OF SCHIZOPHRENIA AND SO ON AND SO ON BUT AGAIN, THIS PAPER I MENTION SAID GOTS MAN AND SHIELDS IN 1967 WOULD HAVE PUT US ON THE RIGHT TRACK, IT'S JUST THAT WE HAD TO WAIT ABOUT 30 YEARS OR MORE TO TAKE SERIOUSLY, WHAT THEY SAID AND THAT--THAT IS REALLY THE--I WILL SKIP OVER THIS. THAT'S REALLY THE PLACE THAT WE ARE NOW IS GRAPPLING AS THOMAS SHOWED YOU WITH THIS POLYGENIC COMPLEXITY. SOMETIMES PEOPLE ASK WHY DO YOU CARE ABOUT GENES OF LOW PENETRANTS AND GIVE ALL KINDS OF ANSWERS BUT THE TRUE ANSWER IS THAT'S WHAT WE GOT SO WE BETTER USE IT WISELY. THE LAST THING I WANT TO SAY IS WE ALSO HAVE TO THINK ABOUT--WE ALSO HAVE TO THINK ABOUT THE RECEPTION OF GENETICS BY SOCIETY ESPECIALLY AS WE BEGIN AS A GROUP TO COLLECT MORE GLOBALLY AND AS A RESULT TEND TO APPEAR IN THE GENERAL PRESS. MANY SOCIAL SCIENTISTS STILL DO NOT ACCEPT THE GENETIC INFLUENCE, THEY DO ON MENTAL ILLNESS, WHEN YOU GET SICK YOUR GENES ARE RIGHT THERE SCREWING YOU UP BUT THEY REALLY DON'T WANT TO ACCEPT THE IDEA THAT GENES DON'T HAVE SOMETHING TO DO WITH COGNITION AND BEHAVIOR AND I THINK WE HAVE AN IMPORTANT EDUCATION AND EFFORT TO UNIVERSITY STUDENTS AND OTHERS EVEN IF SOME OF OUR COLLEAGUES ARE BEYOND HELP. OFTEN GENES TO PEOPLE SUGGEST DETERMINISM AND IT'S REALLY HARD. I MEAN THIS IS A--I'M SURE ALL OF US IN SOME WAYS ARE BRISK FOR COMMON AND TURSKEY, AND PEOPLE HEAR GENES AND DETERMINEANISM AND IT'S REALLY IMPORTANT THAT WE EXPLAIN THIS PROPERLY TO PEOPLE. SOME ACADEMICS TRY TO FEND OFF DETERMINISM BY CLEANING TO UNDERPOWERED REDUCTIVE GENE BY ENVIRONMENT STUDIES, THESE--THESE ARE--THESE ARE CANDIDATE GENE STUDIES SQUARED EXCEPT THAT THE ENVIRONMENT IS AN INFINITE SPACE SO IF ANYTHING THEY ARE EVEN LESS INFORMATIVE AND WORSE. AND I THINK IMPERICALLY, MAYBE IT'S NOT THE PEOPLE IN THIS AUDIENCE BUT HOPEFULLY, WE WILL GAIN INFORMATION ABOUT THIS, ABOUT HOW PATIENTS FAMILIES AND SOCIETY INTERNALIZE AND THINK ABOUT GENETIC INFORMATION. DOES IT INCREASE OR DECREASE STIGMA. ARGUMENTS HAVE BEEN MADE BOTH WAYS. ON 1 HAND IT GETS PEOPLE OFF THE MORAL HOOK, ON THE OTHER HAND IT MAKES THEM SEE OTHER AND DIFFERENT AND AGAIN FOR THOSE WHO HAVE FALLEN TO DETERMINE THE TROAPS MAKE ITS SEEM HOPELESS. DOES GENETICS DECREASE SOMEBODY'S BLAME. WELL CERTAINLY THEY'RE NOT MORALLY RESPONSIBLE BUT MAYBE THEY'RE SEEN AS CURSED OR TAINTED IN SOME SOCIETIES AND MOST IMPORTANTLY, I THINK FOR PATIENTS IS DO GENETIC EXPLANATIONS, MAYBE THEY DECREASE SELF-BLAMING BUT THEY MAY ALSO UNDERCUT SOMEBODY'S SENSE OF AGENCY. AND THOSE ARE REALLY IMPORTANT ISSUES FOR US TO TAKE ON AND ADDRESS. AND WITH THAT LET ME WELCOME YOU AGAIN LEGALITY ME THANK YOU AND LET ME--LET ME THANK AND YOU LET ME ASK MARK MARK DALY, HE'S AT THE MGH AND THE BROAD INSTITUTE. MARK. >> THANK YOU. >> [ APPLAUSE ] ALL RIGHT, SO STEVE GAVE YOU THE BACKGROUND ON WHY MANY OF US ARE INVOLVED AND FOCUS ON GENETICS. SO I'M GOING TO BREEZE THROUGH A LOT OF THE POINTS THAT HE'S ALREADY SO--WELL ELOQUENTLY MADE THAN I WILL BE ABLE TO MAKE BUT I THINK IN PARTICULAR AS IT PERTAINS TO EVERYTHING ELSE THROUGH ANY OTHER MECHANISM HAS BEEN PROFOUND AND GREATER THAN ALMOST ANY OTHER MEDICAL AREA AND SO THIS GIVES US REALLY THE GREATEST EMPHASIS TO FOCUS AS A SOWRT OF THOSE FUNDAMENTAL BIOLOGICAL INSIGHTS AS LONG AS WE HAD DESCRIPTIONS OF THESE CONDITIONS. SO DESPITE THE HISTORICAL AND CONTINUED CHALLENGES IN DIAGNOSIS, COMMON PSYCHIATRIC DISEASES ARE EXTREMELY HERITABLE JUST SHY OF HEIGHT AND AT A FEW OTHER THINGS AND THIS IS A--THIS IS A REALLY REMARKABLE POINT BECAUSE IT MAKES IT CLEAR THAT WHILE THERE ARE CONTINUEUA, WE WILL ALWAYS SORT OF HAVE DISAGREEMENTS ABOUT THE CRITERIA WE USE TO CATEGORIZE PATIENTS AND MANY OF US DOUBT THE VALUE OF SPECIFIC CATEGORIZATION OF PATIENTS IN THE WAY THAT THE DSM DOES IT, WHAT WE HAVE IS MORE THAN ADEQUATE TO APPROACH GENE DISCOVERY AT SCALE BECAUSE WHATEVER IT IS THAT WE HAVE WHETHER YOU LIKE IT OR NOT IS EXTRAORDINARILY HERITABLE AND THAT'S WHAT WE NEED TO MOVE AHEAD WITH OUR GENETIC STUDIES SO SCHIZOPHRENIA, AUTISM, BIPOLAR ALL VERY MUCH NEAR THE TOP AS IT WAS IN STEVE'S SLIDES AND YOU KNOW REALLY MORE THAN 50 YEARS OF EPIDEMIOLOGY, IF NOT MORE HAVE BACKED UP THESE NUMBERS. THESE ARE NOT NEW MEMBERS BY ANY STRETCH OF THE IMAGINATION, THEY HAVE BEEN REMARKABLY CONSISTENTLY APPROVED ACROSS ALL POPULATIONS, DIFFERENT STUDY DESIGNS, TWIN STUDIES AND STUDIES OF ALL TYPES OF RELATIVES. OF COURSE AS YOU'VE ALREADY HEARD, THE WAY WE GREW UP THINK BEING GENETICS CONCEIVING OF THE SORT OF SINGLE GENES UNDER LYING SEVERE GENETIC DISORDERS WAS, YOU KNOW THAT WAS OUR MIND SET AND OUR TOOL KIT AS WE BEGONE TO APPROACH PSYCHIATRIC GENETICS 20, 30, 40, YEARS AGO DEPENDING ON HOW OLD YOU ARE AND IT WASN'T UP TO THE TASK BUT IT DOES PROVIDE THE UNDERPINNINGS AND THE RULES BY WHICH WE NEED TO APPROACH GENETICS, IT'S NOT LIKE WE'RE INVENTING NEW RULES HERE, WE JUST THINK TO THINK ABOUT HOW TO APPLY THEM A LITTLE MORE JUDICIOUSLY AND WE SORT OF FORCED OUR WAY THERE BECAUSE WHILE WE ALL WANTED TO IMAGINE THAT THEY WERE HIGH PENETRANTS, SINGLE GENES AND THEY TRAVEL PREDICTIVELY IN FAMILIES AND THAT WE GAIN BIOLOGICAL INSIGHT BECAUSE OF THEM, YOU KNOW QUITE SIMPLY THAT HASN'T HAPPENED WITH RESPECT TO MAJOR MENTAL ILLNESSES AND I THINK WE'RE STILL, GREATLY LACKING THESE TYPES OF OBSERVATIONS FOR VERY GOOD REASONS THAT WILL COME TO BUT THIS IS WHERE WE HAD THE MAJOR SUCCESS IN GENETICS AND WE APPLIED ALL OUR EFFORTS THROUGH THE 80S AND 90S IN AN ATTEMPT TO LEVERAGE WAS A REMARKABLE SUCCESSFUL PERIOD OF TIME IN HUMAN GENETICS RESPECT TO MENDELIAN DISORDERS AND AS STEVE GLOSSED BY VERY QUICKLY THIS ERA IN THE 1990S AWE A REMARKABLE EXPLOSION IN OUR ABILITY TO TACKLE AND REALLY BRING DOWN TO A ROUTINE, THE DISCOVERY OF THE GENES AND MUTATIONINGS URNTD LYING SINGLE GENE MENDELIAN DISORDERS AND DESPITE AN OVERWHELMINGLY GREATER INVEST NMENT THAT EFFORT IN COMPLEX DISEASES AT LARGE AND MENTAL ILLNESS PARTICULARLY AND WE HAD ESSENTIALLY NO SUCCESSES ACROSS THE BOARD IN COMPLEX AND COMMON DISEASES. SO WHY THIS WAS, WAS ACTUALLY PREDICTABLE IN ADVANCE BUT WE HAD TO GO THROUGH THE PROCESS AND LEARN IT FOR OURSELVES IMPERICALLY BECAUSE THE THEORY OF WHAT WE'RE NOW ENCOUNTERING HAS BEEN AROUND FOR A GOOD HUNDRED + YEARS. THE ABILITY TO ACT ON THAT AND TACKLE THE COMPLEXITY OF THE GENOME ON A SCALE OF PATIENTS HAS NOT EXISTED FOR A HUNDRED YEARS, THAT'S EXISTED FOR MAYBE ONLY 10 YEARS AND SO THAT'S WHY MORE WERE HAVING THIS AS A 10 YEAR ANNIVERSARY RATHER THAN A HUNDRED YEAR ANNIVERSARY, I GUESS, SO WE RECOGNIZED FOR A LONG TIME THAT MANY TRAITS, HERE'S GOLF AND AGAIN, ARE EXTREMELY HERITABLE AND REALLY VERY FEW MORE SO THAN ADULT STATURE, BUT QUITE OBVIOUSLY AND IT WAS OBVIOUS AT THE TIME AND IT WAS A PERIOD OF TIME WHEN THERE WAS A GREAT VEXATION BECAUSE CLEARICALLY THERE WAS NOT A SINGLE GENE FOR HIGHS, YOU COULD NOT GET AN 80-90% HERITABLE BUT PERFECTLY DISTRIBUTION NO MATTER HOW YOU ARRANGED THE GENO TYPES SO THERE WAS A DEBATE BETWEEN ELDERLY WHITE MEN WHO WANTED TO TRY TO FIGURE OUT AND YOU KNOW HOW THESE 2 THINGS COULD CO-EXIST AND MEIN, DEL'S LAW OF INHERITANCE AND THEN PLANTS AND MODEL SYSTEMS WITH THE CONTINUOUS PRESENTATION OF THINGS LIKE HEIGHT, OR MAYBE THINGS LIKE DISEASE RISK IN A POPULATION. AND SO, IT ACTUALLY TURNED OUT TO BE INFLUENTIAL PLANT EXPERIMENTS AND LATER FLY EXPERIMENTS FROM THE LABS AT COLUMBIA THAT BEGAN TO DEMONSTRATE WHAT WE NOW COME TO UNDERSTAND AS A POLYGENIC MODEL AND IF YOU TOOK 2 PLANTS OR 2 FLIES THEY MIGHT HAVE REMARKABLILY DIFFERENT FOON O TYPES IN 1 PARTICULAR AREA IN THIS CASE, COROLLA, LENGTH, AND WE'VE ALL SAW THESE MOUSE CROSSES AND SO FORTH AND YOU FOUND THAT THOSE 2 FOUNDING PLANTS THAT WERE HOMOZYGOUS IN LINES DIFFERED DRAMATICALLY IN THIS FENEE TYPE NOT BECAUSE OF A SINGLE GENE OR GENO TYPE BUT BECAUSE OF MYRIAD, SMALL ACTING GENES ACROSS THE GENOME AND YOU KNOW WE DIDN'T HAVE GENOMES OR GENES AT THE TIME BUT THE FUNDAMENTS OF THE POLYGENIC MODEL OF INHERITANTS REALLY CAME TO LIGHT IN THE EARLIEST PART OF THE LAST CENTURY. AND FISHER PUBLISHED A VERY INFLUENTIAL SYNTHESIS OF HOW YOU COULD HAVE CONTINUOUS OR EXTREMELY HERITABLE WITHOUT ANY VIOLATION OF MENDEL LAWS BY THE SIMPLE OBSERVATION THAT YOU COULD HAVE INSTEAD OF 1 MEN MEN--MENDELIAN, OR OTHERS BY A MILLIMETER OR HALF A MILL MITER OR RISK PROTECTION OF INFLUENCE AND DISEASE WITH A SMALL MEASURED MANNER. SO, OUR FAILURE TO UNDERSTAND TO DISCOVER GENES FOR COMPLEX DISEASES AND GENES FOR MENTAL ILLNESS IS NOT DIFFICULT TO UNDERSTAND AND ACTUALLY THIS IS--SHOULD HAVE BEEN REASONABLY WELL KNOWN BUT AS I SAID, WE HAD TO TRY IT ANYWAY. SIMPLY THE EXISTENCE OF MANY GENES RATHER THAN A SINGLE GENE WILL ELIMINATE THE EFFICIENCY OF LINKAGE ANALYSIS UNTIL YOU GET TO LARGE DENSELY EFFECTED FAMILIES THAT CAN BY THEMSELVES SUPPORT GENE DISCOVERY. UNFORTUNATELY WE DON'T HAVE THAT IN PSYCHIATRY. WE DON'T HAVE THOSE IN MANY COMPLEX DISEASES AND SO THAT BRINGS THE SECOND EVEN MORE COMPLICATING FACTOR OF INCOMPLETE AND LOW PENETRANTS RURAL RATHER THAN THE EXCEPTION AS PERTAINED TO COMMON AND COMPLEX TRAITS. SO OBVIOUSLY THE PROGRESS IN OUR FIELD REQUIRED MANY FUNDAMENTAL PARADIGM SHIFTS, SOME OF THEM TECHNICAL, SOME OF THEM JUST UNDERSTANDING THE PATTERNS OF VARIATION IN THE GENOME, I THINK THE TECHNICAL REALLY HAVE OUTWEIGHED EVERYTHING. AS I SAID, NONE OF THE IDEAS THAT WE'RE APPLYING AND REALLY VERY FEW OF THE IDEAS THAT YOU'LL HEAR ABOUT TODAY ARE NOVEL IDEAS OUR ABILITY TO EXECUTE ON THEM IS WHAT'S UNIQUE ABOUT THIS POINT IN TIME. WHICH MAKES ITS A GOOD TIME BUT WE WERE FACED WITH THE REALITY OF IF THERE WAS 1 GENE, THERE MIGHT BE 10S OF HUNDREDS OF GENES AND STEVE SAID IF THIS IS WHAT WE HAVE, WE HAVE TOW GO AFTER IT BECAUSE THAT'S THE ABOUT I ONY LOGICAL CLUES WHETHER WE LIKE IT OR NOT. I DON'T THINK WE DID LIKE IT BUT IT IS WHAT IT IS. SO GENOME WIDE ASSOCIATION STUDIES TO COMMON VARIANTS WITH A FIRST OF THE EFFORTS THAT WE'VE REALLY UNDERTAKEN IN EARNEST, PART OF THAT WAS TECHNOLOGICALLY DRIVEN AND PART OF IT WAS AIAN RATIONAL THEORETICAL SET OF OBSERVATIONS. AS WE BEGAN SEQUENCING THE GENOME AND DISCOVERING VARIATION IN THE HUMAN POPULATION AT LARGE SCALE IT WAS RECOGNIZED THAT AS POPULATIONS GENETICISTS HAD PREDICTED FOR DECADES IN ADVANCE OF THIS, THE VAST MAJORITY OF DIFFERENCES BETWEEN ANY 2 INDIVIDUALS, ANY 2 CHROMOSOMES, MY 2 CHROMOSOMES, 1 OF MINE, 1 OF YOURS ARE WRAPPED UP IN COMMON GENETIC VARIANCE, WRAPPED UP IN ANCESTRY 10S OF THOUSANDS OF YEARS AGO IN AFRICA AND THIS IS WHY COMPARABLY SMALL AMOUNTS IN THE HUMAN POPULATION THAT HAVE RISEN UNIQUELY IN EUROPEAN POPULATIONS OR UNIQUELY IN EASTAZZIAN POPULATIONS OUT OF AFRICA, BOTTLENECK. SO AS A CONSEQUENCE A SYNTHESIS MIGHT BE IF THIS IS A VAST MAJORITY OF GENETIC INDIVIDUALS IN A POPULATION, THIS MIGHT WELL BE THE SOURCE OF THE VAST MAJORITY OF PHENOTYPICALLY RELEVANT GENETIC VARIATION, NOT A LEADERSHIP BUT NOT A GUARANTEE. AS STEVE POINTED OUT THIS WAS NOT A NEW IDEA EVEN TO PSYCHIATRIC GENETICS WHERE IT'S PROBABLY THE AMONG THE MOST DIFFICULT OF FOON O TYPES TO THINK OF IN AN ALL--WE THINK OF IT TRADITIONALLY IN AN ALL OR NOTHING, YOU HAVE SCHIZOPHRENIA OR YOU DON'T. THE FACT THAT GENETICS PROMOTIONAL SUBJECT WORKING LIKE THAT AND THE FACT THAT WE'RE LIMITED IN OUR ABILITY TO SORT OF IMAGINE CONTINUOUS BIOLOGICAL VARIATION UNDERNEATH THE HOOD AS IT WERE, THAT IT'S DRIVING AT, SHOULDN'T BE A--YOU KNOW A FREE TICKET TO GET AROUND WHAT ARE COMPELLING OBSERVATIONS OF HERITABILITY AND FREQUENCY AND NATURAL SELECTION AND WHAT GOTTESMAN AND SHIELDS OBSERVED WAS THE EXTREMELY HIGH HERITABILITY THAT WE ALL OBSERVED TODAY ACROSS CULTURES, BACKGROUNDS, RELATIVE HAIR TYPES AND WHAT THEY RECOGNIZED BASED ON VERY RECENT SORT OF WORK EXTENDING TRADITIONAL QUANTITATIVE HERITABILITY MODELS WAS THAT, YOU KNOW A VIABLE MODEL AND REALLY THE 1 THAT FIT THE DATA BEST GIVEN THE FREQUENCY AND THE HERITABILITY THAT WE'RE LOOKING AT IS THAT THE VAST MAJORITY OF THAT HERITABILITY IS WRAPPED UP IN A CONSTITUTIONAL PREDISPOSITION OR LIABILITY TO DEVELOP SCHIZOPHRENIA, AND THAT THERE MIGHT BE OTHER EVENTS AND ENVIRONMENTAL FACTORS THAT PUSH YOU 1 WAY OR ANOTHER, BUT THAT CONSTITUTIONAL INHERITANCE IF DISTRIBUTED OVER MANY, MANY GENES WOULD INFACT EXPLAIN ALL THE EPIDEMIOLOGIC OBSERVATIONS. STEVE ALSO POINTED OUT, I DON'T THINK ANY OF US OR ANYBODY PAID MUCH ATTENTION TO THIS AND CERTAINLY WE DIDN'T MUCH HEAT OF IT IN THE 80S AND 90S AS THE MENDELIAN GOLDEN AGE TOOK OVER ALL OF GENETICS BUT AS IT TURNS OUT THIS ULTIMATELY COMES TO PASS AS BEING THE SOURCE OF MOST OF THE HERITABILITY OF COMPLEX TRAITS ACROSS THE BOARD SO EVERYONE RECOGNIZES THAT NOW THE SUCCESS OF GENOME WIDE ASSOCIATION STUDIES AS APPLIED TO COMPLEX TRAITS THAT IT WASN'T SUCH A LONG TIME AGO THAT THE FIRST AND PSYCHIATRY ACROSS THE BOARD WERE PRETTY, YOU KNOW UNIMPRESSIVE OR DEPRESSING DEPENDING ON WHERE YOU WERE COMING FROM. AND IN FACT, FAIRLY SIZABLE STUDIES OF A SIZE THAT HAD BEEN SUCCESSFUL FOR EXAMPLE, IN IMMUNE MEDIATED DISEASES, DID NOT REVEAL ANY NEW INSIGHTS INTO THE BIOLOGY OF SCHIZOPHRENIA, AND SO THIS THEN RE-RAISED THE THIRD OF THE SORT OF MAJOR EPIDEMIOLOGIC OBSERVATIONS THAT PLAY INTO THE GENETIC ARCHITECTURE OF A TRAIT AND THAT BEING NATURAL SELECTION AND WHAT MIGHT BE DIFFERENT IF WE WERE THINKING ABOUT THE HIGHLY HERITABLE COMMON DISEASES LIKE DIABETES, LIKE SEVERAL IMMUNE MEDIATED DISEASES, THAT GOTTESMAN AND SHIELDS CITED AS QUITE SIMILAR TO WHATEE SEE IN SCHIZOPHRENIA WHY THOSE MAY DIFFER IS OF COURSE THAT THERE'S A VERY STRONG AND SIGNIFICANT AND RELIABLE, THIS HAS BEEN REPLICATED ACROSS, MANY, MANY DIFFERENT COUNTRIES AS WELL. REDUCES FITNESS DIAGNOSIS OF SCHIZOPHRENIA OR AUTISM IN PARTICULAR. THESE 2 STAND OUT AS THEY DO WITH AN ALMOST IDENTICAL HIGH INHERITABILITY AND PREVALENCE, THEY ALSO STAND OUT WITH A REDUCED FITNESS IN MALES WITH THIS DIAGNOSIS HAVE 1 QUARTER AS MANY OFFSPRING, FEMALES HAVE HALF AS MANY OFFSPRING SO REMARKABLE EPIDEMIOLOGIC OBSERVATIONS ACROSS THE BOARD THAT MIGHT CONNECT AUTISM AND SCHIZOPHRENIA IN A WAY THAT WE MAY NOT HAVE APPRECIATED EARLIER. BUT THE RAMIFICATIONS OF THIS ARE QUITE SUBSTANTIAL WITH RESPECT TO WHAT WE CAN FIND GENETICALLY. AND WITHOUT GOING INTO THE MATH, WHAT THESE--THESE TELL US IS NOT SIMPLY THAT YOU CAN'T HAVE COMMON VARIATION BECAUSE THERE'S SELECTION AGAINST IT, IT PUTS A CAPOT EFFECT SIZE AS THE FREQUENCY AS A FUNCTION OF THE EFFECT SIZE, SO WHEN THE MUTATION ARISES IN THE POPULATION AND CONFERS A CERTAIN RISK, IT WILL NOW BE CAPPED AS TO HOW FREQUENT IS EXPECTEDDED TO GET OR CAN GET AT ALL. SO AS IT TURNS OUT, THE CAP THAT IT PUTS IS ACTUALLY WORSE IN THESE INTERMEDIATE FREQUENCY RANGES AT VERY HIGH FREQUENCIES, CAN YOU STILL ACHIEVE AALLOWED TOS THAT CONFER RISK FOR THE DISEASE BUT THEY HAVE TO AN EXTREMELY WEAK EFFECT ON RISK, SO 1.1, SO 10% IN. CREASED RISK FOR ALLELE, MAYBE A LITTLE BIT MORE, MOST OF THEM A LITTLE BIT LESS THAN THAT. WHERE GENETICS HAS MADE TREMENDOUS PROGRESS IN REVEALING REASONABLY IMPACTFUL AND STILL REASONABLY COMMON ALLELES IN CARDIOVASCULAR DISEASE, ALZHEIMER'S MEDIATED IMMUNE DISOAZS HAS BEEN IN THE INTERMEDIATE RANGE WHERE ALLELES OF SAY EFFECTS, ODDS RATIOS OF 2-10 RISE TO A CERTAIN PREQUENCY AND GET PICKED UP AND PROVIDE FUNDAMENTAL INSIGHTS. THESE ARE SO PUNISHED BY THAT FORCE OF NATURAL SELECTION, HAVING ODDS RATIOS IN IN RANGE THAT THEY ALMOST NEVER ACHIEVE FREQUENCIES OF 1 IN 10,000, SO IT'S NOT THAT THEY DON'T EXIST, IT'S JUST THAT WHEN THEY ARISE IN THE POPULATION, THEY'RE NOT PERMITTED TO GET TO A FREQUENCY THAT OUR SAMPLE SIZES REMOTELY CAN ACHIEVE AND THIS GETS WORSE AND WORSE AS YOU GO TO HIGHER AND HIGHER IMPACT VARIANTS. TO THE POINT WHERE AT THE OPPOSITE END OF THE SPECTRUM, THE ONLY THING WE CAN FIND ARE HIGH ARISING ALLELES THAT ARE DE NOVO AND WE STARTED TO LEARN THIS 10 YEARS AGO AT THE TIME WE WERE BEGINNING WITH GWAS WITH HIGH IMPACT AND FREQUENTLY RECURRENT COPY NUMBER OF VARIANTS. SO THE CHALLENGE THOUGH WAS AS WE COLLECTED THOSE GWAS DATA SETS' WE RECOGNIZE THE MATHEMATICS BEHIND ALL THIS AND HOW IT MIGHT FIT TOGETHER IT WAS SOMEWHAT DAUNTING OR DEPRESSING AGAIN, DEPENDING ON YOUR VIEW POINT. THE SCALE THAT WOULD BE REQUIRED TO GAIN ACCESS TO THE INDIVIDUAL BIOLOGICAL AND GENETIC ELEMENTS IN A POLYGENIC MODEL REQUIRED THAT WE MADE A FULL COMMITMENT TO TEAM SCIENCE, THE ONLY WAY WE WOULD ACHIEVE THE SAMPLE SIZES REQUIRED, THE ONLY WAY WE WOULD ACHIEVE THE RESOURCES REQUIRED WOULD BE TO TACKLE THIS AS A COMMUNITY AND SO IT'S BEEN ENORMOUSLY SATISFYING AND EXCITING TO SEE THE PROGRESS OF THINGS LIKE THE PGC, YOU WILL HEAR ABOUT, I'LL SHOW YOU IN A FEW MINUTES BUT I THINK WE ALL OWE AN ENORMOUS DEBT OF GRATITUDE TO MANY INDIVIDUALS AND THOMAS IN PARTICULAR FOR HIS COMMITMENT AND ADVOCACY TO BELIEVING THAT THIS WAS GOING TO BE A PRODUCTIVE ACTIVITY HELPING US PAVE THE WAY FOR TREMENDOUS PROGRESS IN THE PGC THAT NICK WILL TALK ABOUT MOVING SCHIZOPHRENIA FROM THAT FIRST SAD LOT WITH 0 GENES DISCOVERED TO A REMARKABLILY FLOR FLOWER EDUCATIONAL GARDEN OF GENETIC ASSOCIATIONS EACH OF WHICH HARBORS BIOLOGIC CLUES AND IN NO SENSE WAS A VIOLATION OF ANY OF THOSE PRINCIPLES THAT I DESCRIBED EARLIER WITH RESPECT TO FREQUENCY PREVALENCE, EFFECT SIZE, GIVEN THE SELECTION, THE DIRECT SELECTION AGAINST THE TRAITS OF SCHIZOPHRENIA AND PSYCHIATRIC DISEASES WAS ANY OF THAT VIOLATED AND IN FACT, IT SIMPLY TOOK OFF BECAUSE WE FINALLY REACHED THE SAMPLE SIZE THAT ENWAIBLED US TO FIND--ENABLED US TO FIND ONLY THE VERY, VERY MODEST EFFECTS THAT NATURAL SELECTION IS LEAVING US WITH. SO, TREMENDOUS NUMBERS OF BIOLOGICAL CLUES FROM THESE STUDIES ARE BEGINNING TO EMERGE. IT SIMPLY REQUIRED A SCALE AND COMMITMENT FROM MANY IN THIS AUDIENCE TO BRING THEM TOGETHER. A NUMBER OF OTHER ENERNLGING LESSONS THATIME SURE YOU WILL HEAR SOME OF THESE THROUGHOUT THE DAY, SO I WILL NOT GO THROUGH THEM IN ANY GREAT DETAIL HERE, BUT THERE ARE ALSO MANY, MANY CHALLENGES AND IT'S NOT BEEN EASY AT ALL TO MOVE FROM THE EXCITING LOOKING PLOTS HERE, TO ACTUALLY FUNDAMENTAL BIOLOGICAL INSIGHTS. THAT'S I THINK THE CHALLENGE OF THE NEXT 10 YEARS THAT WE REALLY HAVE TO INVEST IN AND I THINK A LOT OF WHAT YOU WILL HEAR DURING THE DAY TODAY WILL BE FOCUSED ON THAT. SO, OBVIOUSLY WE CAN'T BE SATISFIED WITH JUST DISCOVERING THAT THERE IS APOLOGY GETTIC MODEL AND EXEXPLAINS HERITABILITY AND WE HAVE TO PRESS ON TOWARDS THOSE GOALS AS WAS FORETOLD YEARS AGO. THERE'S ALSO AN EQUALLY EXCITING SET OF WORK GOING ON TO PROBE THE SAME POLYGENIC MODEL WITH RESPECT TO RARE VARIANCE AND AND YOU WILL HEAR A LOT ABOUT THAT TODAY AND IN PARTICULAR BASED OFFICE OF DIVERSITY THIS BENEFICIAL EXCEPTION DE NOVO MUTATIONS WHICH ESCAPE THE FORCES OF NATURAL SELECTION BECAUSE THEY HAVEN'T YET HAD--NATURAL SELECTION HASN'T YET HAD 1 WHACK AT THEM, GROUPS SUCH AS THE SEQUENCING CONSORTIUM THAT JOE BUCK WILL TELL YOU ABOUT LATER ON HAVE MADE SIMILAR COMPILING THAT ULTRA RARE COMPONENT AND THAT IS NOT EXPLAINING THE HERITABILITY, EXPLAINING THAT 80% COMOPPOSITE BEHAVIORIAL PHENOTYPENT AND STILL PROVIDES GENETIC INSIGHTS THAT WE CAN ULTIMATELY BRING TOGETHER IN THAT WAY. SO THANK YOU FOR THAT AND THANK YOU THOMAS FOR BRINGING US ALTOGETHER AND YOU AND YOUR COLLEAGUES, THE LATE PAT BENDER, ANGIE AND SO MANY OTHERS FOR MAKING THIS DAY AND THE PROGRESS IN OUR FIELD POSSIBLE. [ APPLAUSE ] >> I WOULD RESOLVE THAT MEDICAL SCHOOLS SHOULD STOP HAVING GENOMICS TAUGHT BY OFF THE RYAN FARMER MONKS AND JUST HAVE A VIDEO OF MARK EXPLAINING WHAT HE JUST SAID AND THAT WOULD HELP A LOT OF THINGS. THE NEXT SPEAKER BENJAMIN NEALE, IS A POPULATION SPECIALIST AND BACKGROUND IN MATHEMATICS AND IS ALLYING THIS TO COMPLEX BIOLOGY. >> THANK YOU. IT REALLY IS A PLEASURE TO COME HERE TODAY AND SPEAK AT THIS SYMPOSIUM. REALLY THOMAS HAS DONE MORE FOR MY CAREER IN SCIENCE THAN ALMOST ANYONE ELSE AND HAS BEEN A GREAT SUPPORTER AND I THINK A LOT OF THAT SUPPORT HAS REALLY BEEN A CONSEQUENCE OF SOME OF THE PECULIAR WAYS THAT I TRY AND LOOK AT THE WORLD AND THINK ABOUT HOW WE USE MATH AND DATA AND KIND OF DATA SCIENCE AND STATISTICS TO GET TO THE BOTTOM OF WHAT'S ACTUALLY CAUSING THESE DISEASES AND THINK ABOUT UNDERSTANDING THE BIOLOGY AND THE BASIS OF THOSE THINGS AND FOR THAT I JUST WANT TO SAY THANK YOU THOMAS. STEVE'S ALSO BEEN A CHEERLEADER OF LATE SINCE HE'S TAKEN OVER AT THE STANLEY CENTER. I WOULD BE REMISS TO THANK HIM AS WELL. AND THINK ABOUT THE PRINCIPLES OF EXPERIMENTAL DESIGN AND THE IDEAS AROUND STATISTICS AND THE WAY WE THINK ABOUT DATA AND ANALYSIS THAT REALRY GET US TO THAT KIND OF RIGOROUS FIRM GROUPING IN GENETICS THAT WE REALLY NEED IS THE PLATFORM TO START TO EXPLORE THE BIOLOGICAL BASIS OF DISEASE. SO, WHEN WE THINK ABOUT GENETICS AND WHY WE'RE DOING THESE THINGS, REALLY FINDING THOSE GENES THAT DRIVE DISEASE, THAT'S REALLY CRITICAL, GENETICS OFFERS US A LOT OF STUDENT TO UNDERSTAND THE PHENOTYPIC STRUCTURE OF THESE NEUROPSYCHIATRIC DISEASES TO KIND OF TAKE AIM AT THAT DICHOTOMY OR LACK THEREOF. I THINK OTHERS HAVE WRITTEN EXTENSIVELY ABOUT THAT. UNDERSTANDING HOW WE PUT THESE DIFFERENT GENETIC EFFECTS INTO A MORE COHERENT VIEW OF BIOLOGY AND THEN TAKE THE NEXT STEP FORWARD IN TAKING GENETICS TO EXPERIMENTAL SYSTEMS AND HOW WE THINK ABOUT LEVERAGING THINGS LIKE IPS TECHNOLOGIES AND JUST TAKE A CELLULAR AND EVENTUALLY MAYBE PHYSIOLOGICAL VIEW OF HOW GENETIC VARIATION TRANSLATES INTO BIOLOGICAL ACTIVITY. AND I THINK ONLY THROUGH THOSE STEPS WILL ACTUALLY START TO UNDERSTAND WHAT'S REALLY HAPPENING UNDER THE HOOD WHEN WE THINK ABOUT THESE MENTAL ILLNESSES. NOW I THINK FOLLOWING ALONG FROM A LOT OF WHAT STEVE AND MARK WERE TALKING ABOUT, I WILL TAKE US BACK IN TIME TO MORE OR LESS WHEN I WAS STARTING OUT IN GENETICS. THIS WAS THE VERY EARLY DAYS, I WAS IN RICHMOND VIRGINIA WORKING IN PAT SULLIVAN'S LAB AS A RESEARCH TECH HAVING DROPPED OUT OF COLLEGE. AND 1 OF THE FIRST THINGS THAT PAT AND KEN HAD ME DO WAS RUN A LINKAGE SCAN THAT WAS SIMILARLY UNSUCCESSFUL AS MARK MIGHT HAVE PREDICTED BUT THEN WE DID AN ATTEMPT TO REPLICATE A CABBEDIDATE JEEP STUDY AND I THINK IT WAS REPLICATING THE REG LYNN ASSOCIATION BY DECODE AND THE FIELD BACK THEN WAS CHARACTERIZED BY PICKING A GENE OR A HANDFUL AND THEN TYPING A HANDFUL OF MARKERS FROM THAT KIND OF GIVEN GENE AND THEN YOU TEST EACH 1, 1 BY 1 AND YOU LOOK FOR A P-VALUE OF 0.5 AND YOU GIVEN GENES AND MARKERS YOU WERE ALMOST GUARANTEED TO FIND IT AND THEN YOU WRITE A PAPER AND THEN YOU GO, YAY, I WROTE A PAPER AND THEN EVERYBODY WOULD GO, HEY WE'RE WE'RE EXCITED ABOUT THIS GENE AND BECAUSE WE WROTE A PAPER AND PEOPLE ACCEPTED IT AND IT WAS ALL WELL AND GOOD AND DANDY. AND THEN IN 2002 AND HIS DAD WROTE THIS LOVELY REVIEW PAPER TAKING HAIM OF THIS CANDIDATE ASSOCIATION, AND GOING AFTER THAT ASSOCIATION AND WHAT THEY LOOKED AT WAS 600 SUCH PUBLICATIONS. THIS WAS THE GENETICS LITERATURE FOR COMPLEX TRAITS AT THAT TIME AND AND OF THE 600 MOST HAD TRIED TO BE REPLICATED SO THERE WAS NONE OF THIS REPRODUCIBILITY OF SCIENCE AND DATA AGGREGATION AND SHARING AND NONE OF THE CONCEPTS TO PERCOLATE THROUGH AND GENETICS AND OF THE 166 THERE WERE AT LEAST 3 REPORTS, 6 LOOKED RELIABLE. OKAY? SO THAT'S NOT A STABLE FOUNDATION FOR FOLLOW UP WORK. THAT IS NOT WHAT I WOULD WANT TO DO. IF I WERE A POST DOC WORKING IN BIOLOGICAL CHARACTERIZATION OF A GENE OR OUTSIDE OF GENETIC VARIANCE, I WOULD NOT WANT TO SPEND MY TIME ON THE 160 OUT OF THE 166 THAT WERE TESTED MULTIPLE TIME IN ANY RIGOROUS WAY. OTHERS PUBLISHED ON THIS, JOHNNY AND ALMOST MADE A CAREER OF TRYING TO ADDRESS THIS QUESTION OF REPRODUCIBILITY IN SCIENCE. THESE IDEAS OF HOW DO WE CONVINCE OURSELVES SOMETHING IS TRUE VERSUS SOMETHING NOT BEING TRUE AND I THINK MOST OF THE GENETIC CYSTS OF THE TIME, THIS IS WHAT THEY WERE DOING, THIS IS WHAT WE DID AS GENETICISTS AND THEN THAT JUST DIDN'T SEEM TO WORK. SO, WHAT--WHAT WENT WRONG, WELL, 1 OF THE THINGS THAT PROBABLY WENT MOST WRONG WAS THE MULTIPLE TESTING ISSUE AND THIS IDEA OF HOW WE HAVE TO BE HONEST WITH OURSELVES ABOUT WHAT WE'RE DOING WHEN WE CONDUCT GENETIC EXPERIMENTS. SO THIS WHAT YOU WOULD GET FOR A INTRIEWKS OF NULL P-VALUES SO THIS IS JUST THE SIMULATION OF A MILLION INDEPENDENT TESTS SIMULATED UNDER THE NULL NOT TOO DISTIMULAR THIS FROM WHAT YOU WOULD HAVE DOING A GWAS AND UNDER THE NULL THEY FOLLOW THIS NICE UNIFORM INTRIEWKS BUT WHAT HAPPENS IF I LOAD THE DICE AND PICK THE WINNER FROM SOMETHING WHERE I DO A HUNDRED TESTS AND IF I DO A HUNDRED TESTS AND I PICK THE BEST 1, THIS IS THE DISTRIBUTION THAT I GET, RIGHT? AND IF I PUT DOWN MY P-VALUE OF .05, BASICALLY EVERY EXPERIMENT WAS GUARANTEED TO GET AT LEAST A P-VALUE OF .05 AND THIS IS WHAT HAPPENED IN THE CANDIDATE GENE ASSOCIATION ERA THAT WE WOULD RUN THESE EXPERIMENTS NOT BE HONEST WITH OURSELVES ABOUT THE NUMBER OF TESTS WE WERE DOING AND OTHER PROBLEMS IN TERMS OF CONTROLLING FOR COMPOUNDING AND BIAS AND THINGSLY LEARNED SUBSEQUENTLY BUT REALLY THIS SORT OF MULTIPLE TESTING ISSUE AND THIS IDEA THAT IF YOU DON'T ACTUALLY HONESTLY LOOK AT WHAT YOU'RE EXPLORING SCIENTIFICALLY, IT'S DIFFICULT TO TELL WHAT YOU SHOULD EXPECT TO SEE AND SO WE WOULD SEE THIS BIAS, THIS GUARANTEE OF GETTING A .05. AND HOW WE KIND OF TURN TO GWAS, SO RIGHT? WHAT ABOUT GWAS, HOW DO WE THINK ABOUT GENOME WIDE ASSOCIATION IN THIS SORT OF CONTEXT, HOW DO WE INTEGRATE THOSE SORTS OF MULTIPLE TESTING ISSUES, WELL WE HAVE, MAYBE HALF A MILLION MARKERS ON THE ARRAY, WE CAN USE IMPUTATION TO FILL IN THE COMMON VARIATION SCATTERED ACROSS THE GENOME. IT'S A REALLY BIG MULTIPLE TESTING PROBLEM AND IT GETS WORSE BECAUSE WE CAN DO THINGS LIKE TEST FOR DOMINANCE, SO& MAYBE THERE'S LIKE ADDITIVE EFFECTS AND NONADDITIVE EFFECTS TO GO BACK TO THE WELCOME TRUST CONTROL PAPER FOR NOW A DECADE AGO, THEY TESTED THE 2-DEGREE FREEDOM MODEL AND THE 1 THING THEY CLAIM THAT DIDN'T REPLICATE WAS ACTUALLY THE DOMINANCE ASSOCIATION, THAT MIGHT BE A LESSON IN THERE, BUT SOMETHING TO THINK ABOUT. WE CAN THINK ABOUT TESTING FOR GBACE, CONDITIONING ON THINGS LIKE SMOKING OR OTHER RISK FACTORS THAT ARE RUNNING AROUND THE POPULATION OR DOING SUBGROUP ANALYSIS AND KIND OF ON AND ON AND ON AND ON. AND I THINK THAT THESE SORTS OF THINGS ARE NATURAL DESIRES IN TERMS OF EXPLORATION OF DATA, THEY ALSO POSE PROBLEMS WITH RESPECT TO HOW DO WE MAKE SURE THAT WE CAN MAINTAIN THESE SAME LEVELS OF STRINCHENCY. WHERE FOR THE VANILLA GWAS PROBLEM, THE KIND OF VERY STRAIGHT FORWARD APPROACH IS THE IDEA OF DISCIPLINE, SO IT'S 1 I WROTE WITH MARK AND DAVID AND ROM AN AND AT EXACTLY THE SAME TIME IN GENETIC EPITEEMIOLOGY ERIC AND OTHERS WROTE A PAPER AND A DID A VERY SIMPLE THING, THEY TOOK A GWAS AND THEY TRIED TO FOR OUT HOW TO SET THAT GWAS THRESHOLD AND THE WAY THEY DID THAT WAS THEY ESTIMATED THE NUMBER OF TESTS THEY WERE GOING TO BE PERFORMING. NOW THE GENOME'S A COMPLICATED PLACE, RIGHT? SO THERE ARE--YOU KNOW MAYBE 30 MILLION COMMON VARIANTS, SCATTERED ACROSS HUMAN POPULATIONS BUT THESE DIFFERENT VARIANTS ARE PRETTY CORRELATED SO THAT DEGREE OF CORRELATION MEANS THERE'S A LOT OF REDUNDANCY IN THE ASSOCIATION TEOF THING SO TODAY WE'RE MORE SOPHISTICATED ABOUT THAT BUT THEY BASICALLY ESTIMATED THAT IF YOU DO A GWAS, YOU'RE DOING A MILLION INDEPENDENT TESTS AND YOU ARE GUARANTEED TO SEE A WHOLE BUNCH OF .05S IF YOU DO A MILLION INDEPENDENT TESTS. TAKEN--THEY'S PRETTY MUCH A GUARANTEE. SO WHAT WE NEED TO DO IS WE NEED TO TAKE THIS .05 AND APPLY THE CORRECTION AND BE DISCIPLINED ABOUT THIS MULTIPLE TESTING AND UNDERSTAND WHAT WE'RE DOING WELL AND THEN IF YOU DO THAT YOU END UP WITH OUR 5 TIMES 10 TO THE MINUS 8 AND THAT'S OUR GENOME WIDE SIGNIFICANCE AND THIS SAME LAW CAN BE EXTENDED NOT ONLY TO THE TESTING OF COMMON VARIATION, BUT RARE VARIATION AND YOU CAN DO THE SAME THING FOR TESTING GENE BURDEN AND EXOME SEQUENCING AND CORRECTING FOR THE NUMBER OF GENES AND AGAIN WITH WE THINK ABOUT THOSE RARE VARIANT ASSOCIATION TESTS WE HAVE TO THINK ABOUT THE FACT THAT WE MIGHT TEST LOSS OF FUNCTION MUTATIONS AND TEST MUTATIONS AND WE MIGHT TAKE A LOT OF DIFFERENT CRACKS AT THESE KINDS OF ASSOCIATIONS SO WE SHOULD BE AGAIN EXPLICIT ABOUT THE KINDS OF TESTS AND APPROACHES WE'RE TAKING. THOU OTHER APPROACH IS TO SAY THIS WHOLE GENOME WIDE SIGNIFICANCE THING IS ATRINCHENT AND IT'S REALLY HARD TO GET TO AND WHAT I REALLY HEAR THEM SUBTEXTURALLY SAYING IS I HAVE A HARD TIME WRITING PAPERS BECAUSE I CAN'T CLAIM SPLASHY RESULTS THAT'S HOW IT SOUNDS TO ME. THAT MIGHT BE A LITTLE UNGENEROUS BUT THAT'S KIND OF WHAT I HEAR IS LIKE HOW DO I WRITE A PAPER ABOUT A NULL ANALYSIS AND IF YOU GO BACK AND LOOK AT WHAT I STARTED WITH, IT WAS THE EARLIEST GWAS OF ADHD AND KIND OF NEGATIVE CANDIDATE GENE ASSOCIATIONS THESE WERE ALL NEGATIVE PAPERS AND THEY WERE NOT IN GOOD JOURNALS. IT WAS NOT A SUCCESSFUL WAY TO GET SPLASHY RESULTS AND GET A LOT OF NAME PREC NITION OUT THERE BUT WE SHOULD VALUE THAT AS A COMMUNITY MORE THAN WE DO. WE SHOULD TAKE A VIEW THAT THE WELL CRAFTED ANALYSIS AND A WELL COLLECTED DATA SET IS THE VALUABLE CONTRIBUTION TO THE COMMUNITY, NOT NECESSARILY THIS DISCOVERY OF A LOCUST THAT WE ALMOST PERHAPS FETISHIZED AS A COMMUNITY BUT NEVERTHELESS, PEOPLE WANT SOME KIND OF 0 AND THEY WANT SOMETHING THEY CAN TALK ABOUT AND WRITE ABOUT AND THEY WILL TAKE THIS P-VALUE INTRIEWKS OF A MILLION NULL TESTS AND THEN MAYBE DOSE IN SOMETHING LIKE 10,000 TRUE EFFECTS OVER HERE IN ON THE RIGHT HAND SIDE SO THESE ARE NOW TRUE ASSOCIATIONS THAT WE'RE NOT SUPER WELL POWERED TO DETECT BUT WE HAVE SOME INDICATION AND IF YOU PUT THESE TOGETHER, YOU GET THIS BULGE, SO WE GET THIS LITTLE BUMP AND THEN WE CAN BE CLEVER AND DRAW THIS LINE AND SEE WHERE THIS DEVIATION IS FROM THIS PARTICULAR BUMP AND THIS IS WHAT FALSE DISCOVERY APPROACHES ARE TRYING TO DO. THEY ARE TRYING TO 0 IN ON THE SET OF THINGS THAT ARE, YOU KNOW MAYBE IN THIS BULGE. THE CATCH IS, IN THIS SIGNAL, SO FOR THE THINGS THAT ARE 10 TO THE MINUS 3 OR 10-44 SCHIZOPHRENIA RIGHT NOW,A I BUNCH OF THEM ARE REAL AND A BUNCH OF THEM AREN'T AND WE REALLY CANNOT TELL THE DIFFERENCE BETWEEN THOSE THAT ARE REAL AND THOSE THAT ARE NOT AT THIS JURCHGZ. WE CAN BE MORE CONFIDENT AT THE THINGS THAT ARE VERY WELL ESTABLISHED, THEY ARE STRONGLY ASSOCIATED BUT THOSE THINGS THAT ARE IN THAT BUBBLE ZONE THAT ARE A MIXTURE OF TRUE AND NOT TRUE EFFECTS, WE CAN'T TELL WHICH ARE TRUE AND WHICH AREN'T SO BE SUSPICIOUS WHEN PEOPLE CLAIM SOMETHING AS BEING UNEQUIVICALLY TRUE TO SAY 10 TO THE MINUS 3 OR 10 TO THE MINUS 4 BECAUSE OF THAT CERTAINTY. SO WE GO TO CUE PLOTS AND HISTOGRAMS AND WE HAVE THE MINUS 10 LOG P-VALUE ANDEE ON THE Y-AXIS WE HAVE WHAT WE OBSERVE BASED ON EXPECTATION IF YOU DRAW FROM THE NULL AND YOU DO A FLAT GWAS, YOU WILL GET THE INTRIEWKS AND--INTRIEWKS AND--DISTRIBUTION AND YOU VISUALIZE THAT AND THE IDEA BEHIND FALSE DISCOVERY IS THAT WHEN YOU HAVE A MIXTURE OF TRUE AND NOT TRUE EFFECTS, YOU GET THINGS THAT START TO SKIP ABOVE THESE LINES AND WE CAN CLAIM SOME OF THEM AS TRUE AS A GIVEN FALSE DISCOVERY RATE. NOW WHEN WE CLAIM THESE AS TRUE AS A FALSE DISCOVERY RATE, GIVEN THE SET OF EFFECTS SOME MIXTURE OF THEM ARE REAL AND SOME ARE NOT REAL. AND WE CAN ESTIMATE WHAT THAT MIXING PROPORTION IS BUT IT DOESN'T GUARANTY THAT BASICALLY ALL OF THEM WERE REAL. THEY WERE NEVER REALLY, REALLY GUARANTEED BUT WE CAN BE PRETTY CERTAIN. SO THIS--YOU KNOW THERE'S THAT ASSUMPTION AND THE OTHER ARKTS SUSMGZ IS THAT OUR OVER ALL INTRIEWKS IS WELLICAL BLAITED SO IF YOU HAVE A BIT OF BIAS, A BIT OF POPULATION STRATIFICATION THAT SCREWS UP THE FALSE DISCOVERY RATE APPROACHES BECAUSE THE WHOLE DISTRIBUTION IS BIASED AND THAT PUSHES EVERYTHING JUST A LITTLE BIT MORE. SO WHAT HAPPENS UNDER THE NULL? I GOT THE SAME DATA SET CONTINUOUSLY ANALYZED IN IN THE PICTURE SO HERE IS THE NULL MARKER AND THIS IS A RANDOM NULL LOCK IN TERMS OF THE ASSOCIATION AS WE SAY INCREASE FROM A THOUSAND PEOPLE TO A THIS HAPPENED THOUSAND PEOPLE AND HERE'S WHAT HAPPENS UNDER THE ALTERNATIVE. SO WE SEE THE WALK, BUT THE ASSOCIATION TEST, THE K-SQUARE SENDS TO GROW. SO UNDER EXPECTATION IF YOU DOUBLE THE SAMPLE SIZE, YOU EXPECT TO DOUBLE THE SQUARE, SO THAT GIVES US CONFIELDFIDENCE ABOUT THE KIND OF TRUTH AND VORACITY OF THE GWAS IS THAT IMPERICALLY THAT'S WHAT WE SAW. SO INCREASING SASM EMP SIZES, WILLING TO INCREASE THE SQUARE, AND SMALL SAMPLE SIZES BY THE VERY NATURE ARE UNTRUSTWORTHY. THERE'S A BASIAN DECODERRIAN ARGUMENT ABOUT HOW MUCH INFORMATION THERE IS IN A SMALL SAMPLE SIZE AND WHY THAT SHOULDN'T BE NECESSARILY RELIABLE REVIEWED IN THE ASSOCIATION. SO THE RETURN TO THE GWAS THAT REALLY SET THE BAR FOR THIS AT LEAST IN PSYCHEIATIC GENETICS YOU CAN THINK ABOUT EACH OF THESE DIFFERENT ASSOCIATIONS, MANHATTAN PLOT, EACH VARIANT TESTED IN THE MINUS LOG 10-PVALUEOT YAXIS, AND YOU ADD EFFORTS THROUGH THE PGC, THAT NICK WILL TALK TO YOU ABOUT AND MARK HIGHLIGHTED YOU SEE THESE POWERS EMERGE AND THEY TEND TO MOVE IN 1 DIRECTION. SO WE SEE THIS ASSOCIATION ON CHROMOSOME 6 TO THE MHC REGION AND THAT CONTINUES IN STRENGTH OF EVIDENCE AND THEN WE ADDED MORE DATA AND OUR P-VALUE CONTINUES TO GO UP AND THAT'S WHAT WE SEE THR THOSE REAL EFFECTS. THOSE REAL ASSOCIATIONS AS YOU ADD MORE DATA, YOU GET MORE SIGNIFICANCE AND THAT'S ACTUALLY IN MANY WAYS, THE STRONGEST INDICATION THAT THESE ARE ACTUALLY RELIABLE ASSOCIATIONS THAT WE SHOULD BE PURSUING, THAT BY THE TIME WE GET TO 10-MINUS 15 AND 10 TO THE MINUS 12, THOSE ASSOCIATIONS AREN'T GOING TO GO ANYWHERE AND AS WE CONTINUE TO ADD SAMPLE, YOU WILL CONTINUE TO SEE THEM STRENGTHENING IN TERMS OF THEIR EFFECT. THIS IS PUBLISHED IN NATURE. SO THIS ISN'T JUST SPECIFIC SCHIZOPHRENIA. OVER THE LAST FEW YEARS WE'VE BEEN WORKING WITH WILL DANISH COLLABORATORS MADS HOLLEGAARD, WHO DEDICATED HIS LIFE TO NEONATAL SPOTS SO THE NEW BORN HEEL PRICKS AND TURNING THOSE BLOOD SPOTS INTO USEABLE DNA RESEARCH. SO IN HIS ROLE IN THE INSTITUTE, HE MANAGED TO TURN DNA AMPLIFICATION INTO A TECHNOLOGY THAT COULD BE ROBUSTLY APPLIED TO THIS KIND OF MATERIAL AND BE SUITABLE FOR BOTH GENOME WIDE ASSOCIATION STUDIES AND SEQUENCING SO WE PARTNERED WITH THE SSI, AND OTHER PIs TO MOVE ADHD INTO THE SAME EXACT KIND OF PLACE, RIGHT? SO WE'VE NOW GOTTEN TO 20,000 CASES AND 35,000 CONTROLS FOR ADHD AND THERE ARE A DOZEN LOCI, SO THAT'S RAYMOND WORKING WITH THE STEVE WHO IS THE HEAD OF THE PGC ADHD GROUP AND WE HAVE ASSOCIATIONS THAT ARE 10 TO THE MINUS 13, THESE ARE 10 TO THE MINUS 10. THESE ASSOCIATIONS ARE NOT GOING INTO. MORE INTERESTINGLY PERHAPS THE ASSOCIATION ON CHROMOSOME 1 IS EXACTLY THE SAME ASSOCIATION AS WHAT WE SLEEP APNEA AND OBESITY IN SCHIZOPHRENIA, IT'S STRONGER, MORE STRONGLY ASSOCIATED AND HAS A BIGGER EEIVELGT ON ADHD, THAN IT DOESIE SCHIZOPHRENIA AND IT DOES CONSISTENTLY SHOW THAT SORT OF SHARED BIOLOGY THAT STEVE AND MARK WERE MENTIONING. SO WHAT DO WE KNOW? WELL WE KNOW THESE DECEASES ARE HIGHLY POLYGENIC. THEY REALLY ARE. THERE ARE HUGE NUMBERS OF GENETIC EFFECTS THAT ARE SCATTERED ACROSS THE GENOME THAT ARE INFLUENCING THESE DISEASES. LARGE SAMPLE SIZES ARE UNEQUIVOCALLY NECESSARY TO DRIVE FORWARD DISCOVERY. AND GIVEN THE DEGREE OF OVERLAP BETWEEN THE DIFFERENT PSYCHIATRIC DISORDERS AND THE KIND OF EMERGING RESULTS WE'RE FIND FREE RADICALS GENERATED INVESTIGATIONS NOT ONLY--INVESTIGATION IN SCETS FRANCIS COLLINSIA OR ADHD OR BIPOLAR OR ACCURACYICISM, IT MEANS WE TAKE A FULL COURT PRESS APPROACH TO THE GENETIC INVESTIGATION. WE CANNOT RESTRICT OURSELVES TO 1 OR 2 KIND OF HALLMARK PHENOTYPES THAT WE SHOULD REALLY GO ACROSS PSYCHIATRIC DISEASE IN AN EFFORT TO REALLY FULLY UNDERSTAND THESE MENTAL ILLNESSES AS A COMPREHENSIVE SET. THE VAST MAJORITY OF HERITABILITY IS IN THE NONCODING VARIATION, THAT COMES FROM THE GWAS VERY CLEARLY, IT COMES FROM A VARIETY OF OTHER SOURCES THAT SUGGEST THAT VARIETY OF ANALYSIS, SO HOW WE INTERPRET AND UNDERSTAND NONCODING VARIATION IS A CHALLENGE WE FACE NOT ONLY IN THE PSYCHIATRIC GENETICS COMMUNITY BUT IN THE BROADER GENERATEDET CLINICAL COMMUNITYS COMMUNITY BUT ON THE OTHER HAND CODING VARIATION HAS PARTICULAR VALUE. IT GIVES US IMMEDIATE GENES WITH WHICH TO START TO EXPLORE SPECIFIC BIOLOGICAL HYPOTHESIS AND DEEPER CHARACTERIZATIONS FOR THINGS WE CAN BE CONFIDENT ARE ACTUALLY ASSOCIATED TO DISEASE SO TO JUST CLOSE AND THINK ABOUT THESE EXPERIMENTAL DESIGN PIECES. AS A IMENET CYST, YOU HAVE 3 MAIN APPROACHES AVAILABLE AT YOUR DISPOSAL. HAVE YOU GWAS, AND YOU HAVE EXOME SEQUENCING AND GENOME SEQUENCES SO YOU AND YOU HAVE UPSIDES SO ARRAYS ARE NOW YIELDING GENOME WIDE SIGNIFICANT HITS, THE MAJORITY IS CAPTURED IN COMMON VARIATION AND IN IS CHEAP AND CHEERFUL AN EASY AND RAPID WAY TO GET TO LARGE INDIVIDUALS IF THERE ARE SAMPLES AVAILABLE. WELL EXOME SEQUENCING IS A FAST TRACK TO GENE ADENTIFICATION BECAUSE CAN YOU DO A BIGGER SAMPLE BECAUSE IT'S NOT AS EXPENSIVE AS WHOLE GENOME SEQUENCING BUT WHAT GENOMES GET YOU IS THE WHOLE CAPTURE, YOU ARE CAPTURING NONCODING STRUCTURAL AND VARIATIONS, AND ADDITION ALGORITHMS GENETIC VARIANTS HAVE A ROLE IN THE GENERATED--GENETIC BASIS FOR THESE, SO IT'S HOW DO WE INTERPRET THESE ASSOCIATIONS HOW DO WE GO FROM A NONCODING VARIANT TO A SPECIFIC BIOLOGICAL MECHANISM, THAT IS A BIG AND MEATY CHALLENGE, WE WILL TAKE AT LEAST A DAY TOW CONFRONT. EXOMES ARE A LIMITED SCOPE, YOU'RE NOT GETTING THE INFORMATION SO YOU'RE STUDENT AND POSTING INTO THE REST OF GENOME LEGAL SOME MIGHT SAY THAT THE UNDERSTANDING AND INTERPRETATION OF NONCODING AND ARRA VARIATION--RARE VARIATION, TO INTERPRET THESE THINGS AND YEAH, WE MAY DISAGROW ON THESE SLIGHTLY ON THIS. AND GENOMES GIVE YOU EVERYTHING. IT IS A DURABLE GOOD. IT WILL STAND A TEST OF TIME IN A SERIOUS WAY BUT IT IS MORE COSTLY. YOU'RE TALKING ABOUT 3 OR 4 FULL DIFFERENCE IN SAMPLE SIZE IN EXOME AND GENOME AND THAT MAKESSA I DIFFERENCE IN HOW MANY PEOPLE YOU CAN PUT INTO YOUR GENETIC STUDY AND IF THERE'S 1 THING THAT IS THE SINGLE MOST IMPORTANT LESSON IS THAT SAMPLE SIZE IS KING, IT IS THE SINGLE MOST IMPORTANT FACTOR THAT GOVERNS HOW SUCCESSFUL YOUR ABILITY TO IDENTIFY NEW REGIONS OF THE GENOME THAT ARE INFLUENCING RIVENG FOR DISEASE. AND SO--RISK FOR DISEASE. SO WITH THAT I WOULD LIKE TO SAY THANK YOU TO THE STANLEY CENTER AND ATGU GENETICS TEAM AND NOW WE HAVE DISCUSSION ON THE BOARD BUT MOST OF ALL THANKS TO THOMAS IN HIS TIRELESS EFFORTS OF HERDING THE CATS OF THE PGC, IT'S NOT A THANKFUL TAKEN--THEY TASK BUT IT'S 1 THAT IS SORELY NEEDED AND WITH THAT I WILL CLOSE. >> OKAY. [ APPLAUSE ] >> OKAY, TIMES FOR QUESTIONS, COMMENTS, REBUTTAL. >> THANKS THIS IS FOR BEN. EXCELLENT TALK AND YOU MAKE EXCELLENT POINTS ABOUT THE IMPORTANCE OF BEING STATISTICALLY RIGOROUS AND HIGHLIGHTING HOW IN THE AREA OF CANDIDATE GENE STUDIES PERHAPS HOW THINGS WEREN'T SO RIGOROUS BUT AT THE SAME TIE I ALSO HERE IN THE BACKGROUND A BIT OF THE HYPOTHESIS IS DEAD LONG WITH THE HYPOTHESIS. SO, THE IMPORTANT ISSUE IS THAT STATISTICALLY VALID ATTEMPTS ARE MADE AND THE APPROPRIATE CORRECTION FOR MULTIPLE TESTING BEING MADE WHEN YOU MAKE A HYPOTHESIS, IT'S NOT NECESSARY TO ELIMINATE HYPOTHESIS ENTIRELY, AND SO I WOULD--I WILL OFFER EXAMPLES IN ADDITION TO THE CANDIDATE GENE STUDIES, I WOULD OFFER SAMPLES FROM THE CMV WORLD WHERE FROM THE BEGINNING RARE VARIANT ASSOCIATION STUDIES WERE DONE WITH VERY SPECIFIC HYPOTHESIS FROM THE BEGINNING THAT HAD VERY STRONG PRIORS. FOR EXAMPLE, CMVs FROM THE VERY BEGINNING. WE DIDN'T TEST EVERYTHING GENOME WIDE AND THE ENTIRE BURDEN OF CMVs WAS WITH RISK CONTROLS WE STRATIFIED ON FREQUENCY AND SIZE. THERE WERE STRONG PRIORS FOR THOSE HYPOTHESIS, AND WE DID THE SAME THING WITH PATHWAYS THAT THERE WERE STRONG BIOLOGICAL, SPECIFIC GENE SETS WITH STRONG A PRIORI HYPOTHESIS WE COULD TEST AND TESTING A DOZEN OR SO PATHWAY HYPOTHESIS, THAT ARE NOW ROBUST AND PROPRODUCIBLE FINDINGS FROM THIS, MVs AND GENOME WIDE ANALYSIS AND PALGT PATH--PATHWAYS AND I DON'T RECALL HEARING A LOT OF OBJECTION ABOUT THAT AT THE TIME SO WHY WAS THAT VALID BUT NOW WE HAVE TO DO AWAY WITH HYPOTHESIS ENTIRELY AND TEST MILLIONS OF HYPOTHESIS AND HOPE THAT WE GET A 10 TO THE MINUS 8? >> SO EXCELLENT QUESTION, JONATHAN, THANK YOU. I THINK I START WITH THE FOLLOWING. GWAS HAS A HYPOTHESIS. IT IS VERY SIMPLY THAT THIS SNP IS ASSOCIATED TO THAT PHENOTYPE. THE THING IS IT'S THE GOT A MILLION OF THOSE. SO IT'S NOT LIKE WE ABANDONED THE HYPOTHESIS, AND WE ARE STILL ENGAGE NOTHING HYPOTHESIS TESTING. WE'RE HONEST WITH OURSELVES ABOUT THE FACT IS THAT WE HAVE A MILLION OF OF THOSE AND WE HAVE TO THINK ABOUT THE SPACE OF TESTING AND STATISTICAL ANALYSIS. I THINK SOME OF THE EARLIER CMV WORK CERTAINLY DID TEST AND ESTABLISH BURDEN. AND IT DID STRATIFY ON SIZE AND IT DID STRATIFY ON FREQUENCY BUT IT DID PERFORM AT LEAST, I THINK JIM STONE'S PAPER WITH THE ISC WITH COPY NUMBER VARIATION AND I'M REASONABLY CERTAIN IT TESTED GENOME BURDEN ON DUPLICATIONS AND STRATIFIED QUESTIONQUENCY AND STRATIFIED THOSE LINES AND WASN'T PERHAPS THE SAME AS THE EARLIER WORK FOR YOURSELF AND MAYBE EVEN WHY 16 P11 THAT KIND OF DE NOVO EXPIRATION THAT WE'VE DONE. BUT I THINK WHAT'S IMPORTANT WHEN WE TART GO DOWN THE ROAD OF STRATIFICATION BY SIZE AND WHEN WE GO DOWN THE ROAD OF TESTING GIVEN PATHWAYS AND NETWORKS THAT WE ACCOUNT FOR THE SPACE OF THOSE THAT WE TEST. AND I THINK IF WE DON'T DO THAT WE WILL END UP IN THE CIRCUMSTANCE WHERE EVERYONE HAS THEIR OWN RULES AND EVERYONE HAS THEIR OWN THRESHOLDS AND NO 1 WILL HAVE ANY TRUTH THAT WE CAN BE CONFIDENT IN THE ASSOCIATION. AND IF YOU LOOK AT WHAT WE WORK TOGETHER ON WITH THE SCHIZOPHRENIA ANALYSIS, THAT'S EXACTLY HOW WE PROCEEDED. WE WERE DISCIPLINED ABOUT DESCRIBING THE SPACE OF THINGS WE TESTED AND WE TALKED ABOUT THINGS THAT HAD BEEN PUT OUT INTO THE LITERATURE BEFORE, VERSUS THINGS THAT WERE NOVEL BUT WE ALSO DID AN UNBIASED EXPERIMENT AND I THINK THAT ANALYSIS IS EXTREMELY VALUABLE FOR PEOPLE TO INTERPRET AS THEY WOULD SEE FIT. >> OKAY, SO THERE'S 1 OTHER THING, THERE ARE HYPOTHESIS AND HYPOTHESIS SO YOU AND BEN ARE HAVING AN ARGUMENT ABOUT FUNDAMENTALLY GENETIC HYPOTHESIS, THE PROBLEM FOR THE FIELD WAS THAT HAVING EXTRAORDINARY IMPROVEERISHMENT BIOLOGICAL HYPOTHESIS LARGELY FROM PHARMACOLOGY AND THE DIRE NEED FOR--THE VALUE OF UNBIASED OBSERVATION SYSTEM NO 1 HAD ANY IDEA THAT CALCIUM CHANNELS MIGHT BE INVOLVED IN SCHIZOPHRENIA SO WE SHOULDN'T FORGET THE IMPORTANCE, IT'S A BIT LIKE, YOU KNOW BEFORE GALILEO GOT A TELESCOPE, AND SAW THE 4 MOONS OF JUPITER, YOU COULD FIX THE WORLD BY ADDING MORE ASTON ASTON--ASTRONOMY, SO IN BIOLOGY IT'S BEEN THE SAME. >> RIGHT, IT CAN BE UNBIASED TRANSPARENT AND STATISTICALLY RIGOROUS BUT STILL HAVE A HYPOTHESIS. >> OF COURSE. >> TERRY FROM NHGRI, ASKING ALL OF YOU ACTUALLY, YOU'VE TALKED A LOT ABOUT THE EVOLUTION IN THE GENETICS WHICH MAKES ME VERY HAPPY BUT I REMEMBER AT THE BEGINNING OF THIS FIELD THERE WAS TALKING ABOUT THE PHENOTYPING BEING A CHALLENGE AND IF YOU COULD COMMENT ON PROGRESS THAT'S BEEN MADE WITH THAT. IF YOU HEARD THE BEGINNING OF MY TALK, YOU WILL SEE I'M STILL A BIT NIHILISTIC AND TERRY BACK IN 1997 OR 8, YOU HELPED ME AND OUR GENETICS GROUP SORT OF FIGURE OUT HOW TO PROCEED. I ACTUALLY LOOKED AT THE 1997 COUNCIL GENETICS REPORT AND IT'S AMAZING HOW MUCH IS STILL GERMANE TODAY AND REALLY WAS ARGUING FOR TEAM SCIENCE. I MEAN YOU SAW IT AS WELL. IT HAD AN EARLY DATA SHARING POLICY WHICH LED THE DIRECTOR OF A DIFFERENT INSTITUTE TO SAY, GET YOUR AUTISM GRANTS FROM US, WE WILL NEVER MAKE YOU SHARE. SO, WE'VE AT LEAST MADE SOME PROGRESS, BUT PHENOTYPING, I THINK, THE VAST DIFFERENCE BETWEEN WHAT YOU CAN DO IS CARDIOVASCULAR DISEASE AND WHAT WE CAN DO REALLY DOES HAVE TO DO WITH THE LACK OF OBJECTIVE MEDICAL TESTS, THE LACK OF TISSUE AND THE FACT THAT--I MEAN SOME SENSE I JUST MENTIONED THAT DIFFERENCE BETWEEN TOMAIC AND KAEPERNCAISE IN ASTRONOMY, THE PROBLEM WITH ALL THE PHENOMENAL LOGICAL OBSERVATIONS WE CAN MAKE IS THAT THEY FIT ON TO MANY DIFFERENT UNDERLYING FORMS OF BIOLOGY. SO I THINK THE PHILOSOPHY HAS BEEN THATY WOO ARE GOING TO HAVE A RATHER UGLY ITERATIVE APPROACH TO PHENOTYPING WHERE GENERATEDEE --GENO TYPES WILL GIVE US NEW HYPOTHESIS ABOUT PHENOTYPES AND THEN WE WILL GO BACK TO THE POPULATIONS, TEST THOSE HYPOTHESIS TO THE GENETICS. I REALLY DON'T SEE ANY OTHER--UNLESS SOMETHING MAGICAL HAPPENS ABOUT NONINVASIVE NEUROIMAGING OR OTHER TECHNOLOGIES WHERE THEY CAN MANAGE THE HETEROGENEITY AFFORDABLY, TELL BE AN ITERATIVE PROCESS BETWEEN GENETICS AND CLINICAL OBSERVATIONS. >> LAST QUESTION? >> YES, I ENJOYED ALL THE TALKS, ALL GREAT. I HAVE A REQUEST ABOUT BEN ABOUT THE ADHD STUDY. SO IN ADDITION TO FINDING SOME GENES THAT ARE GETTING BOTH THRESHOLD, HAVE YOU ALSO LOOKED AT THE POLYGENIC ISSUES AND ANYTHING EMERGING THERE. >> SURE. IT'S THE SAME STORY AS IT IS FOR BASICALLY EVER MAJOR FORM OF PSYCHOPATHOLOGY THAT WE SEE AROUND 20-30% OF THE VARIATION AND LIABILITY, EXPLICABLE BY COMMON VARIATION, THAT'S AGAIN SCATTERED ACROSS THE GENOME, POLYGENIC RISK SCORING DOES THE SAME THING IN ADHD AS IT DOES IN SCHIZOPHRENIA THAT WE CAN LEAVE SAMPLES OUT AND PREDICT. THOSE PREDICTIONS ARE PRETTY POOR. YOU KNOW 1-2% CURRENTLY IN PSEUDOR-SQUARED AND IT'S SOMETHING THAT WILL IMPROVE AS OUR RESOLUTION OF WHAT VARIANTS MATTER AND WHAT DON'T IMPROVE AND AS WE DRIVE DOWN THE STANDARD AREAS ON THE ESTIMATES ON THE INDIVIDUAL EFFECTS THEMSELVES. THERE'S OVERLAP, THE OVERLAP BETWEEN ADHD AND SCHIZOPHRENIA AND BIPOLAR AND MAJOR DEPRESSIVE DISORDER AT LEAST FROM THE COMMON VARIANT, AVERAGES AROUND 30% OR SO WITH RESPECT TO RG, ABOUT YOU ITERALATES--CORRELATES WITH A BUNCH OF OTHER THINGS LIKE SMOKING AND WHETHER OR NOT YOU TRY SIGNATURES--SIGNATURES CIGARETTES OR THINGS LIKE THAT. THERE ARE LOTS OF OTHER THINGS THAT HUMANS DO BEHAVIORIABLY THAT ALSO RELATE TO PSYCHEEE PATHOLOGY, THEY'RE NOT IDENTICAL AND THERE'S A LOT OF THIS CONTINUUM RUNNING AROUND THE POPULATION FOR A LOT OF DIFFERENT THINGS AND WE SHOULD BE OPEN TO THE INTEGRATION OF THOSE KINDS OF APPROACHES AND WAYS OF MEASURING PHENOTYPE TO LEARN MORE ABOUT MENTAL ILLNESS. >> THANK YOU. SO OUR NEXT SPEAKER IS MIC 'ODONOVAN BUT HE'S KNOWN TO A LOT OF PEOPLE HERE AS CHAIRING THE SCHIZOPHRENIA GROUP IN THE PGC WHICH IS NOT ONLY A SCIENTIFIC AND PAPER RIGHTING TASK, BUT UNDOUBTEDLY REFLECT HIS FAIRNESS AND HIS MANAGEMENT SKILLS BECAUSE, HUMAN NATURE IS WHAT IT IS AFTERALL. THE OTHER THING IS THAT MIC I THINK HAD THE LONGEST SCHLEP TO GET HERE OF ANYBODY IN THE AUDIENCE. SO WITH THAT. [ APPLAUSE ] >> SO ED, THANK YOU FOR THAT. I'VE BEEN TASKED WITH TALKING ABOUT THE PROGRESS MADE BY THE PGC AND BECAUSE OF THAT, I HAVE TO DO THIS DECLARATION. THERE'S NOT MUCH CREDIT OR BLAME ATTRIBUTED TO ME BUT YOU CAN TELL WE MUST BE DOING A GOOD JOB BECAUSE WE HAVE A LOGO NOW. - LOOKS A BIT SINNISTER TO ME, IT LOOKS LIKE THAT SOMEONE IN A JAMES BOND FILM USED. SO WE HAVE HEARD A LOT ABOUT CANDIDATE GENES. THIS WAS 1 OF MY BETTER CANDIDATE GENES STUDIES AND I WAS IMMENSELY PROUD OF IT AND YET MANY YEARS PASSED AND I DON'T KNOW IF IF--IF IT'S TRUE. SO THIS IS ACTUALLY WHAT SOMETHING EXPENSIVE LOOKS LIKE, AND SOMETHING THAT IS EXPENSE AND I HAVE NEVER ACTUALLY DELIVERS SOMETHING, THERE'S ALWAYS RUMORS THAT IT'S PREGNANT, IT'S ABOUT TO DO SOMETHING GOOD, BUT PANDAS NEVER DO ANYTHING. I AM NOT HAPPY WITH TERM POOP, BUT I'M CANCER CENTER INCAs LO SAXON AND SPECIFIC TERMINOLOGY AND--[LAUGHTER] --THE SAME WAS THE CASE WITH GENETICS TO BE SAID. SO THIS IS--NOW I WASN'T THERE WHEN THE PGC STARTED BUT THIS IS WHAT I TOLD THAT THE GAIN PIs WHO WERE STEVE, PAUL, AND PAT, I'VE GOT TOGETHER IN SOME SORT OF KABALL AND THE APPROACH THAT MARK AND PAMELA AND DOUG, THEY WERE DOING OTHER BIG STUDIES TO FORM A CONSORTIUM AND RECOGNIZING THE AMERICA CANNOT STAND ON ITS OWN, THE THEY APPROACHED NICK KRAIDUC IN THE UK WHO WAS LOOKING AFTER THE CASE CONTROL SON SORTIUM, AND IT WAS VERY WISE THAT THE AMERICANS DID APPROACH EUROPE BECAUSE CERTAINLY IN THE EARLY STAGES AND IS PROBABLY STILL THE CASE, THE VAST MAJORITY OF THE PGC SAMPLES COME FROM EUROPE. SO I THINK WE HEARD ENOUGH ABOUT STATISTICAL POWER FROM MARK AND BEN ALTHOUGH WE CAN NEVER HEAR ENOUGH ABOUT IT WE HAVE TO KEEP IT IN THE BACKS OF ONURE MINDS AT ALL TIMES BUT THIS WAS SAID TO BE THE MOTIVATING FACTOR WAS TO ACHIEVE POWER TO DELIVER RESULTS. YOU'LL NOTICE, IF YOU CAN SEE THAT EVEN THESE--WAS THE KIND OF THE WEAKEST GENETIC EFFECTS BUT OBVIOUSLY THIS IS REALLY OPTIMISTIC COMPARED WITH WHAT WE'RE FINDING NOW, SO THE MOTIVATION WAS SAMPLES, SAMPLES AND GETTING PEOPLE TOGETHER. SO JUST BRIEFLY THIS IS A FIRST TRENCH OF PGC STUDIES THERE WERE 5 DISEASES, ADHD, AUTISM BIPOLAR, MAJOR DEPRESSION, DITS FRENNIA ACROSS THIS DISORDER AND THERE WAS A NUMBER OF DELIVERABLES, I MEAN THERE WAS A NUMBER OF REASONABLY STRONG ASSOCIATIONS. BUT THE MAIN--THE MAIN POINT HERE IS THAT THEY SHOWED THAT THE CONSORTIUM ACTIVITY COULD DELIVER. AND THEN, THIS IS A SLIDE THAT STEVEN MADE NOT FROM ME, BUT FROM HIMSELF AND SOMEHOW IT FELL INTO MY HANDS SO THIS WAS A PRESENTATION THAT WAS GIVEN SOMEWHERE ELSE AND I APOLOGIZE TO STEVEN, I DIDN'T ASK FOR HIS PERMISSION TO USE IT, BUT ANYWAY, THIS IS THE PROGRESS THAT SCHIZOPHRENIA WAS MAKING SO ROUND ABOUT THEN WAS WHEN THAT CANDIDATE GENE STUDY THAT I SHOWED YOU RIGHT AT THE BEGINNING WAS, WE PUBLISH AID SMALL GWAS OF SCHIZOPHRENIA ROUND ABOUT HERE WHICH IDENTIFIED 1 THING COMBINED IN SCETS FRENNIA BIPOLAR DISORDER, THE ISD WORK AND THEN THINGS START TAKING OFF ROUND ABOUT 2011 AND GO ON TO THE SLIDES THAT YOU'VE SEEN. NOW PROGRESS HAS NOT STALLED EVEN THOUGH--PROGRESS HAS NOT STALLED. THIS IS WHERE WE ARE ON SCHIZOPHRENIA. NOW THIS LOOKS LIKE A DIFFERENT LOGO BUT SO I DON'T KNOW WHY WE HAVE 2 LOGOS, BUT, THIS IS WHERE PROGRESS IS NOW. SO THERE WAS 2008, TO 2014, NOW GWAS STUDIES ARE THE GIFT THAT KEEPS GIVING AND I THINK ALL OF YOU WHO ARE GENETICISTS ALL YOU MEN AND WOMEN WHO ARE GENETICISTS KNOW THIS BUT A LOT OF PEOPLE OUTSIDE OF GENETIC FIELD DON'T APPRECIATE THIS, THEY SAY YOU'VE DONE A STUDY OF 20,000 PEOPLE AND YOU FOUND X THINGS WELL THAT'S NOT MUCH, THEY FAILED TO PROASHT THAT THOSE DATA ARE THERE TO BE ADDED TO--WE ALL KNOW THIS BECAUSE WE WORK IN THESE LABS BUT WE HAVE PEOPLE BUT IT USES ALL THE PGC DATA THAT WAS SUBMITTED IN 2016. AND IF ANY OF YOU GUYS OR WOMEN ARE REVIEWING IT CAN YOU LEAVE A HOLE AND SUBMIT YOUR REVIEW AND PUT US OUT OF OUR MISERY 1 WAY OR ANOTHER BECAUSE IT'S BEEN UNDER REVIEW SINCE THE BEGINNING OF MARCH. SO IMRAI, THE PGC HAS INSTALLED AND THIS IS VERY MUCH THE ROYAL WE, CONDUCTING ANALYSIS WHEN WE COMBINE LARGELY EUROPEAN DATA AND EUROPEAN DATA WE HAVE 70,000 CASES ON THE FIRST LIGHT WE MIGHT HAVE 300 OR 350 LOCI. THIS SAY GRATIFYING DEVELOPMENT, GRATIFYING DEVELOPMENT. A WHILE AGO, SOME OF US PRODDED OUR COLLEAGUES FROM SOUTHEAST ASIA WHO FORMED A WORKING SUBGROUP OF SCHIZOPHRENIA, IT WOULD GENETICALLY REPRESENT THE MAJOR POPULATIONS. THEY'VE NOW GOT A PAPER AND DEVELOPMENT WITH A LOT OF NEW LOCI AND AS MARK, I THINK IT WAS SAID THERE'S EVIDENCE FOR LIMITED HETEROGENEITY COMPARED WITH EUROPEANS. BUT, THE PGC IS NOT JUST ABOUT SCHIZOPHRENIA, SO THERE'S BEEN NO NEW EXPANSION IN THE NUMBER OF WORKING GROUPS, INCLUDES SUBSTANCES, ABUSE, OCD, TERRETS, PTSD, ACROSS DISORDER WORKS, CMVs AND STATISTICAL ANALYSIS. WITH A VERY LARGE DANISH EYE SIGHT REPRESENTATION WHICH HAS BEEN PIVOTAL AS YOU'VE HEARD FROM THE CHILD ONSET DISORDS OF ADHD AND AUTISM. AND WE HAVE A NEW GRUP THAT HASN'T MADE IT INTO THIS GROUP YET AND THIS IS ANXIETY AND ALZHEIMER'S DISEASE, AND WE DEPEND ON A LOT OF SUPPORT ON NIMH AND THOMAS AND PETER AND ANGIE. SO ANOTHER ACHIEVEMENT OF THE PGC WAS THE DEVELOPMENT OF SIGHT CHIP, I GOT THIS SLIDE FROM PAMELA, AND THESE 2 CRIMINALS IN IT BECAUSE I ADDED PAMELA BECAUSE I REMEMBER THAT SHE WAS STRONGLY INVOLVED IN THE CHIP, SO THIS IS BEEN USED IN GWAS LEVEL, MORE THAN 150,000 INDIVIDUALS FROM A LOT OF COUNTRIES THROUGHOUT THE GLOBE AND THIS WAS DONE A LOT WITH THE NIMH FUNDING, A LOT FROM THE STANLEY FUNDING AND INDEPENDENT GROUPS THEMSELVES. AS A RESULT, OTHER DISORDERS ARE BEGINNING TO GAIN SOME OF THE SUCCESSES THAT SCHIZOPHRENIA HAD. SO OBVIOUSLY, THIS IS ANKER EXPECTATIONSIA NERVEOSA, AND THIS THEY ARE GETTING RESULTS AGAINST SMALL SAMPLE SIZES SO IT REMAINS TO BE SEEN IF THIS GOES UP OR DOWN BUT IT'S A SIGN OF PROGRESS. THE MOST INTERESTING THING ABOUT THE ANOREXIA NERVEOSA AS PEOPLE MENTIONED WE'RE NOT USED TO SEEING CORRELATIONS BETWEEN A PSYCHIATRIC DISORDERS AND BETWEEN PSYCHIATRIC DISORDERS AND COGNITIVE PERFORMANCE AND I HAVE NOTHING TO GO BUT THIS WORK AND NOT ON THE PAPER BUT SO THIS IS OTHER PEOPLE'S INTERPRETATION, SO WHAT SEEMS TO BE QUITE INTERESTING HERE IS AFTER YOU ADJUST THE POPULATION STUDIES FOR BMI, WEIGHT AND SO ON, THE POLYGENIC SCORE OR THE GONEETIC CORRELATIONS WITH THE LDSCORE LOOK AT ANOREXIA NERVEOSA HAS A LOT OF METABOLIC PROBLEMS, SO WHAT THAT PLEENS FOR ANKER EXPECTATIONSIA, I DON'T WORK IN ANKER--AN ANOREXIA,--IT'S ADH D IS MAKING PROGRESS, WE HEARD ABOUT THAT FROM BEN. BUT GRATIFYING AND A HEROIC NOTE, MAJOR DEPRESSIVE DISORDER IS KICKING OUT A LARGE NUMBER OF LOCI. THIS SLIDE AT 1 POINT HAD NO TWEETING SO THAT TELLS YOU IF ANY OF YOU ACTUALLY TWEET, SO MANY IN THE GROUP WOULD PREFER IF YOU DIDN'T TWEET IT. BIPOLAR DISORDER, AGAIN, ALSO MAKING REASONABLE PROGRESS SAMPLE SIZES, ARE STILL LAGGING BEHIND SCHIZOPHRENIA BUT RAFTER FINDINGS, 30 LOCI, TWIN STUDIES MANY OF WHICH ARE NOVEL AND MANY CONSTRUCTION AND DESPITE ME URGING THE CHAIRS OF THE RECOLLECT GROUPS USING SOMEWHAT THREATENING LANGUAGES, I DID GET SLIDES FROM THE WORKING GROUPS SO THAT'S WITHOUT PROGRESS, WITH RESPECT TO THE GWAS STANCE, WE ALSO AT EARLIER ON 2016 PUBLISHED NOT JUST COMMON VARIATION, BUT I SUPPOSE WHAT WOULD BE ARGUABLILY CURRENTLY THE DEFINITIVE STATEMENT ON CMVs IN THE MOMENT IN SCHIZOPHRENIA, WHERE THE CMV GROUP SPENT AGES AND THAT'S NOT A CRITICISM DOING METICULOUS QC AND HODGEOG NOWS, TO GET WHAT LOOKS LIKE A CLEAR SET OF RESULTINGS, WHERE THERE'S 8 CLEARLY ASSOCIATED CMVs IN ANOTHER 8 THAT'S WHERE THEY'RE SMUGGLED IN UNDER THE FDR MANEUVER THAT BEN MENTIONED LATER ON. NOW, THAT'S THE STAMP COLLECTING AND OF COURSE, THERE'S NOTHING WRONG WITH STAMP CLICKING IF THAT'S YOUR OWN PERSONAL HOB BUT IT'S NOT THE MOST SOCIAL USEFUL ACTIVITY IN THE WORLD AND WE ARE DOING RESEARCH TO TRY TO HELP PEOPLE SO THE QUESTION S&P ANY OF THIS STUFF OF ANY VALUE? ARE WE GETTING ANY USEFUL KNOWLEDGE? SO REMOVING FROM, I SUPPOSE AREAS OF--OFTEN, WHAT I WILL TALK ABOUT NEXT ARE AREAS THAT--WHERE'S THE GENETIC FINDINGS ARE ARE BUST, SOME THINGS I WILL OFFER ARE A BIT SPECULATIVE AND I WILL TRY AND MENTION A FAIR MEMBER WHICH ARE A BIT SPECULATIVE. BUT FIRST OF ALL IN ORDER TO MAKE FULL USE OF THE PGC DATA, I MEAN AN EARLY DECISION WAS MADE THAT THESE OLD GENOME WIDE SUMMARY RESULTS WOULD BE MADE AVAILABLE UNRESTRICTED WITH NO MESSY APPLICATIONS, BASICALLY ANYONE CAN DOWNLOAD THE SUMMARY STATISTICS. AND THAT'S SOMETHING THAT NOT ALL CONSORTIA DO AND I SAY THAT AS SOMEONE WHO SPENT QUITE A LONG TIME GETTING DATA OUT OF THE OTHER CONSORTIUM SO THE OTHER IS IN ORDER HERE FOR THE PGC. SO, USEFUL THINGS. WELL, I MEAN EVERYTHING WILL TAKE TIME TO MATURE TO REALLY DELIVER THINGS THAT HELP PATIENTS. SO I'M JUST GOING TO FLAG UP SOME OF THE STUFF GIVING YOU INSIGHTS, IDEAS, NEW WAYS OF THINKING ABOUT SI PSYCHEIATIC DISORDERS SO 1 OF THE INTERESTING EARLY QUEPMENTS WHEN THERE WERE NOT LOCI IDENTIFIED IS THE ASHTS MOUNT OF HERITABILITY OF THE TRAIT LIABILITY THAT WAS ATTRIBUTABLE TO COMMON VARIANTS THAT WERE ACCESSIBLE IN GWAS AND I THINK AS BEN MENTIONED BETWEEN 20-30%. BUT MORE INTERESTING WE STARTED GETTING CO-HERITABILITY SO EVIDENCE THAT BIPOLAR, MAJOR SHARE GENETIC, RISK AND COMMON. NOW THIS IS SO MUCH--THIS IS BASICALLY DOGMA NOW THAT I THINK NO 1 FINDS THIS INTERESTING ANYMORE. I CAN WILL IT YOU WHEN THE ISC PUBLISHED THIS SCHIZOPHRENIA WITH OVERLAP BIPOLAR DISORDER IT WAS EXTREMELY CONTROVERSIAL. THIS EARLY WORK HAS MOVED ON, TO BIGGER SAMPLES AND I WILL NOT GO THROUGH THIS IN DETAIL. I SUSPECT OTHER PEOPLE MAY HAVE SLIDES IN THIS, AND THIS IS REALLY JUST DEMONSTRATING THAT THE SUBSTANTIAL OVERLAP IN GENETIC LIABILITY TO A RANGE OF DISORDERS, NOW I DON'T KNOW WHERE THIS WILL END UP, THIS TYPE OF WORK, I CAN'T REALLY PREDICT THE FUTURE BUT YOU KNOW UNDERSTANDING THE SIMILARITIES AND DIFFERENCES BETWEEN THE CATEGORIES THAT WE HAVE NOW MAY FORM THE BASIS FOR DIFFERENT WAYS OF IDENTIFYING DIMENSIONS OF SITES, SYMPTOMS AND BIOLOGY THAT'S REALLY UNDERLYING THESE DISORDERS. NOW, THIS IS A BIT MORE OF THE KIND OF SPECULATIVE STUFF, SO WE HEARD A LOT ABOUT PATHWAY ANALYSIS AND HOW THEY HAD TO BE CONDUCTED CAREFULLY. THIS IS WORK DONE BY THE NETWORK AND PATHWAY ANALYSIS SUBGROUP OF THE PGC, AND BASICALLY THE 2--THE CROSSED DISORDER FINDINGS AND THE ENRICHMENT USING METHODS THAT WE HAVE TO SEE ARE NOW NO LONGER STATE-OF-THE-ART BUT THEY LOOK FOR ENRICHMENT IN PATHWAYS FOR VARIOUS BIOLOGICAL PROCESSES IF THE SIGNAL FOR THIS CROSS DISORDER COULD BE ATTRIBUTED TO SYNAPSE, HISTONE METHALATION, IMMUNE NEURONAL NEUROTROAPIC PATHWAYS. YOU KNOW YOU HAVE TO TAKE SOME OF THIS STUFF WITH A HEALTHY DEGREE OF SKEPTICISM AND MOST OF US DO, DO IT BUT IT SEEMS PERHAPS NOT A COINCIDENCE THAT THE FIRST REAR VARIANT BY REAR AREA EXOME SEQUENCE IMPLICATED GENE AS SET D1-A, IN THE ARM, HISTONE METHALATION OF ARM BIOLOGY AND THE FIRST COMMON VALID AND VARIANT THAT WAS BIOLOGICALLY NAILED IN TERMS OF CAUSAL VARIANT AND CAUSAL GENE BELONGS TO THIS LENGTH OR THIS LENGTH. BUT WE ALSO HAVE THIS VERY MURKY AND BROAD CATEGORY CALLED THE SYNAPSE AND I DON'T KNOW IF I CAN SAY THIS, BUT IT'S BEING VIDEOS BUT THE SYNAPSE SINCE I SAW THIS SHOULD IS THE NO-SHIT-SHERLOCK MOMENT IN PSYCHIATRIC GENETICS WHERE THE GENETICS POINTING TO A BIT OF BIOLOGY WHICH IS ESSENTIALLY WHAT THE BRAIN IS ABOUT SO IT REALLY JUST SAYS THAT THIS IS A BRAIN DISEASE BUT WITHIN THE SYNAPSE, THE PATHWAYS ARE DRIVING THESE GLUTA METERGIC DENSITY AND MEMBRANE, NOW THESE ARE NOT--THESE ARE--I THINK THESE WERE THE CORRECTED P-VALUES TRUNCATED SO THESE ARE NOT--THEY'RE NOT SIGNIFICANT IN THIS STUDY WHEN YOU CORRECT FOR ALL THE MULTIPLE TESTING SO THESE WERE JUST THE PROMISING 1S. BUT WHEN--I THINK THIS IS WHAT JONATHAN WHEN HE WAS RAISING QUESTIONS WITH BEN WAS BEGINNING TO ALLUDE TO, THE CNV STUDY, WHEN YOU LOOK FOR VARIOUS BIOLOGICAL PROCESSES THAT ARE ENRICHED WITH CNVs, AGAIN, THIS POINTS THESE KIND OF STRUCTURES, THE--THE NMDA COMPLEX, WERE THE ONLY THING THAT REACHED DUPLICATIONS, ARE COMPLEX, BUT I PUT THIS PICTURE UP AND THIS IS A PICTURE WE HAD MADE FROM AN AREA PAPER WHERE WE ARE LOOKING AT DE NOVO CMVs AND WE SUBSEQUENTLY FOUND EVIDENCE AGAIN THAT'S NOT CORRECTED FOR HUGE WRAPS OF MULTIPLE TESTING FOR AXONIC VARIANT AND DE NOVO MUTATIONS AND NOW THIS BIG STUDY OF CNVs POINTS TO THESE. THE RELEVANCE OF THE LIGHTLY RELEVANCE OF THIS AMONGST OTHER AREAS OF BIOLOGY. IT'S KIND OF GENERAL POINTING TO THE GLUTA METERGIC TESTING AND SITES AND NOT EXPLAINING THE WHOLE OF SCETS--SCHIZOPHRENIA. THIS SLIDE IS A BIEWEVERL PIECE OF WORK DONE BY STEVE Mc COLLEAGUES AND EVERYONE, EVERYONE WHO'S INVOLVED IN GENETICS KNOWS THIS BUT I PUT IT UP AS A PRODUCT, AS A BENEFIT OF THE PGC, THIS WAS NOT PGC WORK, STEVE TOOK THE RAW DATA FROM THE PGC THAT HE WAS GIVEN ACCESS TO AND ABLE TO SHOW THAT MUCH OF THIS SIGNAL IN CHROMOSOME 6 WAS DRIVEN BY COMPLEMENT COMPONENT 4 WHICH IS A WELL--WELL KNOWN INVOLVED IN THE VIDEOS IMMUNE O PATHWAYS BUT AGAIN, THE BODY REPURPOSES LOTS OF PROTEINS, YOU KNOW JUST BECAUSE IT WAS WHEN THESE WERE ORIGINALLY DISCOVERED THEY WERE CALLED IMMUNE PROTEINS, THAT'S ALL THEY DO AND CERTAINLY IN THIS PAPER, THERE WAS EVIDENCE THAT COMPLEMENT WAS INVOLVED IN SYNAPTIC PRUNING WHICH WAS NOT, YOU BEGIN DO COME INVOLVED IN THE SEMANTIC INVOLVEMENT, IS THAT IMMUNITY OR SYNAPTIC SCULPTURE THROUGH AN IMMUNE PROCESS SO IT DOESN'T IMPLY A MORE GENERAL ROLE FOR IMMUNITY IN SCHIZOPHRENIA. AND THEN THIS IS--IT HAS TO BE SAID, AND INDEED IT WAS SAID EARLIER ON THAT WE'RE MILES AWAY FROM TAKING GWAS FINDINGS AND MAPPING THEM ON TO BIOLOGY. SO I'VE JUST GIVEN A FEW OF THE EARLIER ATTEMPTS TO DO THIS. THIS IS A PAPER THAT WAS LED BY PAMELA SCLAR TAKING THE MAIN CONSORTIUM GENE EXPRESSION, GROUPING THE JEERN EXPRESSION INTO MODULES, LOOKING FOR MODULES OF COLD CORRELATED GENES IF YOU LIKE IN TERMS OF EXPRESSION AT LOOKING FOR EVIDENCE THAT THEY ARE ENRICHED FOR GWAS SIGNALS AND INDEED THEY WERE AND THE SAME MODULE WAS ENRICHED FOR CMVs AND RARE VARIANTS AND THAT MODULE WAS RARE GLUTAMATERGIC THINGS AS WELL AS TARPGETS AND IN THAT PAPER THEY ALSO REPORTED AT LEAST THERE WERE CREDIBLE CAUSAL, SNPs THAT WERE CREDIBLE IN THE SENSE THAT THERE'S A FINITE PROBABILITY THAT THEY MIGHT BE EXPLAINING THE ASSOCIATION SIGNAL THAT WE'RE ALSO APPEAR TO BE ACTING AS QTLs AS 22 LOCI. NOW THESE ARE EARLY STEPS IN TRYING TO SEE WHAT THE DRIVERS OF THE GENETIC ASSOCIATIONS ARE AND I'M NOT--I DON'T KNOW IF THESE WILL HOLD UP. WE'VE EXTENDED THIS TYPE OF WORK TO SOME OF THE NEWER DATA WE GENERATED AND REPLICATE WILL EXTEND MOST OF THEM RATHER THAN REPLICATE AND FIND A FEW MORE. AND THEN THIS IS ANOTHER EXPLOITATION OF PGC, EXPLOIT IS USUALLY A BAD WORD. THIS IS EXPLOIT IN A GOOD WORD HERE, EXPLOITATION BY DON AND HIS COLLEAGUES WHO AGAIN TRIED TO FIGURE OUT FUNCTIONALITY FROM SOME OF THE SCHIZOPHRENIA, GWAS BY SEEING WHETHER OR NOT THE INCREDIBLE SNPs MAPPED FOR CHROMATIN INTERACTION AND I THINK THERE WILL BE TALKS ABOUT THAT LATER ON SO I WILL NOT BELABOR THE POINT. ONE OF THE OTHER USEFUL TOOLS, I MEAN, I THINK WE HEARD FROM STEVE HYMAN EARLY ON THAT THERE WERE NO BIOMARKERS BUT ARGUABLY I THINK THE POLYGENIC SCORES REPRESENT A BIOMARKER ALBEIT IT'S NOT THE SORT OF BIOMARKER THAT FLUCTUATES IN RESPONSE TO TREATMENT SO YOU CAN'T USE IT TO MONITOR TREATMENT EFFECTS. THIS IS REALLY THE APPLICATION OF THIS METHODOLOGY IS BREAKING OUT ACROSS THE WHOLE OF PSYCHEIATIC RESEARCH AND MEDICAL RESEARCH AND SOCIAL SCIENCE THAT ARE EMBRACING OR AT LEAST THOSE WHAT ARE EMBRACING GENETICS SO THIS IS--I'M PRETTY SURE I DON'T NEED TO EXPLAIN POLYGENIC SCORES AND YOU COUNT UP THE NUMBER OF MODISKS YOU COUNT UP THE NUMBER OF RISK ALLELES AND THAT'S YOUR SCORE. AND THEN YOU CAN TO AN EXTENT, HAVE YOU THE CASES AND CONTROLS, MORE IMPORTANTLY, FOR THE RESEARCH, YOU CAN--YOU HAVE A KIND OF CONTINUOUSLY DISTRIBUTED MEASURE OF LIABILITY THAT CAN BE MEASURED IN ANYONE REGARDLESS OF WHETHER OR NOT THEY'RE MANIFESTING SYMPTOMS AND THIS IS--ACTUALLY WHAT OUR RESULTS LOOK LIKE WHERE THE BLUE IS THE CASE AND SCHIZOPHRENIA CASES AND THE CONTROLS, BUT THIS IS THEIR OWN USES, BUT CURRENTLY TAKING THE BIGGEST DATA SET WE COULD IS TESTING INTO PRETTY DESSENT SIZED--DESCENT SIZED SCHIZOPHRENIA CASES THAT ARE IN THE PGC WE GET AN ODDS RATIO, THIS IS A SLATE MESSAGE, THIS IS FROM 1986 BUT THAT'S FROM THE LOWEST SENTILE TO THE TOP SENTILE AND IT'S CAPTURING SOMETHING THAT'S USEFUL AND I ONLY WANT TO ILLUSTRATE, I MEAN THIS--THIS TYPE OF APPROACH IS SO MANY USES THAT YOU COULD SPEND DAYS TALKING ABOUT THE DIFFERENT OPERATING GLOBALLY GLOBALLY--APPLICATIONS. I WILL GIVE 2 EXAMPLES. SO THIS WAS A SAD EFFECT, THIS WAS NOT A MAJOR POINT OF OUR STUDY BUT WHAT WE DID IS WE TOOK OUR--IT WAS ABOUT 10,000 PEOPLE WITH A MIXTURE OF DIAGNOSIS, SCHIZOPHRENIA, BIPOLAR DISORDER AND BIPOLAR, AND WE LOPPED THEM IN A BIG POT AND I MEAN JUST RANKED THEM BY DECILES OF THE SCHIZOPHRENIA POLYGENICS AND AS YOU CAN SEE, AS YOU EXPECT, THE PROBABILITY OF BEING--HAVING SCHIZOPHRENIA OVER BIPOLAR DISORDER RISES ALONG WITH THE CELLS AND PROBABLABILITY OF BIPOLAR GOES IN THIS DIRECTION, SO IF YOU'RE--IF YOU'RE MODERATELY LOW STITS FRENNIA, YOU ARE MORE LIER BIPOLAR DISORDER AND SCETS FRENNIA AND VICE VERSA, NOW CAVEATS HERE, THESE ARE NOT INCIDENT CASES. THE DISTRIBUTION OF SAMPLES IN THE POT DO NOT NECESSARILY REFLECT THE POPULATION PREVALENCES. BUT YOU CAN JUST ABOUT IMAGINE THAT IF YOU ARE A BIG EMERGING SAMPLE OF PEOPLE BEGINNING TO EXHIBIT SIGNS OF SEVERE PSYCHOPATHOLOGY AND IT CAN OFTEN TAKE A WHILE TO FIGURE OUT WHICH DIRECTION THIS IS GOING IN, PERHAPS, JUST PERHAPS, ALTHOUGH THE POLYGENIC SCORE IS NOT DIAGNOSTIC IT COULD FAVOR ALONG WITH OTHER THAT THIS PERSON IS MAYBE ABOUT TO DEVELOP SOMETHING THAT WE COMMONLY CALL SCHIZOPHRENIA AND NEEDS TO BE MONITORED MORE CAREFULLY THAN SOMEONE ELSE. THAT'S WHAT I WANTED TO SHOW, I WANTED TO SHOW ANITRIC OXIDE'S' WORK EXPLOATING THE WEAK ADHD POLYGENIC SCORE AND SHE LOOKED AT A SAMPLE CALLED ALSBACK, AND THIS MEASURES ADHD TYPE SYMPTOMS AND SHE WAS ABLE TO DIVIDE THE SAMPLE UP INTO AGI, MAY HAVE DONE THE TRAJECTORY ANALYSIS, AND THEY WERE ABLE TO DIVIDE THAT INTO 4 CLUSTERS. PEOPLE WHO HAD HIGH SYMPTOMS EARLY ON AND WHO STAYED HIGH. PEOPLE WHO WERE HIGH EARLY ON IN CHILDHOOD, PEOPLE WHO WERE MODERATE ALONG THE LINE AND THE MAJORITY OF THE SAMPLE DIDN'T HAVE MUCH IN THE WAY OF ADHD SYMPTOMS AT ALL AND REALLY, ALTHOUGH THIS IS NOT CLINICALLY APPLICABLE BECAUSE THE EFFECTS ARE SMALL, IT WAS THE GROUP THAT STAYED WITH ADHD INTO THEIR--SO AT LEAST INTO THE LATE TEENS, THE 1S THAT THE DISORDER REMAINS CHRONIC AND THESE COMPARE TO OTHER GROUPS SO THIS IS HIGHER THAN ANYTHING ELSE AND THIS IS INCLUDING THE 1S AT CHILDHOOD WHO HAD A HIGH SYMPTOMS FOR THE 1S OF THE LOW SCORE SEND TENDED TO DROP OFF. IT'S JUST AN INDICATION OF SOMETHING THAT MIGHT--THIS POTENTIAL, THIS POTENTIAL APPLICATION FROM BEGINNING TO IDENTIFY PEOPLE AT HIGH RISK OF PERRENT SYMPTOMS AND--PERRISTENT SYMPTOMS AND INTERESTINGLY ENOUGH, WE ARE USED TO SEEING AND VERY ACCUSTOMED TO SEEING IF YOU HAVE POLYGENIC SCORE OR GENETIC OVERLAP WITH 1 CONDITION OR OVERLAP WITH EVERYTHING. IF YOU APPLY THE SCETS FRENNIA SCORES THEY DON'T DO THIS SO THIS SEEMED TO BE RELATIVE SPECIFIC TO THE ADHD LIABILITY. AND FINALLY, THE PGC HAS HIGH OUTPUT PRODUCTIVITY AND THE OUTPUT COMMUNITY THIS IS PUT TOGETHER BY PAT SULLIVAN, SO IF THERE'S ANY HYPE OR EXAGGERATION, HE PUT THESE TOGETHER SO HE CLAIMS WE HAVE 23 PRIMARY PAPERS 51 SECONDARY ANALYSIS PAPERS AND 141 ADDITIONAL PAPERS OF THESE PAPERS USED PGC RESULTS. SO AS YOU CAN SEE, THIS PGC HAS UNLOCKED COMMON VARIANT GENOMICS IN INCREASING NUMBERS OF DISORD ARERS. THE GENERAL CONVERGENCE WITH SOME ASPECTS OF RARE VARIATION, AND THE FINDINGS HAVE NOT BEEN EXTENSIVELY NAILED THE BIOLOGY, THAT'S GOING TO TAKE A LONG TIME. PROVIDE TDZ THERE ARE IMPROVING RESOURCES THAT MIGHT HELP THEM; THE EARLY RELEASE OF ALL SUMMARY DATA IS BEING WIDELY TAKEN UP AND IT'S ROCKET FUEL FOR PSYCHIATRIC RESEARCH FOR THE NEXT 20 YEARS AND PROVIDING ALSO, THAT. AND MORE ATTENTION ALTHOUGH MORE ATTENTION IS NEEDED IN EUROPEAN ANCESTRIES AND THAT'S IT. THIS IS WHO FUND THIS IS WORK AND ALL THE PEOPLE WHO YOU CAN'T SEE ARE MOST OF THE AUTHORS UNT DOUBTLY MOST OF THE AUTHORS MISS BECAUSE THERE'S ONLY 500 THERE. [ APPLAUSE ] >> SO, THANKS ON THE BIOMARKER ISSUE, YOU'RE RIGHT, I WAS THINKING ABOUT TREATMENT BIOMARKERS BUT I THINK PRS A RISK BIOMARKER AND I THINK TERRY WOULD BE USED TO THE NEED FOR MARKERS, MANY--THE SUM OF MANY DIFFERENT MARKERS, RIGHT BUT WE DON'T YET HAVE OUR EQUIVALENT OF SMOKING AND LDL LEVELS BUT IT'S TRUE, IT REALLY IS PROGRESSION IN THAT REGARD. I ALSO OF-1 BRIEF RETRACTION, YOU DIDN'T HAVE THE LONGEST RIDE HERE BUT I THINK DAN STEIN IS HERE FROM SOUTH AFRICA. SECOND PLACE, OKAY? OUR NEXT SPEAKERS ARE GOING TO TALK ABOUT THE AUTISM SEQUENCING CONSORTIUM AND SPATIAL, TEMPORAL AND LEVEL DISORCHS AND THAT'S JO BUXBAUM FROM ICAHN SCHOOL OF MEDICINE AT MOUNT SIGH NIGH AND THEN JEREMY WILLSEY UNIVERSITY OF CALIFORNIA, AND YOU WILL WORK OUT YOUR TANGO, RIGHT? >> RIGHT I WILL TALK ABOUT SOME OF OUR EARLY WORK AND SUCCESSES AND THEN I WILL TURN IT OVER TO JEREMY TO TALK ABOUT GENE EXPRESSION AND TYPES TO GENETIC DATA. SO WITHOUT FURTHER ADO, I WILL SHUT THE SLIDES OFF. YOU'RE WELCOME. LET ME TRY TO UNDO THAT. LET'S SEE. IT WAS ON THE LAST SLIDE. SO ASY HAS A LOGO WE DESIGNED SO, NOT AS CATCHY, I DON'T THINK AS PGC LOGO, BAIRVEGLY THE PIs ARE LISTED HERE, AND THE EXECUTIVE COMMITTEE, INVOLVES THE PIs BUT ALSO THOMAS LANNER AND AUDRY AND PEOPLE HAVE BEEN INVOLVED SINCE TD VERY FIRST DAYS OF THE ASC AS I WILL SHOW YOU. AND THEN WE HAVE STANDING COMMITTEES AROUND STATISTICS, DATA, PHENOTYPES, MEMBERSHIP AND AUTHORSHIP AND THEN WE HAVE A LOT OF WORKING GROUPS. WHICH I WILL GO THROUGH VERY QUICKLY. AND JUST TO KIND OF GIVE YOU A BIT OF A CONTEXT ABOUT HOW THE ASC CAME INTO BE AND WE HAD SOME--WE HAD A MEETING IN 2010 WHICH WAS SNOWED OUT AND 1 OF THE PEOPLE IN THE AUDIENCE HERE FLEW IN BECAUSE HE DIDN'T KNOW THERE WAS SNOW FALL IN NEW YORK AND HE LANDED AND HEADED HOME AND WE RESCHEDULED FOR APRIL 7th AND THAT WAS THE MEETING WHERE I SAID WE HAVE TO PUT OUR DATA TOGETHER AND DO IT PROSPECTIVELY. THERE'S NO POINT OF HAVING A BUNCH OF SMALL SEQUENCING STUDY, THIS IS ABOUT WHOLE EXPOEM SOQUENCING. THERE'S NOT A LOT OF REASON TO DO BIG STUDIES IF THEY GO ON TO PUT THEM TOGETHER LATER. SO THIS IDEA OF PERSPECTIVE DATA SHARING BECAME A FOUNDATIONAL CONCEPT OF THE ASC. WE HAD A MEETING IN 2010 OF THIS BUILDING AND EQUALLY ATTENDED WERE STAKEHOLDERS ACROSS THE SPECTRUM FOMLYS, ADVOCACY GROUPS, GENETICS PHENOTYPING AND THERE WAS A WHITE PAPER THAT CAME OUT IN 2012 THAT DESCRIBED THE FIELD AND TALKED ABOUT THE IDEAS THAT THE ASC WAS GOING TO BE UPON. THE CONTEXT AGAIN WAS THAT THE IAC AND OTHERS REALIZE THAD GENETICS WAS A MAJOR UNDERSTANDING AS PART OF OUR BIOLOGY AND THERE WAS THIS ASPIRATIONAL GOAL, LONG-TERM OBJECTIVE OF FINDING GENETIC CAUSE AND 50% OF THE PEOPLE AND WE KNOW THAT'S NOT TRUE BUT WE CAN EASILY GET TO A RARE VARIATION FOR EXAMPLE, 50% OF THAT RISK, AND FOR COMMON VARIATION, A BIT FURTHER AWAY. AS I SAID A LOT OF PEOPLE WERE DOING SMALLER SCALE STUDIES AT THAT TIME AND--THIS IS SOMEBODY ELSE'S SLIDE BECAUSE IT'S AN ANIMATED AND THESE ARE THE 4 STUDIES THAT CAME OUT IN 2012 TAKING ADVANTAGE OF THE FACT THAT DENOAFEE--DENOVAE VARIATION IS RARE ENOUGH IN THE POPULATION AND COMMON ENOUGH IN THE AUTISM THAT WE COULD MAKE SIGNIFICANT FINDINGS WITH MODERATE SAMPLE SIZES. SO THIS IS WHAT WE CALL THE PILOT PART OF THE ASC. THAT'S HOW I REFER TO IT. AND THOSE PAPERINGS AND THE WHITE PAPER MADE THE FOLLOWING POINT WHICH WAS THAT NO MATTER--WE HAD ABOUT A THOUSAND SAMPLES OF THAT, NO MATTER THE MODEL OF GENES THAT COULD BE DISCOVERED USING SUCH ASBESTOS PROACHS AND WE THINK IT'S A THOUSAND ACTUALLY, YOU KNOW WE WERE VERY MUCH IN THE EARLY, EARLY LYNNIAN PHASE AND SO WE REALLY NEED TO PUT SAMPLE SIZES TOGETHER TO GET APPRECIABLE FINDINGS AND THIS IS A MODEL OF GREATER THAN MORE THAN 2--THIS IS NOT--THIS IS NOT USEFUL FOR LARGER SAMPLES AS A CUT OFF. SO JOWR DISCOVERY MIGHT LOOK MORE LIKE THIS IF YOU JUST USE DE NOVO INFORMATION. SO WE CAN DO BETTER THAN THAT. SO DATA HAS TO COME TOGETHER AND COME TOGETHER AS A FIRST STEP RATHER THAN AS AFTER AFTERTHOUGHT. AND THIS IS THE MEETING THAT WASN'T SNOWED OUT. THERE WERE A LOT OF PEOPLE WHO RAN OFF QUICKLY. A LOT OF GRIM FACES. THOMAS LEONARD WAS THERE LOOKING EXTRAORDINARILY HAPPY BECAUSE HE LIKED THIS IDEA AND WAS VERY SUPPORTIVE OF IT THAT YOU KNOW WE SHOULD WORK TOGETHER PROSPECTIVELY AND HE PRETTY MUCH SAID THAT THAT HE WOULD EXPECT PEOPLE TO DO THIS IF THEY EXPECT FUNDING. HE DIDN'T QUITE SAY THAT PARTIC PAITIONZ WAS REQUIRED BUT YOU--PARTICIPATION WAS REQUIRED BUT THE RESOURCES WILL GO TO THINGS THAT ARE SUCCESSFUL AND THIS IS 1 OF THOSE THINGS OF THE OTHER FUNDERS WERE IN THE ROOM AS WELL AND MANY FOUNDATIONS. AND 1 OF THEM DEBORAH HILDEBRAND SAID WHO IS WITH AUTISM SPEAKS AND ARE FOCUSED ON GET, SHE SAID I SEE HISTORY BEING MADE. THIS IS THE FIRST TIME SOMEBODY GOT TOGETHER AND SHARE EXPERIENCES IN ADVANCE OF PUBLICATION. SO THE GRIM FACE IS BECAUSE IT WAS AN UNUSUAL CONCEPT AND HAD TO BUILD AN INFRASTRUCTURE WHERE PEOPLE FELT A, THEY COULD GET CREDIT FOR WHAT THEY PUT IN AND B-THAT THERE'S NOT GOING TO BE TAKEN ADVANTAGE OF AND ALL THE DATA LIVES IN 1 PLACE AND IT'S PROTEBLGHTED AND PEOPLE--ANYBODY CAN ANALYZE IT BUT ONLY WITH PERMISSION AND SO FORTH SO THAT TOOK DOING OTHER CONSORTIA LIKE THE PGC LAID AGREEMENTS ACROSS INDIVIDUALS RATHER THAN ACROSS UNIVERSITIES WHICH IS A DIFFERENT THING. SO, IN 2013 WE GOT OUR FIRST NIH GRANT AND IT WAS FUNDING INFRASTRUCTURE THAT WAS FUNDING AROUND JUST BASICALLY THE PIPELINES, AND THEN BRINGING IN STATISTICAL METHODOLOGY WHICH IS A BIG PART OF THE ASC AND I WILL TALK ABOUT THAT IN A MINUTE TO IDENTIFY AUTISM GENES AND THERE WAS SOME MONITOR THAT'S WENT INTO SEQUENCING. SO AS I'LL SHOW YOU WE GOT A LOT OF SEQUENCE DATA NOT FROM ANY ASC, DIRECT INDIFFER. SO THE ASC FUNDING AS I SAID IS PRIMARILY TALK TARGETED TO ANALYTIC IT IS METHODS, GENE DISCOVERY AND THOSE TYPES OF STUDIES, NOTHING IS REALLY--CURRENTLY NOTHING GOES TO SEQUENCING OR DATA GENERATION. AND SO THERE'S VERY, VERY STRONG STATISTICAL CORE OF PEOPLE ACROSS THE WORLD. THE LEADERS ARE HERE, MARK, CATHY AND BERNIE ARE HEADS OF THAT COMMITTEE BUT THEY HAVE CLOSE RELATIONSHIPS WITH EMORY AND ALL SORTS OF SITES. AND SO THEY CAN THINK ABOUT STATISTICS, METHODOLOGY, INTEGRATING DATA, TRANSCRIPTOMICS WHICH YOU WILL HEAR ABOUT FROM JEREMY IN A FEW MINUTES AND PUTTING PEOPLE TOGETHER, THINKING ABOUT DATA YOU WILL HAVE STRONG EMERGENTS. PIFORT PAPER WE WORKED ON WITH THIS GROUP TO PUT TOGETHER DATA SETS TO MAKE A LARGER ANALYSIS. AND DONNA WHORLING AND COLLEAGUE VS--COLLEAGUES HAVE A PAPER. THE 5 GRANTS THAT WENT OUT TO ANALYZE THE SIMONS GENOMES 3 WENT TO ASC SITES AND THE ASC SITES AGREED TO HAVE A CLAIBERATION AND THE ASC INFRASTRUCTURE SUPPORTED THAT AND THESE ARE THE WORKING GROUPS THAT ARE IN THAT PAPER, IS--I'LL SHOW YOU A SLIDE FROM IT IS IMPRESSED, SO VARIATIONS AND PHENOTYPES AND HOW TO INTEGRATE THAT INFORMATION. X-LINKED VARIATION, STRUCTURAL VARIATION, INTEGRATING WGS, AND ALL THOSE ARE IN THE ASC THINKING ABOUT STRONG ANALYTERAL APPROACHES TO THIS AND THEN APLOIFING THEM. -- APPLYING THEM. THIS IS A DE RUBEIS PAPER THAT GOT TAKEN BACK BUT PUT AWAY, AND CHROMATIN REMODELING, SEEMS TO BE A MAJOR, MAJOR PART OF THE AUTISM RISK. THIS IS THE PAPER WHERE STEF AND OTHERS PUT TOGETHER THE SAMPLE CELLS, THE ASC AND THE SSC AND NOTABLY THE 2 MAIN FINDINGS WERE THE--HOW DID YOU CALL IT, NO-SH IT-SHERLOCK CONCEPT AND THEN THE MONITOR DOMAIN. SO THIS IS THE PAPER THAT'S THE PRESS AND YOU HEARD THE NAME MENTIONED AROUND THIS EARLIER IN THE DAY SO THIS IS JUST 1 EXAMPLE OF A GENE WHERE THE MUTATIONS THAT YOU FIND IN ASD ARE ARE MOW ZAYAIC MUTATIONS MANY OF THESE THAT WE'RE SEEING ARE IN THIS BECAUSE THEY'RE--THEY'RE NOT REALLY COMPATIBLE WITH THE HEALTHY DEVELOPMENT OR MUTATIONS SOMEONE SAYS THIS IS A WAY BACKING OFF ON THE DELETERIOUSNESS AND THESE ARE MUTATIONS AND SHOWS IT EFFECTS THE MODULES THE AUTISM GENE, THIS IS THE ASDID, AND AND ALL THE PUBLISHED AND ASC AND DOING THINGS KIND OF GETTING SENSE ABOUT THE DIMENSIONS OF THIS AUTISM AND THIS IS 1 OR THE OTHER, WHAT THEY LOOK LIKE, OTHER GROUPS DOING THE SAME THING BUT THIS IS AN AREA WHERE WE CAN KIND OF LEARN A LOT ABOUT NOT ONLY THE RELATIONSHIP FOR THE MORBID DISORDERS BUT ALSO BEGIN TO ASK QUESTIONS ABOUT WHAT IS THE RELATIONSHIP BETWEEN COGNITION IS SOCIAL COGNITION FOR EXAMPLE. THIS IS AN ACTIVE WORK GROUP AND HIGH WILL HIGHLIGHT TADA, WITH COLLEAGUES WITH THE TOOL CALLED IT, AA AND IT'S BEEN TRANSFORMATIVE EVEN AT THE OUTSET OF THE ASC, BECAUSE TADA IS A MEANS OF PUTTING TOGETHER VARIOUS CLASSES OF SIGNALS AND WAITING THEM BY THEIR EFFECTS SIZE AND THE FIRST VERSION THAT BECAME SO USEFUL FOR US, INTEGRATES DE NOVO AND INHERITED DATA AS WELL AS CASE CONTROL DATA. SO THE ASC WAS CONFRONTED WITH THE ISSUE THAT ASCERTAINMENT IN EVERY SAMPLE WAS DIFFERENT. SOME OF IT WAS CASE CONTROL, SOME OF IT WAS FAMILY BASED DATA, SOME OF IT WAS LARGER PEDIGREES AND WE WANT TO INTEGRATE ALL DATA AND TADA IS A MEANS THAT ALLOWS YOU TO BRING IN ALL DATA IRRESPECTIVE OF HOW IT WAS ASCERTAINED SO IT WAS THE BASIS OF THE PAPER, SO THIS' BEEN A NEW FLAVOR WHICH IS PART OF STEPH SANDERS PAPER AND I WILL SHOW YOU A BRIEF INTRIEWKS OF THOSE. SO TADA AS BECOME A WORK HORSE FOR THE GROUPS TO TACKLE THE SAME DATA SET FROM OTHER MERSPECTIVES. SO THE WGS IS SOMETHING THAT XIN PUT TOGETHER AND IT'S A FUNDED STUDY AND IT TAKES CLASSES OF VARIATION AND ANNOTATION OF NONCOATING SEQUENCE AND ATTRIBUTES WEIGHTS TO THEM AND THOSE THAT ACTUALLY SHOW SIGNIFICANT WEIGHT, YOU CAN ADD BACK TO THE TADA MODEL AND IT'S EARLY STAGES BECAUSE AS YOU WILL HEAR LATER FROM STEPH STANDERS, MODELS ARE UNDERPOWERED BUT THEY'RE GETTING BIGGER WITH HEART AND OTHER THINGS BUT PUTTING THE DATA TOGETHER WITH OTHER INFORMATION FROM OTHER SAMPLES AND PUTTING IN GENOME DATA FROM OTHER SAMPLES WHICH ALSO VARMIS FLANKING INFORMATION OR HAVE NONCODING INFORMATION, YOU BEGENERATED TO SEE THINGS LIKE THIS WHERE THIS GENE KIND OF CREEPS THE TOP OF THE LIST AND BECAUSE OF A LOSS IN FUNCTION MUTATION AND THIS SENSE OF YOU HADITATION AND DE NOVO CMV,--SSNV AND DOMAIN SO I THINK WE'RE AT A POINT NOW WHERE AT LEAST IN THE CONTEXT OF KNOWN AUTISM GENES WE CAN BEGIN WITH PUTTING TOGETHER THE CURRENT GENOMES WE CAN BEGIN TO MAKE SENSE OF THE FLANKING SEQUENCE BUT WE HAVE TO PUT THEM TOGETHER, I THINK. TADA-X, WE DON'T HAVE PUBLICATIONS BECAUSE THE X OF A BIT OF A SURPRISE FOR US. WELL, FOR ME, NOT FOR ANYBODY WHO'S THINKING ABOUT IT WHO WILL THINK ABOUT THESE THINGS, 2 THINGS ABOUT THE X ARE 1 MOST OF OUR PATIENTS ARE MALE AND MOST OF OUR POWERFUL FINDINGS OF THE GENES ARE DE NOVO. SO WE DO A GOOD JOB FINDING X GENES IN GIRLS BECAUSE YOU CAN HAVE EXPLURIBUITATION WHICH IS REASONABLE PROBABILITY RIGHTS AND MOST X-MUTATIONS COME FROM THE DATA. AND THESE ARE PRIMARILY BY GIRLS. BUT THAT'S BECAUSE THEY'RE IN THE X-CHROMOSOME FROM DAD THAT HAS THE MUTATION. DID I GET IT RIGHT? SO IT'S BEEN AN INTERESTING THENG AND WE'RE READY TO SUMMARIZE WHAT WE KNOW ABOUT POTENTIAL VISIBILITY WITH TADA-X. SO JUST TO FINISH UP, ASC TODAY HAS ABOUT 50 RESEARCH GROUPS, 550 ACTIVE MEMBERS. WE RECENTLY GOT 5 YEARS OF FUNDING TO CONTINUE THIS WORK, THROUGH THE NIMH. SO LET'S REMARKS SIEWMEUR BASE LINE IS 20,000 EXOMES RIGHT NOW. WE WILL PRODUCE AND ANALYZE AN ADDITIONAL 50,000 TO GET THIS NUMBER. WE WANT TO SPEND MORE TIME GETTING TO THE MORE DIFUL THINGS, MISSENSE VARIATION OR INHERITED VARIATION, THERE'S LOTS OF FUNCTION, DE NOVO LOSS OF FUNCTION AND WE WANT TO DO BETTER AND ULTIMATELY INTEGRATE NONCODING VARIATION AND ALSO COMMON VARIATION THROUGH SNPs. THIS IS AN ELABORATION OF THAT WHICH I PROBABLY WILL NOT GO INTO IN GREAT DETAIL BUT I WANT TO TALK ABOUT THIS COLLABORATION ALL THE WEST DATA COMING IN IS FUNDED BY OTHER SOURCES AND A LARGE PART IS FUNDED BY NHGRI AS PART OF THE ASC, SO THIS HAS BECOME AN NIMH-NHGRI AND ASC AND THEY WERE FUNDED IN THE PAST BY THE CCG, AND EVEN WHEN THEY WERE A SEQUENCING CENTER, THEY WERE TAKEN BY THEM TO RUN THROUGH AND AS PART OF THE CKDG WHICH HAS A NEURODEVELOPMENTAL FOCUS FOR THE FIRST TIME IN MY LIFE, MARK COMES TO ME AND SAYS WE HAVE CAPACITY WE NEED MORE SAMPLES. SO THE BROAD IS RUN TO MY ESTIMATE NOW, WELL, OVER 12,000 SAMPLES FOR THE ASC. THROUGH THESE MECHANISMS, 8000 ARE UPLOADED RIGHT NOW. YOU CAN SEE THAT THIS COMES FROM ALL OVER THE WORLD. THEY ALL HAVE OF HAD-THEY ARE ALL CONSENTED TO BE SHARED WITH DB GAP. THE 1S THAT AREN'T ARE BECAUSE THEY COME FROM SEQUENCING AND THEY LIVE ON THEIR SERVER, BUT THE THINGS THAT COME TO THE ASC, BY DEFINITION, INSIST THEY HAVE A DUL THAT SAYS THAT DATA CAN BE SHARED. AND SO, THIS IS ABOUT 12,000 SAMPLES. SO ALL THESE BATCHES, BATCHES 1-11, 11-12, YOU KNOW, THE BROAD, MARK STEVENS AND OTHERS HAVE TAKEN THESE THROUGHOUT THE WORLD AND GIVEN THEM BACK TO THE ASC AND'S ABOUT 12,000 TO DATE. SO THIS HAS BEEN A VERY, VERY SUCCESSFUL NHGRI AND NIMH COLLABORATION. THIS IS A SAMPLE THAT'S COMPLETE, THIS WAS A TASK SAMPLE AND THE AUTISM GENOME PROJECT WHERE SEVERAL SITES GOT TOGETHER AND SDZ THEY WOULD DO THE STEEP PHENOTYPING OF FRIED STEAK OS, THEY ARE IN THE TASK SAMPLE, THOMAS WAS SUPPORTIVE OF THAT, HE MADE SURE THAT ALL THE SAMPLES COULD MAKE IT TO THE REPOSITORY BECAUSE WE DIDN'T HAVE FUNDING FOR THAT, AND THEN, HE PUT THE BLOOD DERIVED DNA ASIDE AND HE SAID WHEN THE TIME COMES AND SEQUENCING WILL MAKE THAT AVAILABLE AND SO IT CAME BACK TO THE ASC FOR SEQUENCING. IS THAT FLASHING BECAUSE I'M AFTER? 29,600 SPHEVENT SAMPLES, THE DATA GETS BETTER, AND THIS IS A TRUE FREEZE, WE WILL RUN ANALYSIS ON THIS AND PUBLISH PAPERS AND JOIN US AND JUST AS A PLUG, WE ARE VERY EXCITED TO BE USING HAIL AND YOU WILL HEAR ABOUT HAIL AT 5:30. I WILL STOP THERE AND TURN IT OVER TO JEREMY WHO IS ACTUALLY DOING IMPORTANT WORK ON THE ASC. I'M JUST THE FIGURE HEAD SOMETIMES. [ APPLAUSE ] >> GREAT THANK YOU FOR THE EASY INTRODUCTION INTO THE GENETICS. IT SAVES ME TIME. I WANT TO THANK THE NIH, NIMH AND COMMITTEE ORGANIZERS FOR ALLOWING ME TO BE THERE AND THE ASC FOR GIVING THIS TALK. HAS ACCELERATED IN THE LAST COUPLE YEARS AND NOW WITH THE PAPERS IN 2015 WE'RE UP TO 65 STRONGLY ASSOCIATED AUTISM GENES SO THE NEXT CHALLENGE HERE IS TO USE THESE GENES TO UNDERSTAND SOMETHING ABOUT THE UNDERLYING BIOLOGY. NOW OF COURSE THIS IS NOT OUTLYING THE PAST, WE ARE TRYING TO OUTLINE THE CHALLENGE. SO FESTER IS HETEROGENEITY. WE HAVE A THOUSAND INDIVIDUAL GENES THAT CONTRIBUTE TO RISK. THE SECOND LEVEL IS GENETIC PLIE O TROPY, SO THESE GENES HAVE DIFFERENT CONTEXT, CELL THEY'RE IN, REGION OF BRAIN AND IT MAKE ITS CHALLENGING ABOUT WHAT'S KNOWN IN THE LITERATURE IN THE GENES AND A SYNTHESIZE A CO HERENT BIOLOGICAL PROCESS. AN UNDERSTATEMENT AND BRAIN DEVELOPMENT IS COMPLEX AND SO UNDERSTANDING HOW THESE ARE PERTURBED WITHIN AN INDIVIDUAL CELL. WE NEED TRANSLATE THIS TO THE CIRCUIT LEVEL PHENOTYPES. AND FINALLY WE KNOW THAT EVEN WITHIN A GIVEN GENETIC RISK FACTOR THERE ARE ARE DIVERSE OUTCOMES IN THE PHENOTYPE SO TOGETHER THIS MAKE ITS QUITE CHALLENGING AND THEN WHEN WE LAYER ON THINGS LIKE GENETIC BACK GROUPEDS, ENVIRONMENTAL EFFECTS, STOIKASTIC EVENTS THIS BECOMES A DIFFICULT PROBLEM TO GRAB. SO 1 OF THE SOLUTIONS AND APPROACHES SOLVING THE IDEA OF CONERG ENSEL SO BASICALLY THAT IA LARGE NUMBER OF JOONS WILL CONVERGE ON A SMALLER NUMBER OF PATHWAYS. AND SO IF WE IMAGINE THAT, YOU KNOW EACH 1 OF THESE RAYS HERE AS A GENE AND WE THINK THAT ALONG WITH RAY WE HAVE A SPECTRUM OF DIFFERENT PHENOTYPES MAY BE OBSERVED DEPENDING ON CONTEXT AND THE CELL, MAYBE SAME CELLS THIS WOULD BE A MYRIAD OF EEIVELGTS YOU WOULD SEE FROM KNOCKING OUT A GENE SO WHAT WE'RE LOOKING FOR ARE WHERE ARE THESE RAYS POINTING TO SO THIS IS AN OOH DEALIZED EXAMPLE. WE ARE NOT XESKTING THEM TO POINT AT THE SAME PLACE BUT IF WE LOOK AT WHAT THEY'RE POINTING TO THIS GIVES US TRACTION TO UNDERSTANDING THE UNDERLYING BIOLOGY. SO OBVIOUSLY TO DO THIS TYPE OF ANALYSIS, WE NEED THE RIGHT DATA SET. SO THESE DATA SETS NEED TO BE FROM THE BRAIN, IDEALLY HUMAN BRAIN AND SPAN THE COURSE OF DEVELOPMENT SO WE CAN BEGIN TO IDENTIFY SPECIFIC SPATIAL, TEMPORAL POINTS OF CONVERGENCE ACROSS BRAIN DEVELOPMENT. SO THIS WAS GENERATED AS PART OF THE CONSORTIUM SO THIS DATA SET SPANS 15 PERIODS OF HUMAN BRAIN DEVELOPMENT STARTING EARLY ON AND 4-6 WEEKS CONCEPTION AND ALSO THE WAY TO 60 + YEARS OF AGE AND THIS IS GENE EXPRESSION AND 16 REGIONS OF THE BRAIN, AND THESE ARE HERE SHOWN IN RED AND 5NY O CORTICAL REGIONS SO THE QUESTION WE ASKED SEVERAL YEARS AGO NOW IS CAN WE USE THIS DATA SET TO IDENTIFY SPATIAL AND TEMPORAL CONVERSIONS AMONG AUTISM RISK GENES SO HOW WE CHOSE TO DO THIS WAS TO SUBSET THE DATA IN DIFFERENT SPATIAL AND TEMPORAL WINDOWS. THE IDEA BEING THAT WE CAN ANALYZE EACH WIN WINDOW OR CONVERGENCE AND IDENTIFY WHETHER OR NOT THERE WERE SPECIFIC POINT WHERE IS YOU HAD STRONG CONVERSIONS. SO WE FIRST PUT THE DATA UP INTO 4 CLUSTERS OF REGIONS, SO WE HAVE 2NY O CORTICAL CLUSTERS HERE AND WE HAVE 2 NONNY O CORTICAL CLUSTERS AND WE ALSO SPLENT THE DATA INTO 13 WINDOWS OF TIME AND WE START EARLY ON, AND PERIOD SAYS 1-3 AND SHIFTED OVER 1 WINDOW AT A TIME. AND SO WE CHEESE TO DO THESE 3 PERIOD WINDOWS INEREDDER ON HAVE ENOUGH DATA FOR A ROBUST HOW WE THEN APPROACH THE ANALYSIS WAS CO EXPRESSION NETWORK SO LOOKING AT CORRELATION BETWEEN GENE EXPRESSION PROFILES AND HERE WE BUILD OUR CO EXPRESSION NETWORKS AROUND THE SET OF MOST STRONGLY ASSOCIATED AUTISM GENES SO THEA THE TIME THIS WAS 9 HIGH CONFIDENCE GENES AND SO REMEMBER WE HAVE 4 BRAIN REGIONS AND 13 WINDOWS OF TIME SO THESE ARE 252 DIFFERENT NETWORKS AND WITHIN THESE CO-EXPRESSION NETWORKS WE TESTED FOR ENRICHMENT OF AN UNRELATED SET OF AUTISM GENES SO THESE ARE THE LOWER SET OF AUTISM GENES. SO WHAT WE'RE LOOKING FOR HERE ARE SPECIFIC POINTS OF ENRICHMENT THAT WOULD INDICATE THE SPATIAL AND TEMPORAL FEATURES OF AUTISM PATHOGENESIS. THIS IS THE FIRST RESULT PUBLISHED 4 YEARS AGO NOW, HERE WE HAVE THE 4 GROUPS OF GRAIN REGIONS ON THE LEFT, DIFFERENT WINDOWS OF TIME ON THE BOTTOM AND HERE'S THE NEGATIVE LOG OF THE 10 TO THE P-VALUE AND SO HIGHER NUMBERSER MORE SIGNIFICANT AND WHATTA JUMPED OUT RIGHT AWAY IS THE PEAK OF STRONG ENRICHMENT IN THE PREFRONTAL CORTEX, SO THIS IS BASICALLY 10-TWEF CONCEPTUAL WEEKS. WE ALSO HAD THE SECOND POINT OF ENRICHMENT SO THIS IS IMMEDIATELY AFTER BIRTH TO 6 YEARS OF AGE AND THIS IS THE IN THALAMUS AND CELLBELAR CORTEX. SO AT THE TIMES I MENTIONED WE HAD 9 HIGH CONFIDENCE GENES AND THEN 116 PROBABLE ASD GENES SO NOW THAT WE'RE UP TO THE SET OF 65 HIGH CONFIDENCE AUTISM GENES WE ASK THE QUESTION OF WHETHER OR NOT WE COULD REPLICATE THE SIGNAL IN AN NPT WAY SO THE SHORT ANSWER IS REQUESTS WE REPLICATE THE SIGNAL BUT I WILL TALK YOU THROUGH THE ANALYSIS A BIT. SO FIRST OFF HERE, WE USE THE FALSE DISCOVERY RATE THRESHOLDS AS THROUGH THE PAPER. SO, AND THEN WE ACTUALLY EXCLUDED ALL THE GENES THAT WERE USED IN THE PHASE 1 ANALYSIS SO NONE OF THE HIGH CONFIDENCE GENES OR PROBABLE ASD GENES ARE INCLUDED HERE SO USING THESE SAME THRESHOLDS THAT LEADS HICONFIDENCE GENES AND 269 PLOBBABLE GENES SO SIMILAR IN THE INITIAL ANALYSIS AND WE SEE THAT WE REPLICATE THE PREFROBTAL CORTEX ENRICHMENT HOWEVER WE DO NOT SEE A REPLICATION OF THIS, THALAMUS AND CEREBELLAR CORTEX AND WE'RE DIGGING INTO THIS SAMPLE IN THE TIME PERIOD FOR THESE BRAIN REGIONS AND THIS IS ENCODE, THEY'RE CONTINUING TO INCREASE RESLIEWGZ OF THESE DATA SETS, SO TD NEXT WE WILL DIG INTO THAT QUESTION DEEPER. SO THE OTHER THING THAT WE CAN DO IS LOOK FOR ENRICHMENT OF CELL TYPE SPECIFIC MARKERS. SO BASICALLY THESE ARE GENES THAT ARE HIGHLY SPECIFICALLY EXPRESSED IN A CELL TYPE AND THEY INTEGRATE ENRICHMENT IN A CELLS WITHIN A NETWORK IF YOU HAVE MARKET GENES FOR PARTICULARLY CELL TYPES OVERREPRESENTED AND WE OBSERVE THAT WE SEE THESE MARKER GENES FOR GLUTAMATERGIC NEURONS AND THESE ARE THE DEEP LAYER 1S, THE LIER 5-6. SO FAST FORWARD A COUPLE OF YEARS THERE'S A LONGER LIST OF GENES BUT THERE'S ALSO NEW GENERATED DATA THAT'S LOOKING AT PROFILING MANY THINGS INCLUDING SINGLE CELL TRANSCRIPTOMES IN THE HUMAN BRAIN AND THE NEXT QUESTION IS CAN WE LEVERAGE THE SINGLE CELL DATA TO REFINE HIPOGHTD SIS ABOUT THE SPECIFIC CELL TYPES INVOLVED AND ALSO THE SPATIAL AND TEMPORAL DYNAMICS. SO YES, THIS ANALYSIS WAS DONE BY NENAD SESTAN'S GROUP, WE'RE LOOKING AT 2 TIME POINTS, 6-SIT CONCEPTUAL WEEKS, SO FETAL DEVELOPMENT AND HERE CELL TYPES ARE DEFINED AGAIN BASED ON THE CELL TYPE SPECIFIC MARKER GENE, SO AFTER--GROUP THEM INTO 26 DIFFERENT CELL TYPES, WHICH ROUGHLY CORRESPONDS TO 13 CLASSES OF CELLS AND THIS IS BASICALLY REPRESENTATION OF HOW WE CAN THINK ABOUT PARSING THESE CELLS. SO THE FIRST TASK I WILL POINT OUT THIS WAS DONE BY BERNIE AND KATHRYN'S GROUPS. SO HERE, THEY'RE DEFINING EXPRESSION OF THE 6200 GENES OF PAST QC AND I'LL ALSO POINT OUT THAT 29 OF THE 69 AUTISM GENES ARE REPRESENTED HERE IN AND HERE THE FIRST TASK IS TO DEFINE EXPRESSION. SO HERE'S A BIENAR EXPRESSED NOT EXPRESSED AND THIS DEPENDS ON THE EXPRESSION LEVEL WITH A--WITHIN A GIVEN CELL AND HOW THAT COMPARES TO IMMEDIATE EXPRESSION OF THAT GENE ACROSS ALL THE DIFFERENT CELLS SO HERE YOU'RE SAYING IF IT'S EXPRESSED, HIGHER THAN MEDIAN AND ALL THE DAILY BASIS THEA NOT EXPRESSED. THIS IS A SIMPLE MEASUREMENT. IT HELPS US GET AROUND THIS ISSUE OF THE NOISE AND SINGLE CELL RNA SEQUENCING DATA. AND SO, THE FIRST QUESTION THAT THEY ASKED IS WHETHER OR NOT, THERE IS ENRICHMENT OF AUTISM GENE EXPRESSION IN PARTICULAR GROUPS OF THESE CELLS. SO HERE WHAT WE'RE SHOW NOTHING IN HEAT MAP IS THE FRACTION OF CELLS EXPRESSING THE GENE, SO THESE SQUARES THAT ARE HIGHLY RED MEANS THAT A LOT OF THE CELLS OF THAT TYPE ARE EXPRESSING THIS GENE. SO THE FIRST THING THAT JUMPS OUTED IS THE BAND HERE AT THE BOTTOM WHERE HAVE YOU AN EXTREMELY HIGH FRACTION OF THE AUTISM GENES OR HIGHLY EXPRESSED SO THEY'RE EXPRESSED MOST OF CELLS OF THIS IMMEDIATE PROGENITOR CELL TYPE AND WE CAN QUANTIFY THAT WITH A P-VALUE WITH THE AUTISM GENES TO THE NONAUTISM GENES IN XERM--TERMS OF EXPRESSION GENES AND WE SEE THOO AIAN SIGNIFICANT RESULT. THE SECOND BAND THAT'S INTERESTING IS IN THE 20-18 CONCEPTION WEEK IN THE EXCITATORY NEURONS SO THIS IS--YOU CAN CLUSTER IN DIFFERENT SUBGROUPS SO THIS IS JUST A CLUSTER OF THEM HERE. AND AGAIN WE SEE THAT YOU KNOW IT'S NOT AS EXTENSIVE AS THE POINT OF ENRICHMENT IN THE INTERMEDIATE PROGENITORS BUT WE SEE THIS POINT WHERE A REASONABLE NUMBER ARE HIGHLY EXPRESSED. SO DIGGING INTO THIS FURTHER AND LOOKING SPECIFICALLY AS THESE INTERMEDIATE PROGENITORS AND THESE EXCITATORY NEURONS WE CAN ASK THE QUESTION, WHAT IS CO EXPRESSION AMONG THESE GENES LOOK LIKE IN THE CELLS. HE WERE CO EXPRESSION IS BASICALLY HOW OFTEN ARE THEY ON IN THE SAME CELL TOGETHER? AND WHAT WE SEE IS THAT JUMPS OUT RIGHT AWAY IS THAT THESE INTERMEDIATE GROUP OF GENES THAT ARE CO EXPRESSED EXPR THE REALLY INTERESTING THING IS IF WE COLOR THESE GENES ON THE BOTTOM BY WHETHER OR NOT THEY'RE DNA REGULATORY GENE OR A NONDNA REGULATORY GENE, SO A LOT OF THESE ARE SYNAPTIC GENES OR GENES THAT FUNCTION AT OTHER POINTS IN THE NEURON. AND HERE WHAT WE SEE ACTUALLY IS THAT A LOT OF THE PROAMA TIN MODIFIERS ARE CO-EXPRESSED IN THIS TIME POINT. SO IF WE COMPARE AUTISM TO NONAUTISM, IF WE LOOK AT REGULATORY VERSUS NONREGULATORY WE ARE--SEE THE THAT THE REGULATORY ARE HIGHLY EXPRESSED. HERE WE SEE THERE'S MORE REPRESENTATION OF THESE NONDNA REGULATORY GENES AND HERE IT'S TRENDING TOWARDS SIGNIFICANCE THAT WE HAVE THIS HIGHER EXPRESSION LEVEL IN GENERAL OF THESE NONREGULATORY GENES SO INTERESTINGLY THIS SUGGESTS THESE CHROMATIN MODIFYING GENES, DNA REGULATORS MAY BE COMING ON LINE EARLIER AND MAY BE REGULATING EXPRESSION OF THESE GENES AND NOW OBVIOUSLY THAT ISN'T BEEN MADE HERE BUT A LOT OF PEOPLE HAVE POSITIVE IN THE PAST, IT'S INTERESTING HYPOTHESIS TO FOLLOW UP. SO THAT WILL JUST FINISH WITH A QUICK SUMMARY. YOU KNOW OBVIOUSLY THE GENES IDENTIFIED BY THE ASC AND OTHER GROUPS ARE CRITICAL TO GENERATING HYPOTHESIS ABOUT SPATIAL TEMPORAL AND CELLULAR LEVEL CONVERGENCE AND I THINK THE NEXT COUPLE YEARS WILL BE VERY EXCITING IN THAT REGARD. NOW ALONGSIDE THAT WE NEED TO INCREASE THE SPATIAL TEMPORAL DATA REZONINGUTION, IN ORDER TO BE ABLE TO GET DOWN TO VERY SPECIFIC HYPOTHESIS AND NICELY THAT WORK IS ALSO ONGOING AND IN THE GROUPS LIKE ENCODE, SEVERAL OTHER DIFFERENT CONSORTIUM. AND FINALLY, I THINK THE KEY POINT ESPECIALLY WHENY WE THINK ABOUT THIS STUDY, RIGHT NOW WE HAVE--WE ARE UNDER POWERED IN THESE ANALYSIS BUT AS THE DATA ACCUMULATES THE WHOLE DATA, WE HAVE A LARGE LIST OF NONCODING VARIANTS HAVING THESE COMPLIMENTARY REGULATORY ANNOTATIONS THAT ARE SPATIALLY WILL BE CRITICAL TO DISENTANGLING THOSE DISCIPLINARY DIMENSIONS. SO THAT I WANT TO FINISH WITH THE ACKNOWLEDGMENTS. OBVIOUSLY THE NIMH HAVE BEEN GENEROUS IN SUPPORTING THIS WORK AND ALSO INVITING ME HERE TODAY, I'M ALSO VERY HAPPY TO BE PART OF THE ASC AND ALSO WITH THE AUTISM SYSTEMS BIOLOGY GROUP WITH MATT, STEPHAN, AND BERNE EXPE KATHRYN, AS WELL AS ALL OFURE OTHER FUNDERS AND THE MANY PEOPLE I COULD NOT MENTION ON THIS SLIDE. SO THANK YOU VERY MUCH. [ APPLAUSE ] >> ALL RIGHT, SO NOW WE HAVE ANY QUESTIONS FOR THE LAST GROUP OF SPEAKERS BUT IT'S FAIR GAME SINCE MARK AND BEN ARE RIGHT IN THE SECOND LOW, IF YOU HAVE QUESTIONS WE CAN BRING THEM BACK UP. I MEAN I COULD START WITH A COMMENT THAT IT WAS REALLY NICE TO HEAR IN JOE'S REMARKS BRINGING TOGETHER ALL KINDS OF VARIATION, WE HEARD A BIT PREVIOUSLY BY COMMON VARIATION BUT RARE VARIATION, CNVs, AND ALSO SOMATIC MOSAICISM AND HOW THESE COME TOGETHER WHICH ILLUSTRATES THE IMPORTANT POINT I TRIED TO MAKE ON IN THE BRE BEGINNING, THAT PEOPLE FOCUS ON 1 VARIATION TO THE EXCLUSION OF OTHERS BUT WE WILL GET A WHOLE PICTURE IF WE BRING IT TOGETHER. BUT ANY QUESTIONS? YEAH? >> SO, WE ALL KNOW AUTISM HAVE GOOD FINDING OF CNV AND DE NOVO MUTATION SO QUITE CURIOUS WHY WE DID NOT HEAR MUCH ABOUT GWAS SO FAR? SO NOT MUCH GOOD FINDING? >> [INDISCERNIBLE]. >> NICK WAS NOT AGGRESSIVE ENOUGH. SO THERE'S GWAS DATA EMERGING NOW. THERE IS A PAPER IN MOLECULAR AUTISM THAT HAS LIKE A FINDING AND THE ISAAC DATA WILL BUMP THAT UP. IT'S A SAMPLE SIZE ISSUE, THESE ARE MODEST COMPARED TO OTHER SAMPLES. WE DO KNOW FROM 1 STUDY FROM BERNIE'S LAB, A COLLABORATION WITH BERNIE THAT MORE THAN HALF OF THE TOTAL GENETIC RISK FOR AUTISM IS POLYGENIC FOR COMMON VARIATION, WE HAVE NOT YET HAD THE SAMPLE SIZE TO MAKE SIGNIFICANT FINDINGS THERE. BUT BECAUSE AS I SAID THE DE NOVO MUTATIONS ARE SO RARE, EVEN MODEST SAMPLES YOU CAN MAKE FINDINGS THAT ARE STATISTICALLY SIGNIFICANT. >> ANY OTHER--YOU WANT TO ASK ANOTHER QUESTION, IT'S GOOD--THE GROUP. IT'S ONLY 11:30 BUT I GUESS CEREBRAL GLUCOSE LEVELS ARE ALREADY CHALLENGED. >> I WILL CONTINUE THE QUESTION, SO HOW CAN WE COMPARE THE FINDING OF YOUR AUTISM STUDY ON SCHIZOPHRENIA BECAUSE THE LAST STUDY, YOU KNOW STARTS WITH GWAS, BASICALLY YOUR CONCERNS START WITH MORE SEQUENCING, SO I DON'T DON'T KNOW WHETHER YOU WOULD CONSIDER VERY SIMILAR SPECTRUM OF VERYIATION ASSOCIATION OR CONNECTION OR JUST IN DIFFERENT END OF THE SPECTRUM. CAN YOU COMMENT ON THAT? >> YOU CAN FIGHT IT OUT. >> I'LL GET IT STARTED AND THEN-- >> YEAH, SO THEY START AT DIFFERENT PLACES. THE GWAS FOR AUTISM ARE COMING UP, IN TERMS OF ARCHITECTURE WE KNOW DE NOVO MUTATION AND SCHIZOPHRENIA IS A REAL THING BUT PROBABLY NOT AS--WELL WE KNOW IT'S NOT AS KIND OF OBVIOUS AS THEY WILL NOW EXPLAIN AS WHAT WE FIND IN AUTISM. >> SURE. SO I THINK IT'S WORTH BEARING MIND THAT WITH THE AUTISM, THE FIRST COUPLE HUNDRED OR SO TRIOS STARTED TO DELIVER A HANDFUL OF GENES FOR WHICH STRONG ACTING MUTATIONS REALLY CONFERRED DRAMATIC RISK. SO, THE NATURE OF THAT RISK HAS BEEN FURGT INTERPRETED AND SCRUTINIZED BY LOVELY WORK BY ELISE, AND MARK AND JACK AND THOSE MUTATIONS ARE DRIVING NOT ONLY A PRESENTATION OF AUTISM BUT ALMOST ALL CIRCUMSTANCES A PRESENTATION OF INTELLECTUAL DISABILITY AND SEVERE COGNITIVE IMPAIRMENT. NOW WE LOOK AT THE SAME STUDY DESIGN, EXACTLY THE SAME APPROACH, SAME KIND OF CONSIDERATION FOR DE NOVO MUTATION, IDENTIFICATION IN SCHIZOPHRENIA, LOOKED AT 2000 TRIOS NOW AND WE DON'T SEE THE SAME INDIVIDUAL GENES EMERGING IN EXACTLY THE SAME SORT OF FASHION AS WAS THE CASE WITH AUTISM SPECTRUM DISORDERS, AND NOW, IN MY INTERPRETATION, IT'S BECAUSE IF YOU HAVE SEVERE INTELLECTUAL DISABILITY, YOU ARE VERY, VERY UNLIKELY TO GET A DIAGNOSIS OF SCHIZOPHRENIA. BUT YOU ARE STILL GOING TO GET A DIAGNOSIS OF AUTISM BECAUSE OF THE TOLERABLE CO MORBIDITY THAT THE DSM ALLOWS IN THAT CIRCUMSTANCE, FOR SCHIZOPHRENIA AND ALSO PRESENTATION ISSUES AND SO FOR THAT KIND OF MORE SYNDROMIC, MORE SEVERE MUTATION END OF THE SPECTRUM, YOU GET A BIT MORE RUNWAY IN AUTISM THAN YOU DO IN SCHIZOPHRENIA AND I THINK WHERE THAT DISTINCTION IS COMING FROM. IT IS WORTH HIGHLIGHTING THAT THE SAME MORE SUBTLE ODDS RATIOS OF 1.1, 1.2 RARE VARIANT MUTATIONS WILL BE PRESENT FOR BOTH AUTISM AND SCHIZOPHRENIA, BUT THEY WILL HIRE CONSIDERABLY MORE SUBSTANTIAL CIRCUMSTANCES THAN PICKING OFF THESE LATENT SYNDROMES THAT WE THINK OF AS PART OF THAT MORE ON THOSE CLINICAL GENETIC SPACE THAT IS IN THE AUTISM UMBRELLA AS IT STANDS. WHICH IS WHAT JOE MEANT TO SAY. >> YOU CAN SEE TEAM SCIENCE AT WORK HERE. YEAH WITH A LITTLE MENACE. YEAH, GO AHEAD. >> BUT I DO THINK THAT, YOU KNOW THE REAL CLINICAL FINDINGS ARE VERY, VERY FEW AND FAR BETWEEN. AFTER THE FIFTH GENE ON THAT LIST, THE ODDS ARE MUCH THAT IT'S AN ADDITIVE MODEL QUICKLY AND I THINK WHAT'S INTERESTING--I THINK THERE'S A LOT OF SPACE TO EXPLORE, AND ALL THESE DISORDERS AND SHARED RISK. WITHIN FOR EXAMPLE THE IDSD CONTINUUM, I THINK WE WILL FIND GENES THAT ARE MORE ASD AND IAD AND IT'S TRUE THAT THE 1S WE ARE FINDING FIRST ARE THE 1S THAT ARE REALLY DELETERIOUS AND PRESENT WITH A RISK FOR BOTH IDS, AND ISD. >> JEREMY NICE TALK, I HAVE A QUESTION ON YOUR SINGLE CELL SEQUENCING DATA SET. SO HAVE YOU ACTUALLY DONE THE SAME THING NOT ONLY FOR HIGH CONFIDENCE ASD GENES BUT ALSO PROBABLE ASD GENE LISTS BECAUSE HIGH CONFIDENCE GENE LIST ONLY HAVE 13 YEENS RIGHT AND I THINK IT'S KIND OF PRELIMINARY TO CONCLUDE SOMETHING OUT OF THAT? >> SURE. THANK YOU FOR THE COMMENT. SO ACTUALLY IN THE REPLICATION ANALYSIS, WE WERE NOT LOOKING AT SINGLE CELL DATA IT WAS 13 JEERNS BECAUSE WE EXCLUDED ALL THE ORIGINAL FINDINGS SO ACTUALLY WE HAVE 65 GENES AND IN THE SINGLE CELL DATA 29 OF THEM ARE REPRESENT TD BUT OF COURSE THIS IS TEFIT NITRIC OXIDELY PRELIMINARY--DEFINITELY PRELIMINARY DATA. >> THE SECOND QUESTION WAS EVEN THOUGH THERE WAS A HIGH ENRICHMENT IN IPC LIKE INTERMEDIATE PROGENITOR CELLS BUT I COULD AWLSZ SEE SOME OF THE--ALSO SEE SOME OF THE ENRICHMENT VENTRICULAR RAID YOLKLIA, BECAUSE I THINK IT'S INTERESTING BECAUSE RADIAL GLEA HAS AN OVERGROWTH SYMPTOM IN THE BEGINNING SO HAVE YOU ACTUALLY ALSO CHECKED THAT PART TOO? >> I I'M NOT SURE IF BERNE EXPE KATHRYN HAVE LOOKED INTO THAT SPECIFICALLY BUT WE'RE DEFINITELY STILL DIVING INTO THE DATA SEE WE WILL KEEP AN EYE ON THAT. THANK YOU. IT WAS A GREAT IDEA. >> QUESTION? >> YEAH, I THINK THIS REALLY WAS A MARVELOUS SET OF TALKS THIS MORNING AND I CAN SAY HAVING GROWN UP IN THE FIELD DURING THE TIME OF LINKAGE STUDIES WHEN WE WERE STUMBLING IN THE DARK, THE PROGRESS THAT'S GONE FORWARD IN THE LAST 10 YEARS OR SO, SINCE WE MOVED INTO THE GWAS ERA HAS REALLY BEEN BREATH TAKING. BUT LET ME POSE A SOMEWHAT MORE DIFFICULT QUESTION AND IT WAS 1 THAT WAS RAISED A COUPLE WEEKS AGO IN THAT PAPER BY JONATHAN PRITCHARD IN CELL, I DON'T KNOW IF HE'S HERE TODAY BUT PERHAPS IT MIGHT BE A LITTLE UNFAIR TO HIM BUT I WANT TO ASK, WE HAVE THE LEADERS IN THE GWAS PHILOSOPHY HERE AND I--I'M A GREAT FAN OF GWAS TOO, BUT I'VE BEEN TROUBLED BY JONATHAN'S BASIC ANALYSIS AND IF I GIVE IT A FAIR SUMMARY, IT WOULD BE THAT GWAS--TROUBLED BY THE ERRORS OR TROUBLED BY BELIEVING IT? >> TROUBLED BY THE CAUTIONS THAT IT RAISES, I THINK. I'D LOVE TO HEAR ABOUT ERRORS, I KNOW JONATHAN'S A VERY CAREFUL THINKER AND I THINK IT'S BASIC THESIS IS THAT AS SAMPLE SIZES GROW, GWAS WILL IMPLICATE MORE AND MORE OF THE GENOME. SO IN THE END JUST ANY ANY GENE THAT'S EXPRESSED IN THE TISSUE OF THE DISORDER OF INTEREST IS A RISK GENE AND THAT MAY POSE A SERIOUS PROBLEMS IF WE'RE GOING TO MAKE THIS IMPORTANT NEXT LEAP FROM HAVING GENETIC MARKER ASSOCIATIONS TO HAVING GENES AND BIOLOGICAL HYPOTHESIS. SO I WANT TO THROW IT OUT TO YOU FOLKS AND ASK FOR YOUR REACTION. >> SURE. I JUST WANT TO GIVE AN EXAMPLE, I THINK OF WHY I'M NOT PERTURBED BY THE WHATEVER YOU CALL IT POLYGENIC OR I DON'T THINK THERE'S A DISTINCTION AS WITH EVERYTHING, EVERYTHING IS A CONTINUUM AND WE TRY TO DICHOTOMIZE THINGS, WE MAKE MISTAKES. BMI IS A CAUSAL, PROVEN BACKWARDS AND FORWARDS IN THE SAME WAY THAT LDL IS CAUSAL FOR HEART DISEASE, BMI, A LOT OF BAD CARDIOVASCULAR OUTCOMES, BMI IS POLYGENIC, SO THERE'S A CONSEQUENCE, EVERY SINGLE OF THE PROBE DEVELOPMENTABLY THOUSANDS OF VARIANTS THAT INFLUENCE BMI AND OTHER RELATED TRADES ARE THEMSELVES DIABETES, AND HEART DISEASE RISK FACTORS. THAT'S ALL POLYGENERATEDISSITY IS, BUT IF YOU THINK ABOUT IT, IN TERMS OF BMI, MIGHT BE EXPLAINING 10 OR 20% OR MAYBE 5% OF THE GENETICS OF DIABETES, OR HEART DISEASE, THAT 5% IS ENORMOUSLY POLYGENIC AND SO THOSE GMI EFFECTS ARE INFINIT ISES MALLY PROPOSED ON DIABETES AND HEART DISEASE. IT DOESN'T AT ALL SAY THAT THAT SORT OF COMPENDIUM OF GENETICS WHICH WE CAN ACCESS THROUGH BPI AND UNDERSTAND INTERVENEON STUDY, DOESN'T TELL ME THAT'S IN THE INFORMATIVE WHATSOEVER. IT ALSO TELLS, ME ANYWAY, THAT SOME OF THE LARGEST OF THOSE EFFECTS AND REALLY THE SECTIONAL ANALYSIS LARGEST DIABETES GENE IS ACTUALLY THE REGULATORY VALID AND VARIANCE AT THE FDO LOCUST WHICH DON'T ACT DIRECTLY ON THE CELLS OF DIABETES BUT HAVE AN EFFECT ON BMI. THE SAME WITH THE SMOKING BEHAVIOR AT THE CHRNA 3 AND 5, AND LUNG CANCER. THOSE ARE NOT ACTING DIRECTORYOT CELLS OF LUNG CANCER BUT THEY'RE INFLUENCING BEHAVIOR IN A WAY SO THAT'S ALL THAT POLYGENERATEDISSITY IS AND IF WE NEED TO HAVE FANCY NAMES AND MODELS FOR IT GREAT. I DON'T FIND THAT THE INDIRECTNESS OF POLYGENERATEDISSITY IS TROUBLING OR EVEN BIOLOGICALLY UNINFORMATIVE. IT'S VERY INFORMATIVE AND AS FOR YOU KNOW THE REST OF MODELING, THERE'S A BUNCH OF MODELS THAT CAN DO LOTS OF THINGS BUT I DON'T THINK IT'S--I DON'T THINK IT'S A CALL FOR HOPELESSNESS AT ALL. >> I THINK THERE'S 1 VERY IMPORTANT IMPLICIT POINT IN THAT WHICH IS THERE'S A LOT OF DISCUSSION ABOUT CELL TYPE ENRICHMENTS IN THAT PAPER AND A LOT OF BIOLOGICAL ENRICHMENTS AND ETL ENRICHMENTS AND THOSE KINDS OF ENRICHMENTS AND THIS IS SIMILAR TO THE SIM THING AS THE FDR CLAIM THAT AN ENRICHMENT DOES NOT MEAN EVERYTHING THAT IS IN THAT CLASS IS REAL. IT ALSO MEANS THAT ASSOCIATIONS FOR EXAMPLE THAT ARE IN EQTL REGIONS ARE NOT WORK THROUGH AN EQTL SOMETIMES THAT IS A VERY LARGE SPACE OF POSSIBILITY AND WHEN WE THINK ABOUT DECONSTRUCTING WHAT THE GENETIC INFLUENCES ARE REALLY DOING, SOME OF THEM WILL ACT MORE PROXIMALLY TO SPECIFIC, YOU KNOW KIND OF PATHOGENIC PROCESSES THAT WE SHOULD BE, YOU KNOW EXCITED ABOUT, AND WE SHOULD DEFINITELY LOOK FOR THOSE THINGS BUT SOME OF THEM WILL BE VERY DISTAL AND EXACTLY THE MANNER THAT MARK DESCRIBES AND I THINK TO KIND OF ASSUME THAT MAGICALLY WE WILL BE ABLE TO DELINEATE BETWEEN THE DISTAL AND PROXIMAL BIOLOGICAL CHAIN OF CAUSALITY IS MAYBE HIEWBEROUS MIGHT BE HOW I WOULD CALL IT. >> I MEAN WE HAVE SO MANY COUNTER EXAMPLES, THE STRONGEST INHERITED LUNG CANCER PREDISPOSITION LOCUST TO CONTINUE IN THE BRAIN. >> RIGHT. >> BUT YOU KNOW IT DOES PROMPT US TO TAKE A ELECTRIC AT OUR DATA, REEVALUATE IT IN AND IN LIGHT OF THAT DISCUSSION STARTED A FEW WEEKS EARLIER EVALUATING WHETHER OR NOT--YOU KNOW THIS QUESTION OF DO THE--EVEN THE STRONG ACTING MUTATIONS IN INTELLECTUAL DISABILITY AND SYNDROMIC FORMINGS OF AUTISM DO THOSE HAVE REAL SPECIFICITY. THE VAST MAJORITY OF THOSE ARE BROADLY EXPRESSED. THEY ARE OF COURSE EXPRESSED IN THE BRAIN BUT THE MAJORITY ARE ALSO EXPRESSED IN TISSUES SO WE TEND TO WITH A FOCUS ON ENRICHMENTS TO SOME EXTENT ARE MAYBE OVEREMPHASIZING JUST HOW SPECIFIC THE INSIDES OF THAT ARE GOING TO COME FROM THIS. >> MY WORRY IS WILL GWAS TURN UP A LARGE NUMBER OF CORRELATED NONCAUSAL ASSOCIATIONS? AND WILL WE BE ABLE TO SORT OUT THE MORE DIRECT AND INDIRECT? AND I THINK WE PROBABLY WILL BECAUSE WE HAVE A LOT OF SMART PEOPLE THINKING ABOUT IT, BUT IT IS A CONCERN THAT IT MAY BE HARD TO DO THAT PARTICULARLY IN THE BRAIN WHERE WE KNOW SO LITTLE ABOUT THE UNDERLYING CELLULAR PATHWAYS. I THINK THAT WAS THE ARGUMENT IN THE PAPER AS WELL. IF WE HAD AN IDEA OF THE WAY IN WHICH THE GENES ACT TOGETHER, WE MIGHT HAVE A BETTER WAY TO UNDERSTAND WHAT THE GWAS IS TELLING US AND THE ANALOGY THAT OTHER PEOPLE HAVE USED THIS AS WELL, THAT IF YOU WERE ANALYZING PLANE CRASHES WHICH ARE RARE EVENTS AND STUDYING THE WRECKAGE ON THE GROUND, YOU WOULD FIND A NUMBER OF ASSOCIATIONS BETWEEN PLANES THAT CRASH AND PLANES THAT DON'T CRASH. BUT IF YOU DIDN'T HAVE ANY UNDERSTANDING OF THE PHYSICS OF FLIGHT, YOU WOULD HAVE A HARD TIME-- >> YOU KNOW YOU CAN GET LOST IN THESE METAPHORS, THE PIECES OF A WRECKED PLANE ARE EFFECTS OF THE CRASH, WHEN WE ARE LOOKING AT GENE SEQUENCES THAT ARE THERE, THEY ARE AT LEAST PART OF THE CAUSAL OF CHAINS SO WE--YOU KNOW. >> AT LEAST CORRELATED WITH IT. >> NO, NO. >> THEY HAVE TO HAVE SOME CAUSAL ROLE. >> THEY HAVE TO HAVE A CAUSAL ROLE, THAT ROLE MAY BE MANIFEST THROUGH A SPECIFIC VERY PROXIMAL BIOLOGICAL INTERVENTION OR IT CAN BE A CONTRIBUTING FACTOSH TO SOMETHING THAT IS ALSO CAUSEALLY ASSOCIATED TO PHENOTYPE IN THE POPULATION. SO HERE ARE THE EXAMPLE OF HEART ATTACK IS VERY INSTRUCTIVE. ABOUT A THIRD OF THE GENOME WIDE SIGNIFICANT LOCI FOR CORINARY ARTERY DISEASE FALL INTO LIPID LOCI AND WE'RE REASONABLY CERTAIN THAT THE VARMIS MAJORITY OF THOSE ASSOCIATIONS THAT FALL INTO THOSE LOCI WILL ACT ON LIP O GENIC PROCESSES THAT WILL DRIVE ATHEROSCLEROSIS, THAT MAKES A LOT OF SENSE IN AN AREA WITH WE UNDERSTAND A GREAT DEAL OF BOTH BIOLOGY, EPIDEMIOLOGY, YOU MIEWOF THE--MOVE TO REDUCE LDL IN THE POPULATION YOU SEE THE CONCOMITANT REDUCTION IN CORONARY ARTERY DISEASE. THAT'S THE SAME SORT OF THING OF BMI AS MARK WAS ALLUDING TO IN THE INTERVENTIONAL STUDIES THAT ARE CLEEMLY CLEAR WITH REDUCING PUDDERREN OF DISEASE AND IF THERE WERE ENVIRONMENTAL ENVIRONMENTAL COMPONENTS THAT WERE IMONCHED BY GENETICS THAT INFLUENCE BEHAVIOR MUCH THE SAME WAY AS SMOKING INFLUENCES LUNG CANCER, IF WE CAN REDUCE THOSE FOR MENTAL ILLNESS, THAT WOULD BE A SUCCESSFUL THERAPEUTIC STRATEGY AS QUITTING SMOKING COMPAINS HAVE REDUCED LUNG CANCER IN THE POPULATION. DOESN'T MEAN VERMEN INFECTED TO HAVE A SPECIFIC DRUG, IT DOES MEAN IT WILL TEACH US POTENTIALLY SOME OF THE THINGS TO TRY AND INTERVENE ON AND I THINK IT'S THOSE KINDINGS OF CONCEPTUAL FRAMES THAT WE NEED TO KEEP IN MIND WHEN WE INTERPRET THE ASSOCIATIONS THAT ARE EMERGING FROM THESE SORPTS OF THINGS BECAUSE THEY HAVE TO BE CAUSEALLY ASSOCIATED 1 WAY OR ANOTHER TO DISEASE IN THE POPULATION. >> WELL I THINK--THE QUESTION IS? >> I THINK WE SHOULD CONTINUE--I WANT TO GIVE MATT'S A CHANCE TO LAY ALL THIS TO REST AND THEN IT'S LUNCH TIME. , DON'T LOOK AT ME. >> I'LL STAND DOWN, THANK YOU. >> I ACTUALLY WANT TO JUMP BACK AND I WANT TO MAKE SORLT OF A COMMENT. SO WITH REGARD TO A QUESTION ABOUT THE DIFFERENCES IN THE UNDERLYING ALLELIC ARCHITECTURE OF SCHIZOPHRENIA AND AUTISM, I THINK IT'S FACILE TO TRY TO EXPLAIN AWAY THE DIFFERENCES TAKEN--THEY WE'RE SEEING IN TERMS OF PROCESS OF GENE DISCOVERY IN THOSE DISORDERS BY SORT OF CHARACTERIZING AUTISM AS PREDOMINANTLY IDENTIFYING SYNDROMIC AND ID GENES AND SCHIZOPHRENIA AND THEY'RE NOT AND LOTS OF REASONS AS JOE POINTED OUT THE INITIAL DISCOVERIES THERE'S CLEARLY AN OVERLAP BETWEEN ID AND ASD, ABOUT YOU WHEN YOU LOOK THROUGH THE GENE LIST AND LOOK THROUGH DISTRIBUTION OF IQs AND THE SAMPLES AND THE WAY THAT THOSE ARE PHENOTYPE THEY'RE NOT DIN DROAM OF GARDEN CASES, THEY'RE ID COPATHIC AUTISM CASES BY AND LARGE, A SMALL SECTION OF THOSE KIND OF IN RETROSPECT HAVE BEEN IDENTIFY WIDE SYNDROMIC FEATURES BUT MY UNDERSTAND SUGGEST THAT THERE MAY BE SOME SIGNAL CAN REGARD TO DE NOVO MUTATIONS IN RARE HIGHER EFFECT OF MUSEUM TAIGDZS IN ADH D THAT SHOWS THAT THAT'S THE CASE WITH TERRETS. AND WE HAVE A LOT TO LEARN ABOUT THE DIFFERENCES IN ARCHITECTURE BETWEEN DISORDERS THAT ARE EARLY ON AND THAT DISORDERS THAT ARE SOMEWHAT LATER ON, SO I WOULD WANT TO GET AWAY FROM THE IDEA THAT IT'S REALLY, YOU KNOW WE CAN'T DIAGNOSE SCHIZOPHRENIA AND ID AND THAT'S WHY WE'RE SEEING A DIFFERENCE IN THE WAY THAT ALLELES ARE DISTRIBUTED. >> ANYBODY WANT TO COMMENT ON MAOF MATT COMMENT? I DON'T THINK MATT TYRANNYICAL AND LAST COMMENT THEN LUNCH. >> YOU SUGGESTED COMMENTS THAT MIGHT PUT THE DEBATE TO REST, I THINK I MIGHT HAVE 1. SO, THIS IS KIND OF RIFFING ON JONATHAN PRITCHARD'S INFIN TESS MALMODEL, OR REENKARNATED AS THE OMNIO GENIC MODEL. HE'S POSITING THAT RARE VARIANTS ARE ARE MORE LIKELY TO BE CORE GENES BECAUSE THEY HAVE A STRONGER DIRECT EFFECT ON PATHOGENISSITY AND COMMON VARIANTS MIGHT BE MORE LIKELY TO BE PERIPHERAL, THAT IS THE HYPOTHESIS WHICH IS THAT WE DON'T ACTUALLY--IT'S AN IMPERICAL QUESTION THAT WE REALLY HAVEN'T ANSWERED AND 1 CAN LOOK AT RARE VARIANTS AND SEE VERY CLEARLY THAT THOSE ARE NOT NECESSARILY CORE TO AUTISM ALL THE TIME BECAUSE A LOT OF THE CASE WHO IS CARRY THAT PARTICULAR DIAGNOSIS OF AUTISM SO EVEN A RAREIARIANT THAT MIGHT BE CORE MY DISEASE MIGHT NOT BE CORE FOR AUTISM. SOPHISTICATEDY WE'RE STUCK WITH AN INTERESTING DILEMMA OF WHAT IS THE CORE OF AUTISM AND HOW DO RAREIARIANTS ACTUALLY FIT INTO CORE OR IS THERE A COROF AUTISM OR MAYBE THERE'S 17 CORES. WE DON'T KNOW. >> I THINK THIS IS SOMETHING THAT ACTUALLY PROBABLY WOULD BENEFIT FROM A MUCH LONGER DISCUSSION, BUT WE ALSO HAVE TO MAKE SURE WE'RE NOT BEING TRIPPED UP BY AGAIN THIS ARE--ADMINISTRATIVE CONTRARY DIAGNOSTIC NOISOLOGY WHERE YOU MAKE A DOUBLE DIAGNOSIS. BUT I THINK EVERYTHING WILL IMPROVE WITH A LITTLE GLUCOSE SO LUNCH IS UPSTAIRS. >> WAIT, WAIT. >> WE WAIT, WAIT BEFORE YOU GO? YOU HAVE TO NOTICE THAL MEETING ITED IS REALLY NOT CONSTRUCTED AS A WORKING MEETING. ALTHOUGH WE HAVE MEETINGS TOMORROW AND WE HAVE 1 ON FRIDAY SO I CAN'T GO BECAUSE--SCHEDULING BUT BECAUSE OF THE DENSITY OF THE MATERIALS AND THE SUCCESSES THAT YOU ALL BROUGHT FORTH, TODAY IT TO BE STRUCTURED LIKE THIS. SO THIS IS THE LONGEST BREAK YOU'LL GET, MAKE GOOD USE OF IT. LUNCH IS SERVED UPSTAIRS AND FOR THE SPEAKERS PLEASE COME UP HERE FOR PICTURES AND DR. GORDON IF YOU WANT TO JOIN US, YOU ARE MORE THAN WELCOME. SO, HELLO, EVERYONE, WELCOME TO THE AFTERNOON SESSION. I APOLOGIZE FOR THE COMPLEXITIES AT LUNCH, THE LUNCH LINE. SO WE'RE A LITTLE BEHIND. THAT MEANS NO MORE DISCUSSIONS OR BREAKS FOR DISCUSSION OR BIOBREAKING OR ANYTHING. JUST YOU KNOW BE TOUGH. SO WE ARE STARTING THE AFTERNOON SESSION NOW, WE WILL MOVE PROTEIN DISCOVERY INTO THE CELL LINES PHENOTYPIC SCALES WE HAVE THAT WE TALKED ABOUT AND TOUCHED ON IN THE MORNING, BUT BEFORE THAT, I WANT TO THANK DR. GORDON FOR BEING HERE AND THEN HIS WILLINGNESS TO INTRODUCE OR START THE MEETING. SOME OF YOU MAY HAVE WONDERED WHY I DIDN'T DO THIS IN THE MORNING, THE REASON IS HE'S IN [INDISCERNIBLE] JUST AS ME AND HE MADE ALLOWANCES FOR THE INSTRAY JECTORY STRUCTURE, POTENTIAL COMPLEXITIES OF THE INFRASTRUCTURE QUICKLY. >> SO I WAS GOING TO TALK FOR JUST A COUPLE MINNULATES, BUT SINCE YOU KNOW PEOPLE ARE FILTERING IN FOR LUNCH, I WILL TAKE AN HOUR OR SO BUT I'M NOT PARTICULARLY WORRIED BECAUSE I SEE THAT RAQUE L WAS GIVEN AN HOUR FOR THEIR TALK, THEY BECOME SO IMPORTANT FOR THE WORK WE'RE DOING IN THE GENOMICS SPHERE AND IT'S REALLY A LESSON FOR WHAT WE PROBABLY NEED TO DO ALL ACROSS OUR HUMAN BASE RESEARCH AND WE NEED TO INCREASE OUR SAMPLE SIZES SO THAT WE NO LONGER HAVE IMAGING STUDIES IN THE TEENS BUT RATHER IN THE THOUSANDS AND SIMILARLY FOR BEHAVIORIAL STUDIES AND FOR ANY OF THE DIAGNOSTIC WORK WE KNOW WE NEED TO DO. OTHER THAN THAT, I ALSO WANTED TO MAKE THE PLEA WHICH I KNOW THIS COMMUNITY ISACCATIVELY ENGAGED THIS TRYING TO FIGURE OUT IS THAT WE MOVE BEYOND THE SIMPLE DIAGNOSTIC CATEGORIES FOR--FOR DOING GENETICS AND THINK HARD ABOUT HOW WE CAN DO PHENOTYPING DEEPER PHENOTYPING AT LEAST IF NOT DEEP PHENOTYPING ON THIS LARGE SCALE AND JUST, YOU KNOW FOR 1 EXAMPLE PICKING OFF SOMETHING THAT YOU ALREADY SAW, MIC IN ADDITION TO MAKING MUCH CLEVER JOKES THAN I'M CAPABLE OF DOING PUT UP A REALLY NICE SLIDE SHOWING THAT THE POLYGENIC RISK SCORE FOR SCHIZOPHRENIA COULD POTENTIALLY BE USED TO AID IN DIAGNOSIS SHOWING DIFFERENCES IN THE RATES OF INDIVIDUALS WITH DIAGNOSIS OF BIPOLAR OR DIAGNOSIS OF SCHIZOPHRENIA DEPENDING ON THE DECILE RANK IN THE PGS OR PRS OR WHATEVER THE ACKRON IMPEDIMENTS FOR IT IS, YOU THINK, WHAT YOU KNOW WOULD BE REALLY HELPFUL FROM A CLINICAL STANDPOINT IS WHAT DIAGNOSTIC LABEL IS MORE LIKELY GIVEN YOUR RISK SCORE BUT WHAT YOUR MEDICATION RESPONSE IS GOING TO BE AND IF WE KNEW THAT THOSE WITH BIPOLAR DIAGNOSIS WERE MORE LIKELY TO RESPOND TO MOOD STABILIZERS AND THOSE WERE THE SCHIZOPHRENIA DIAGNOSIS WERE MORE LIKELY TO RESPOND TO ANTIPSYCHOTICS AND THAT WOULD THEN BECOME A CLINICAL TEST SIGNIFY KIA TRYSTS WOULD WANT TO USE BUT OF COURSE IN THESE LARGE COHORTS WE DON'T GENERALLY HAVE THAT INFORMATION EXCEPT IN LIMITED CASES AND I THINK, DEVELOPING WAYS THAT WE CAN DO AT LEAST THAT KIND OF LEVEL OF PHENOTYPING AS WELL AS MANY OTHERS THAT THIS GROUP HAS ALREADY CONTEMPLATING WOULD BE A OF TREMENDOUS BENEFIT, NOT JUST IN TERMS OF HELPING IN THE INDIVIDUAL CASE OF DECISION MAKING SUCH AS WEATHER TO GIVE AN ANTIPSYCHOTIC OR MOOD STABILIZER TO ANINE YEAR-OLD WHO PRESENTS FOR PSYCHOSIS BECAUSE MORE IMPORTANTLY FOR THE FIELD BECAUSE WHAT WE'RE LACKING ISN'T POSSIBLE TESTS TO USE IT'S USEFUL TESTS TO MAKE POSSIBLE. SO THANK YOU VERY MUCH FOR GIVING ME THE OPPORTUNITY TO ADDRESS YOU IN THIS SHORT TIME AND WE CAN MOVE ON SO WE CAN GET TO RAQUEL. THANKS. KD--SALLY APLUZ. --[ APPLAUSE ] >> OH I FORGOT 1 THING, STEVE JOBS USED TO SAY OH RIGHT, 1 MORE THING. THIS IS EVEN BETTER THAN THE IPHONE, I WANT TO THANK OUR HELPER THIS MORNING WHO'S BEEN AN IMPORTANT MEMBER OF THE TEAM RUNNING THE TEAM SCIENCE AND ALSO HAD A LOT TO DO WITH TODAY. THAT'S ALL. KD--SALLY APLUZ--[ APPLAUSE ] >> SO AS WE PUSH INTO THE REGULATORY SPACE OF THE GENOME, THE UNDISCOVERED COUNTRY--I THINK THAT'S THE WRONG SLIDE, I THINK? WE STARTED A COUPLE OF YEARS AGO, WE STARTED THIS WHOLE GENOME SEQUENCING FOR GENOME DISORDERS WHERE WE AGAIN TO EXACTLY EXPLORE THAT SPACE AND HAS BEEN AN INTERESTING ENDEAVOR AND I WANT TO INVITE DR. DAVID GOLDSTEIN WHO IS AN EMINENT MEMBER OF THIS GROUP TO SPEAK TO US DAVID, PLEASE. >> SO, JEA, THAT WAS THE RIGHT SLIDE. I DID NOT PUT NIGH--MY NAME ON IF SO I COULD BLAME OTHER PEOPLE FOR ANYTHING I GET WRONG. SO IF WE COULD GET THE SLIDES. SO THOMAS ASKED ME TO INTRODUCE WHAT WE'RE DOING IN 1 OF THE PROJECTS WITHIN THE CENTERS FOR THE GENOMICS OF MENTAL DISORDERS, THAT WAS STARTED NOW, ALMOST 5 YEARS BACK, AND WHAT WE TRIED TO DO IS PUT TOGETHER A STUDY THAT WOULD REALLY DEVELOP THE BASIC TOOL KIT NECESSARY TO IDENTIFY REGULATORY VARIANTS THAT HAVE SOME EFFECT ON GENES AND ULTIMATELY THAT INFLUENCE RISK OF DISEASE. AND THE VIEW WE TOOK IS THAT THIS IS A REALLY HARD PROBLEM AND IN ORDER TO MAKE PROGRESS ON IT, WE HAD TO BACK UP AND CONCENTRATEOT METHODOLOGIES FOR HOW YOU DO THAT AND WE PUT TOGETHER A PROJECT THAT WAS DESIGNED TO IMPROVE ON THOSE UNDERLYING METHODOLOGIES. THE BASIC IDEA WAS TO SEQUENCE A NUMBER OF INDIVIDUALS WITH 1 OF--A NUMBER OF MENTAL DISORDERS IN OUR PARTICULAR CASE WITH THE CASE ON SCHIZOPHRENIA AND AN ATTEMPT WAS MADE TO IDENTIFY THOSE MOST LIKELY TO BE UNDER STRONG GENETIC INFLUENCE WHICH IS HARD TO DO THAT BASED ON CLINICAL CRITERIA. SEQUENCE MOSTLY TRIOS AND THEN USE A VARIETY OF APPROACHES WHICH I WILL QUICKLY INTRODUCE TO IDENTIFY VARIANTS THAT MIGHT HAVE SOME INFLUENCE ON GENE EXPRESSION AND SPLICING INCLUDING SCORING THROUGHOUT THE WHOLE GENOME BUT ALSO INCLUDING THE GENERATION OF RNA SEQ DATA BOTH IN PRIMARY CELLS AND REPROGRAM CELLS AND THAT STEP WOULD ALLOW US TO ASK QUESTIONS ABOUT THE CONSISTENCY OF THE EFFECTS OF SQTLs IN PARTICULAR AND TYPES. THIS IS WHERE WE ARE AS OF RIGHT NOW IN THE SEQUENCING WE WILL GO THROUGH THIS IN DETAIL BUT WE'RE MAKING GOOD PROGRESS, MOSTLY IN TRIOS. WE WERE USING THE STAKES WHICH WE RECENTLY STARTEDUING IN THE RESEARCH EN--STRATEGIESITUTE AT COLUMBIA, AS MOST PEOPLE KNOW NOW REPRESENT A DRAMATIC INCREASE IN OUTPUT SO IT'S REALLY POSSIBLE NOW TO GENERATE A LOT MORE DATA THE MUCH SMALLER LABORATORY FOOTPRINT AND WE'VE BEEN DOING ANALYSIS LOOKING AT ERROR RATES FOR THE NOVA SEQ AND BUT IT'S CLEAR THAT HIGH STAKES IN TERMS OF ERROR RATES. AND THE BASIC AIMS HERE, ARE ARE TO USE THE SEQUENCE DATA TO IDENTIFY DE NOVO MUTATIONS--AND ULTIMATELY RISK OF DISEASE. AND SOME YEARS BACK, THAT WE REALLY NEED TO PULL TOGETHER A VARIETY OF TOOLS IN ORDER TO BE ABLE TO RECOGNIZE VARIANTS THAT HAVE IMPORTANT EFFECTS IN WHOLE GENOME SEQUENCE DATA AND SO 1 OF THE THINGS WE'VE BEEN WORKING ON IN THIS PROJECT IS A WAY TO APPLY THE INCOLRANT SCORING SYSTEMS THAT ARE APPLIED TO GENES TO NONGENETIC REGIONS SO WE DEVELOPED A METHOD WE CALL ORION, SO WE CHOSE THAT NAME JUST BECAUSE OF THE OTHER NAMINGS WE WERE USING PREVIOUSLY AND THE BASIC APPROACH S&P TO LOOK AT VARIANTS IN DEFINED WINDOWS THROUGH THE GENOME AND TO COMPARE THE SITE FREQUENCY SPECTRUM IN THOSE WINDOWS FOR THEIR VARIANTS THAT ARE ACTUALLY OBSERVED AND THAT PREDICTED UNDERNEUTRALLITY AND WE GET A DEVIATION BETWEEN THOSE 2 AS THE MEASURE OF THE STRECT OF SELECTION IN THE REGION AND THEN WHAT WE CAN DO WITH THIS APPROACH IS LOOK THROUGH WINDOWS THROUGHOUT THE GENOME AND ASK WHETHER THEY ARE OR ARE NOT INTOLERANT AND OF COURSE, DEVELOP A DISTRIBUTION OF THE INTOLERANCE PROPERTIES OF ALL OF THESE WINDOWS. --NONFUNCTIONAL FOR THE GENOME AS A WHOLE AND SO WHAT WE DO IS SIMPLY USE THE SITE FREQUENCY COMPARISON AND SHOW THAT IT ACTUALLY WORKS IN AXONS AS YOU WOULD EXPECT WHEN YOU USE A VARIETY OF WAYS TO TEST HOW WELL IT WORKS, IT DOESN'T WORK AS WELL AS YOU USE INFORMATION AND AND ILLUSTRATION WHEN YOU APPLY IT TO GENES, YOU CAN HAVE SOME PREDICTIVE POWER IN WHICH THEY DO DON'T CAUSE DISEASE. AND WHEN YOU GO AND LOOK AT A NUMBER OF GENOMICS FEAT IRBS BEYOND AXONS THEY ACTUALLY REGISTER--THE LARGER NUMBER OF CONTROLS THAT YOU USE, AND WILL BE ABLE TO LOOK AT SMALLER AND SMALLER REGIONS BUT IT'S ALREADY QUITE CLEAR THAT THIS WILL BE ABLE TO IDENTIFY REGIONS THAT ARE UNDER PEWIFYING SELECTIONS AND WHEN YOU COMBINE THAT WITH ANNOTATIONS FROM ENCODE AND SIMILAR APPROACHES IT WILL HELP IN THE PRIORITIZATION IN THE VARIANTS THAT MIGHT BE CAUSING DISEASE. THIS IS JUST 1 OF THE BASIC APPROACHES THAT WE USE WHERE WHAT WE DO TO PATHWAY GIVES RAMITIZE AND PARAMETERIZE HOW WE IDENTIFY REGIONS THAT ARE UNDER STRONGER PURIFYING SELECTIONS OF THE GENOME. WHAT WE DO IS USE THE MOST INTOLERANT AXONS TO SEE WHAT PARAMETERS RECOGNIZE INTOLERANCE EXCELLENCE AND WHEN WE DO THAT, WE CAN ABOUT 4% OF THE GENOME AS BEING ROUGHLY, AS FAR AS APPROACH GOES, AS INTOLERANT AS THESE INTOLERANT AXONS. SO BASIC IDEA THEN IS THAT WE WILL IDENTIFY DE NOVO MUTATIONS IN THESE TRIOS AND RARE VARIANTS OF INTEREST AND THEN WE'RE GOING TO INTERSECT THAT WITH THESE ORION REGIONS THAT ARE IN THESE SELECTIONS AND THESE ARE FOR SWG QUESTIONS ABOUT WHAT THEY DO. ONE OF THE KEY FEATURES OF AAPPROACH AS I INDICATED EARLIER IS THAT WE WILL ASK QUESTIONS ABOUT THE EFFECT OF THESE VARIANTS ON EXPRESSION IN PRIMARY LEUCOCYTES BUT ALSO IN REPROGRAM CELLS AND THAT WAY WE WILL BE ABLE TO DEVELOP AN ESTIMATOR OF HOW OFTEN AN EFFECT THAT REGISTERS IN A RELEVANT CELL WILL REGISTER IN A LESS RELEVANT CELL SO THIS IS MEANT TO PROVIDE GUIDANCE ABOUT HOW APPROPRIATE IT IS OR ISN'T FOR EXAMPLE, USE TRANSCRIPTOMIC DATA TO ASSESS CAUSAL VARIANTS FROM LEUCOCYTES IN A BRAIN RELATED CONDITION. THE FINAL POINT I WANT TO MAKE IS WHAT'S FOW CLEAR IN ANALYSIS OF EXOME SEQUENCE DATA IN LARGE CASE CONTROL SETTINGS THAT SOME KIND OF A COLLAPSING OR BURDEN APPROACH THAT AGGREGATES SIGNAL OVER DIFFERENT VARIANTS IS CRITICALLY IMPORTANT TO DISCOVERY OF GENES AND GENE REGIONS OF INTEREST AND THE REASON FOR THAT IS THE OBVIOUS REASON IT'S HARD TO HAVE SUFFICIENT POWER FOR INDIVIDUAL VARIANTS AND SO WE'VE BEEN THINKING A LOT ABOUT HOW WE CAN& DEVELOP VERSIONS OF COLLAPSING ANALYSIS THAT ARE APPROPRIATE TO APPLY TO WHOLE GENOME SEQUENCE DATA SO JUST A LITTLE BIT ABOUT OUR THINKING HERE JUST A QUICK REMINDER FOR FOLKS ABOUT HOW A COLLAPSING ANALYSIS WORKS FUNDAMENTALLY. WHAT YOU DO IS GO THROUGH YOUR CASE AND CONTROL DATA SET FOR EVERY GENE. DECIDE WHAT IT MEANS TO BE A QUALIFYING VARIANT USING MINOR ALLELE FREQUENCY AND OTHER CRITERIA. IF AN INDIVIDUAL CARRIES A QUALIFIED GENE, IF THEY'RE CODED AS 1 AND THEN IF NOT THEY'RE CODEDAD 0 AND THEN YOUR ANALYSIS IS ON THAT RECODING AND THE REASON WE'RE CALLING IT RECOLLAPSING IS YOU'RE CONSIDERING ALL THESE DIFFERENT RARE VARIANTS AND THAT'S THE SIGNAL, AND THE POINT TO MAKE HERE IS THAT THIS APPROACH WORKS FAR BETTER THAN I THINK MOST OF US WOULD HAVE PREIBLGHT DIDDED THAT IT WOULD WORK. REALLY JUST SIMPLY WORKS AND IN FACT IT'S HARD TO SCREW IT UP TO BE HONEST. HOWEVER YOU DEFINE THESE THRESHOLDS IT KIND OF JUST WORKS WHICH IS INTERESTING. I'M SURE PEOPLE WILL FIND A WAY TO SCREW IT UP BUT IT WILL TAKE EFFORT. ONE ILLUSTRATION OF HOW WELL IT WORKS IS THAT WHAT WE DID IS APPLY THIS APPROACH TO A DATA SET WHERE PREVIOUSLY AS PARTED OF ANOTHER CONSORTIUM WE'RE INVOLVED IN IN EPILEPSY WE HAD DONE TRIO SEQUENCING WHERE YOU SIMPLY COUNT UP THE NUMBER OF MUTATIONS OBSERVED AND SEE IF THERE ARE TOO MANY TO EXPLAIN BY CHANCE ACCOUNTING FOR GENES IN THE GENOME WE HAD DONE THAT WITH THE NUMBER OF EPILEPSY TRIOS, AND SHOWED THAT THERE IS A SIGNIFICANT EXCESS OF DE NOVO MUTATIONS AND SOME NOVEL GENES WERE DISCOVERED THIS WAY AND SO WE TOOK THOSE SAME INDIVIDUALS, WE IGNORED THE PARENTS. WE IMPLEMENTED A GUARDEN VARIETY COLLAPSING ANALYSIS, JUST COUNTING UP ALL THE RARE VARIANTS THAT LOOK FUNCTIONAL AND RECOVERED THE EXACT SAME GENES AND THIS IS REALLY QUITE STRIKING BECAUSE OF A SMALL DATA SET. 509 INDIVIDUALS OF EPILEPSY REPRESENTED REAR HERE, 900 PLUS CONTROLS AND WHAT YOU SEE IS THE QQ PLOT ON GENES AND THOSE GENES THAT COME OFF THE LINE AT THE TOP ARE ALL KNOWN EPILEPTIC ENCEPHALOPATHY GENES. SO THIS DOES WORK WHEN YOU APPLY IT TO GENES AS A WHOLE AND WHAT WE'RE DOING TO ELABORATE THIS FRAMEWORK IS FIRST DEVELOPING AN APPROACH THAT APPLIES TO REGIONS SO THE WAY THE PEOPLE ARE IMPLEMENTING THESE ON GENES AS FAR AS I'M AWARE--A REGIONAL INTOLERANCE SCORE FOR GENES AND THEN WE WILL ACTUALLY USE THE THRESHOLDS FOR THE REGIONAL INTOLERANCE SCORE IN DEFINITION OF VARIANTS ARE QUALIFYING. WE'VE DONE PRELIMINARY WORK SHOWING THERE'S A STRONG RELATIONSHIP BETWEEN THESE REGIONAL INTOLERANCE SCORES WITHIN GENES AND WHERE GENES THAT CAUSE MEN DEALIAN DISEASES ARE FOUND SUGGESTING THIS SHOULD HELP. THIS IS JUST 1 ILLUSTRATION OF THAT KIND OF REGIONALIZATION, ALL THE DISEASE CAUSING MUTATION SYSTEM PARTICULARLY GENE FALL IN 1 PARTICULAR REGION. AND THAT REGION IS INTOLERANT. SO IF WE ACTUALLY HAD A COLLAPSING FRAMEWORK THAT USED THIS INFORMATION, VARIANTS THAT OCCUR IN THE LEFT REGION OF THIS GENE WOULD BE IGNORED AND THE OTHER WOULD BE CLAPTSING IN YOUR ANALYSIS, WE HAVE OTHER ELABORATIONS THAT I WON'T GO INTO TO TAKE A GENOME WIDE APPROACH TO DEFINING HOW EXCEPTIONAL A VARIANT IS. YOU CAN USE THESE KINDS OF GENOME WIDE APPROACHES IN THE FRAMEWORK, TOO, AND THE REASON THAT I MENTION ALL OF THIS, IN THIS CONTEXT IS THAT I THINK WE NOW NEED TO START DEPLOYING THESE KINDINGSS OF APPROACHES AS TO THE DENOAM AS A WHOLE TO GERONTOLOGYSTS FINE THESE WITHIN WHICH WE WILL BE COLLAPSING BECAUSE I DON'T THINK WE WILL BE ABLE TO GET ANYWHERE IN GETTING MUCH SIGNAL IN THE GENOME AS A WHOLE UNTIL WE CAN IMPLEMENT COLLAPSING LIKE METHODS IN THE GENOME AS A WHOLE. WE NEEDED THEM FOR GENES, THERE'S NO REASON TO THINK IT WILL BE EASIER IN SO WE'RE THINKING ABOUT MOW WE CAN ACTUALLY DEPLOY THESE KINDS OF APPROACHES TO PROPERTILY DEFINE REGIONS IN THE GENOME AS A WHOLE AND THEN COLLAPSE THOSE WITHIN THOSE REGIONS AND YOU COULD INCLUDE THINGS LIKE CHARACTERIZATIONS AS TO WHETHER THEY HAVE EFFECTS OR NOT, CONNECTED TO THE OTHER PART OF THE PROJECT. I WILL SKIP THAT IN THE INTEREST OF TIME BUT I WILL SAY THAT ALL OF THIS WORK IN THE CONTEXT OF A VERY BROAD CONSORTIUM THAT WE'RE INVOLVED IN MANY RPGHTS AND YOU WILL HEAR MORE ABOUT THE SCIENCE IN THAT CONSORTIUM NEXT SO THANK YOU FOR YOUR ATTENTION. >> DAVID'S TALKING ABOUT THE GENOME SEQUENCING SPACE, IT WILL GENERATE 20,000 WHOLE GENOMES AND 20 X COVERAGE WITH A FOCUS ON AFRICAN AMERICAN GENOMES AND 5 DISORDERS AND TRUE COLLABORATION WITH OTHER LARGE WHOLE GENOME SEQUENCING PROJECT ACROSS NIH, WE'RE LOOKING NOW AT ABOUT 75 TO A HUNDRED THOUSAND CONTROLS. SO IT WOULD BE AN INTERESTING ANALYSIS. NEXT SPEAKER IS STEVEN SANDERS FROM UCSF. HE WILL TALK TO US ABOUT THE COMPLEXITIES OF ANALYSIS IN THE NONCODING GENOME AND UNDERSTANDING THE EFFECT OF REGULATORY VARIANCE IN THE GENOME. HE WILL TALK TO US ABOUT EARLIER RESULTS IN AUTISM. THANK YOU. >> GOOD AFTERNOON. THANK YOU IT'S A GREAT PLEASURE FOR YOU TO TALK ABOUT THE WHOLE GENOME SEQUENCING THING AND I WILL FIRST LOOK INTO THIS AS PART OF THE AUTISM CONSORTIUM AND WHOLE GENOME SEQUENCING DISORDERS. I'D LIKE TO START OFF BY MAKING THE POINT THAT IT IS INEVITABLE, THE WHOLE GENOME SEQUENCING WILL EVENTUALLY REPLACE AXON AND MICROARRAY. THE SIMPLE REASON FOR THIS IS THAT EVEN THOUGH WE CAN'T FIND ANYTHING IN THE SPACE ULTIMATELY IT WILL BE CHEAPER TO DO THE GENOME SEQUENCING COST FOR, THE BEST ESTIMATE IS THAT THAT DAY WILL BE 20,022. IT MAY HAPPEN SOONER IF WE CAN SHOW MORE UTILITY IN THE WHOLE GENOME. AND SO THE MOMENT I WILL QUESTION AND THINK ABOUT WHOLE GENOME SEQUENCING IS WHEN IS IT USEFUL ENOUGH TO REPLACE THE OTHER SECNOLOGYS AND THE AREA ITS CAN BE MORE USEFUL ARE THE REGIONS WHERE WE'VE NEVER REALLY LOOKED BEFORE. IN PARTICULAR RARE NONCODING VAEVERIANTS AND ALSO SMALL CMVs AND OTHER STRUCTURAL VARIATION. SEE SAW EARLIER FROM JEREMY HOW THINKING ABOUT SPACE, CELL TYPE AND SICKIATRIC DISORDERS AND THOSE DIMENSIONS ARE GOVERNED BY NONCODING VARIANTS THEY SET THE COMPUTER TO DETERMINE WHEN THOSE EXPRESSIONS TAKE PLACE. THIS IS AN IMAGE, SO THIS IS WORK IN 2009, SHOWING THE SPECIFICITY OF 5 DIFFERENT ENHANCERS IN THE BRAIN. IF WE COULD FIND RARE NONCODING LOCI, ASSOCIATED WITH PSYCHIATRIC DISEASES IN A RIGOROUS AND CAUSAL FASHION, WE WOULD HAVE A MUCH HIGHER RESOLUTION UNDERSTANDING OF THE CELL TYPES AND THE WHEN AND THE WHERE OF NEUROPSYCHIATRIC DISEASE. IT MAY BE CRITICAL FOR MAKING PROGRESS. THE WORK I'D LIKE TO SHOW HERE IS DONE AS PART OF THE AUTISM SEQUENCING CONSORTIUM. YOU CAN SEE HERE THE GROUP OF PIs HAVE BEEN INVOLVED MPLET I PARTICULARLY WOULD LIKE TO HIGHLIGHT MIKE AND BERNIE WHO ALONG WITH ME HAVE BEEN LEADING THIS GROUP. NOW AS WE HEAD INTO THE CODING REGIONS IT GETS HARDER TO THINK ABOUT WHAT THE HYPOTHESIS SHOULD BE. WE'VE SEEN EARLIER AND AND--MISSENSED VARIANTS ARE LIKELY TO MEDIATE RISK. AS WE MOVE INTO THE NONCODING AREA, WE DON'T HAVE THIS LUXURY, DO WE KNOW IF IT'S PROMOTERS OR UTRs OR INTRONS THAT CARRY THE KEY. SIMILARLY WE LOOK AT ACTIVE MARKERS OR HISTONE MARKERS OR BINDING SITES OR LOOK AT OPEN DNA, SHOULD WE EXPECT IT TO BE THE REGIONS WHICH ARE THE BEST CONCERNED SORE COULD WE SAY THAT NEUROPSYCHIATRIC DISORDERS ARE SIMILAR TOO HUMANS. WE DON'T KNOW THE ANSWER AS TO WHERE IN THE REGION THIS SIGNAL WILL LIE AND AS A RESULT AS WE STARTED THINKING THROUGH THE ANALYSIS OF THIS, WE REALIZE WE NEED TO TREAT ALL OF THESE HYPOTHESIS EQUALLY. AND THE WAY WE DID THAT WAS THINKING OF MULTIPLE ANNOTATION CATEGORIES. WE ARE LOOKING HERE AT 5 HNTD 19 FAMILIES WITH WHOLE GENOME SEQUENCING AND ACROSS THOSE WE FIND ABOUT 70,000 DE NOVO MUTATIONS. WE CAN SPLIT THEM BY THEIR LOCATION OF GENES BY THE RELATIONSHIP TO KNOWN ASD GENES OR OTHER LIST OF GENES WHETHER THEY'RE CONSERVED OR NOT, AND THEN ALSO USING THIS HUGE AMOUNT OF DATA COMING FROM ENCODE, THE ROAD MATTER EPIGENOME AND FUTURES OF ENCODE. LOOKING AT ALL THESE DIFFERENT CATEGORIES, WE COMBINED ALL POSSIBLE COMBINATIONS OF THE 5 DIFFERENT COLORS. THIS RESULTED IN SOME VERY, VERY LARGE CATEGORIES, SAY FOR EXAMPLE, ALL DENO O REGIONS AND VERY, VERY SPECIFIC 1S. SAY FOR EXAMPLE, INTELS IN THE PROMOTERS OF ASD GENES AND EACH 1 OF THESE, WE ASSESS THE RELATIVE BURDEN AND 519 ASD CASES, AND 519 MATCHED SIBLING CONTROLS. AND HERE'S THE RESULTS. SO HERE ON THE BOTTOM, WE SEE THE RELATIVE RISK. IF IT GRZ TOTS RIGHT THAT'S ANNOTATION IS ENRICHED IN CASES. IF WE GO TOWARDS THE LEFT. THAT MEANS IT'S ENRICHED IN CONTROLS. THE SIGNIFICANCE OF THAT FINDING IS SHOWN ON THE Y-AXIS AS IT WOULD BE IN A GENOME WIDE ASSOCIATION STUDY WITH A LOW MINUS P-VALUE. SO IN THE IDEAL WORLD WE WOULD WANT TO SEE A CATEGORY EXCEEDING OUR LEVEL OF SIGNIFICANCE IN THE TOP RIGHT HERE. THIS IS AN EARLY FINDING JUST LIKE GWAS ANALYSIS, WHERE WE DON'T SEE A SINGLE CATEGORY WHICH IS A CATEGORY WIDE ASSOCIATION. WHAT WE CAN SEE THOUGH ARE CLUSTERS WHICH COME TOP, AND LOOKING ACROSS THE ENTIRE GENOME IN A COHORT WHERE ALL LOSS OF FUNCTION MUTATIONS HAVE BEEN REMOVED ARE TOP PAIRED WITH MISSED SENSE GENOMES IN THE ENCODING GENOME. SO WHEN YOU LOOK ACROSS GENOME IN THE RARE VARIANT SPACE, THE CODING GENOME SEEMS TO HAVE THE PREDOMINANT SIGNAL. NOW WE'VE REPEATED THIS ANALYSIS LOOKINGAL HOMOZYGOUS VARIANTS, SPLIT BETWEEN LOWS INHERITED FROM THE MOTHER AND THE FATHER. PARTICULARLY INTERESTED IN THE MONTHS OF THE MOTHER BECAUSE OF THE SEX IN AUTISM AND WE'VE DONE IT FOR STRUCTURAL VARIATION. SO FOR ALL THESE WE SEE THE SAME RESULT FOR THE NEW SINGLE CATEGORY OF ANNOTATION WHICH EXCEEDS THIS LINE HERE. SO WE DON'T SEE CLEAR EVIDENCE OF THE NONCODING SIGNAL FOR THESE POINTS. EVEN THOUGH WE DIDN'T SEE A SINGLE SIGNAL FROM A CATEGORY YOU COULD ARGUE THAT THERE ARE MORE OF THESE ANNOTATION CATEGORIES ON THE RIGHT THAN THERE ARE ON THE LEFT. WHICH WOULD MAYBE BE A HINT OF AN UNDERLYING SIGNAL AND TENDENCY TOWARDS ENRICHMENT IN AND CASES. AND TO ESSENTIALLY FORMALIZE THAT, WE TAINTED A NUMBER OF THESE CATEGORIES IN CASES COMPARE WIDE A NUMBER WHICH ARE NORMALLY SIGNIFICANT IN CONTROLS. WE THEN VIEW THAT AGAINST 10,000 PERMIRROR IMAGEIATIONS THAT TELL US WHAT TO SEE. AND YOU CAN SEE HERE THE CONTROLS NICELY LINE UP WITH THE EXPECTATION. HOWEVER IN THE CASES WE DO SEE AN EXCESS OF SIGNIFICANCE OVERALL. IN SOME WAYS THIS IS SIMILAR TO THINKING ABOUT A POLYGENIC RISK SCORE IN THE CONTEXT OF CATEGORY WIDE ASSOCIATION. SO THIS SUGGESTS THAT EVEN IF WE CAN'T SEE A SINGLE CATEGORY WITHIN THE DATA SET THERE IS STILL SIGNAL ASSOCIATE WIDE AUTISM RISK. OF COURSE, I'VE BEEN PROMULGATE PROMULGATE--PROMISING HERE NONCODING, NOT JUST NONSIGNAL. SO IF WE SPLIT HAD ON THE LEFT BETWEEN NONCODING AND ALL NONE CODING, WE SEE THE SIGNAL WAS COMING ALMOST EXCLUSIVELY FROM THE CODING REGION. WHEREAS IF WE LOOK IN THE NONCODING REGION, WHICH IS IN 99% OF THESE CATEGORIES WE SEE NO SIGNIFICANT DIFFERENCE. THEREY A STREND IN THE RIGHT DIRECTION. SO THAT'S JUST, WE CAN'T SEE A SINGLE CATEGORY, ON TOP OF THAT OVERALL WE CANNOT MAKE A STRONG STATEMENT ABOUT INDICATING RISK LOOKING ACROSS ALL THE DIFFERENT CATEGORIES. THE 1 POINT HERE OF LIGHT IS IF WE THEN REPEAT THIS LOOKING AT INTELS ONLY. SO THESE ARE SMALL INSERPGZS AND DELETIONS. BECAUSE THESE DISRUPT A LARGER PORTION OF THE GENOME, WE MIGHT SEE THEY HAVE A MORE DISRUPTIVE EFFECT AND THRA WILL HAVE A MORE FUNCTIONAL IRMPACT. IF YOU COMPLETE THE ANALYSIS, CONSIDERING OANLT THE INTELS HERE, WE DO SEE A SMALL EXCESS OF SIGNALS IN THE CASES FOR CONTROLS WITH NONCODING DE NOVO INTELS. NOW THIS--IN DELs. THIS MIGHT BE THE INDELS THAT REPRESENT A SWEET SPOT BETWEEN CATEGORIES. OVERALL IT'S A NEGATIVE FINDING AND I WAS GRATEFUL FOR THE COMMENTS EARLIER FOR THE IMPORTANCE OF NEGATIVE FINDINGS. IF ANY REVIEWERS ARE PRESENT HERE, PLEASE TAKE NOTE. [LAUGHTER] BUT THEN THE QUESTION IS WHAT HAVE WE EXCLUDED. AND THAT ANSWERS A QUESTION. WE SEE HERE THE LINE OF 80% POWER, SO THESE ARE THE DIFFERENT RELATIVE RISKS. OVER HERE WE HAVE A VERY, VERY LARGE EFFECT, OVER HERE WE HAVE A SMALL EFFECT AND THESE ARE A NUMBER OF MUTATIONS IN THE CATEGORY SO A DOT UP HERE MIGHT REPRESENT ALL DE NOVO MUTATIONS WHEREAS 1 DOWN HERE MIGHT REPRESENT MUTATIONS IN ASD GENES AND PROMOTERS. IN PURPLE IS THE LINE OF 80% POWER AFTER WE CORRECTED THE MULTIPLE COMPARISONS. WE WANT TO SEE A SINGLE CATEGORY AS AN ADULT GOING ABOVE THAT LINE AND WE DON'T. THE THING THAT STANDS OUT THE MOST ARE THESE CONSERVED MISSED SENSE MUTATIONS IN THE POST SYNAPTIC DENSITY IN MUTATIONS AND WHERE A FEW ARE EXCEEDING THAT. TO ORIENT YOU TO THIS, HERE'S WHAT WE WOULD SEE IF WE LOOK AT LOSS OF FUNCTION MUTATIONS. NOW WE REMOVED THIS FROM THIS EVEN THOUGH WE KNOW IT'S A STRONG SIGNAL IN AUTISM. SOPHISTICATEDY WE--SO WE WOULD EXPECT IF THERE WAS AN EQUIVALENT SIGNAL FOR IT TO INHABIT THIS SPACE HERE AND IT SEEMS CLEAR THERE'S NOTHING THERE. WE CAN'T BE ABSOLUTELY SURE THERE'S NOTHING THERE BECAUSE OUR 80% POWER LINE IS NOT BELOW IT BUT CERTAINLY THERE'S NOTHING IN THAT SPACE TO LOOK FOR OPTIMISTIC THERE. IN 1 TRAOF THE IF WE LOOK AT A MISSENSED A SIGNAL WE KNOW TO BE TRUE, WE SEE MANY CATEGORIES INHABITING AROUND THERE AND WE ARE NOWHERE NEAR POWERED TO DISTINGUISH A TRUE SIGNAL THERE FROM A FALSE 1. AS WE ADD MORE SAMPLES TO THIS, WE WOULD EXPECT THESE GRAY DOTS TOO REGRESS TOWARDS THIS AREA DOWN HERE UNLESS THEY WERE TRUE SIGNALING, WHERE A TRUE SIGNAL WILL BE MAINTAINED IN THIS. SO THE QUESTION THEN BECOMES HOW MANY SAMPLES DO WE NEED TO GET IN THIS SPACE BETWEEN DE NOVO LOSS OF FUNCTION AND DE NOVO MISSENSE. THE TOP LINE REPRESENT SAYS WHAT WE'VE DONE 519 FAMILIES, THIS BOTTOM BLUE LINE REPRESENTS 4000 FAMILIES. AND SO IT SEEMS LIKELY THAT WE NEED TO GET TO SOMEWHERE BETWEEN 4,010,000 FAMILIES TO BE REALLY RIGOROUSLY TESTING THIS AREA CLOSE TO DENEV O MISSENSE WHICH SEEMS LIKELY TO SEE WHERE THE NONCODING SIGNAL WILL LIE. NOW I'M ONLY SHOWING 500 PEOPLE, TALKING ABOUT 5000 SOUNDS LIKE A MASSIVE STEP UP, BUT, 3 THE EFFORTS OF TEAM SCIENCE AND PARTICULARLY THE WHOLE GENOME SEQUENCING PSYCHIATRIC DISORDERS CONSORTIUM OR WGRRKS SPD, THESE ARE ATTAINABLE DURING A SHORT PERIOD OF TIME. THIS HAS BEEN A COLLABORATION BETWEEN 4 DIFFERENT NIH FUNDED GROUPS, ALONG WITH THE STANDLY CENTER AND THE NIH TO COME TOGETHER TO TRY AND PULL ALL OF THIS DATA INTO A CONSISTENT PIPELINE. SO THE EFFORTS OF THIS GROUP AND OTHERS, WE HAVE ALSO SEEN REACHING OUT TO CCDG AND TOP MED WHICH BRINGS IN HUGE NUMBERS OF CONTROLS, WE SHOULD EXPECT TO SEE SOMETHING IN THE REGION OF 140,000 WHOLE GENOME SEQUENCING CONTROLS ALONGSIDE THE ANALYSIS AND PSYCHIATRIC DISORDERS BY THE END OF THIS YEAR. CRITICALLY, ALL OF THIS DATA HAS BEEN ANALYZED EXACTLY THE SAME WAY. SO WE'VE WE LEARNED OUR LESSONS FROM THE PGC AND THE SEQUENCING CONSORTIUM AND THE PIPELINE AND CONSISTENT AS THE DATA WAS GENERATED RATHER THAN REQUIRING CONSORTIA TO GO BACK AND MAKE IT AFTERWARDS. AND WE HOPE BY COMBINING EFFORTS OF THESE GROUPS HERE TO HIT THAT 5000 TRIOS, POTENTIALLY TO GET ALL GROUPS TO WORK TOGETHER. SO IN SUMMARY WE DO NOT SEE ANY CLEAR EVIDENCE FOR NONCODING SIGNAL IN 519 FAMILIES BUT THAT IS A SMALL NUMBER IF ARE THIS TYPH OF ANALYSIS. WE THINK WE CAN SAY WITH CONFIDENCE THAT THE DE NOVO SIGNAL WHEN WE DO FIND IT IS LIKELY TO BE WEAKER IN THE SIGNAL AND GIVEN THIS IS THE STRONGEST ANALYSIS OF PRESENT BUT IF WE HAVE IMPROVED ANNOTATION AND WITH INTEREST AND THIS WHOLE GENOME SEQUENCING REPRESENTS ENOUGH TO PULL IT ALL OFF TOGETHER. --FINDING FUNCTIONAL'LL MENTORSHIP SKILLS AND IMPROVING THE SIGNAL. AS WE EMBARK ON THIS AREA OF WHOLE GENOME SEQUENCING, WE THINK IT'S CRITICAL TO START THINKING ABOUT MULTIPLE COMPARISONS RIGHT FROM THE START. WE DON'T WANT A REPEAT OF THE CANDIDATE GENE ERA, HE WILL TAKE THE GWAS APPROACH, RIGOROUS STATISTICS AND COMPARISONS AND FROM THAT, WHEN WE DO FIND RESULTS WE WILL BE ABLE TO GIVE THEM TO POST DOCS AND GRAD STUDENTS WITH CONFIDENCE AND I WANT TO END ON 1 POSITIVE NOTE, EVEN AT THIS ANALYSIS, EVEN THOUGH WE FOUND NOTHING CLEAR IN THE NONCODING GENOME, WE WERE ABLE TO MAKE PROGRESS IN THE CODING GENOME AND THIS IS AN EXAMPLE OF A 4000 BASE PAIR BELESION IN A KNOWN ASD GENE WHICH IS REFERRED BACK TO THE FAMILY WHICH WOULD NOT HAVE BEEN DISCOVERED WITH THAT WHOLE GENOME SEQUENCING. I'D LIKE TO END BY THANKING THE POST DOCS AND GRADUATE STUDENT WHO IS WORKED ON THIS. YOU SAW THE PIs EARLIER. THEY DID THE WORK, IN PARTICULAR, DONNA AND JUNE SHOWN OVER HERE WHO ARE IN THE AUDIENCE TODAY, SO YOU CAN SAY HI, THANKS VERY MUCH. [ APPLAUSE ] PROGRESS, AREAS THAT HAVE TRADITIONALLY NOT BEING SUSHTLY SUFFICIENTLY ADDRESSED AND I WANT TO ASK PETER ZANDI TO JOIN US. HE'S HERE FROM HOPKINS AND HE'LL TALK ABOUT BIPOLAR SEQUENCING. >> SO THANKS FOR INVITING ME TO PARTICIPATE IN THIS SYMPOSIUM. IT'S BEEN INSPIRING AND INTERESTING. I AM HERE TO REPRESENT THE BIPOLAR SEQUENCING CONSORTIUM AS AN EFFORT TO BRING TOGETHER THE COMMUNITY OF RESEARCHERS WHO ARE STUDYING THE GENETICS OFBIPOLAR DISORDER THROUGH EXOME AND GENOMIC SEQUENCING STUDIES. WE DON'T HAVE A LOGO BUT AFTER TODAY, I'M THINKING WE SHOULD PROBABLY GET ONE TO KEEP UP WITH THE REST OF THE CONSORTIA OUT THERE. I WANT TO -- I HAVE ONLY A SHORT AMOUNT OF TIME SO I WANT TO BE WISE ABOUT THE USE OF THE TIME. I'M GOING TO TRY TO TALK ABOUT THREE THINGS REAL QUICKLY. FIRST I'M JUST GOING TO QUICKLY MOTIVATE WHY WE NEED A CONSORTIUM FOR BIPOLAR SEQUENCING STUDIES. THE SECOND IS TO TELL YOU A LITTLE ABOUT WHO WE ARE, AND THE THIRD IS TO TELL YOU ABOUT WHERE WE'RE HEADED. IT IS EARLY DAYS FOR US, WE RECENTLY GOT FUNDED SO WE'RE THINKING ABOUT WHERE WE'RE GOING AND I'LL TELL YOU A LITTLE ABOUT THAT. SO FIRST, I APOLOGIZE FOR THIS SLIDE, I'M SURE MOST OF THE FOLKS IN THIS AUDIENCE ARE TIRED OF SEEING THESE TWO FIGURES. WE'VE SEEN THEM SEVERAL TIMES ALREADY. THE FIRST SLIDE OVER HERE SHOWING THE ESTIMATES OF HERITABILITY FOR SOME OF OUR MAJOR PSYCHIATRIC DISORDERS FROM FAMILY AND TWIN STUDIES, THIS FIGURE SHOWING THE SNP BASED REPRESENTING THE COMMON VARIATION CAPTURED BY THESE GWAS SNP PANELS ACROSS OUR MAJOR PSYCHIATRIC DISORDERS FROM THE PGC. I SHOW THIS SLIDE TO MAKE THREE POINTS. THE FIRST IS THAT BIPOLAR DISORDER IS VERY SIMILAR TO SCHIZOPHRENIA IN TERMS OF THE GENETIC EPIDEMIOLOGY OF THE DISORDER. STEVE, IN THE INTRODUCTORY LECTURE, TALKED ALREADY ABOUT THE INTERTWINED HISTORY OF BIPOLAR DISORDER AND SCIZ SCHIZOPHRENIA, WHEN YOU LOOK AT THESE METRICS, THEY DO LOOK VERY SIMILAR. THE SECOND POINT IS, AS HAS ALREADY -- THE POINT THAT'S ALREADY BEEN MADE IS THAT THE HERITABILITY OF THESE DISORDERS ARE AMONG THE MOST -- THEY ARE AMONG THE MOST HERITABLE DISORDERS, PREVALENCE AROUND 1%, HERITABILITY ESTIMATES AROUND 70 TO 80%. AND THE THIRD POINT IS THAT THE GWAS SNP PAMS EXPLAIN AROUND 30% OF THE HERITABILITY, WHICH RAISING THE POSSIBILITY AND THE SUGGESTION THAT RARE VARIATION MAY CONTRIBUTE TO BIPOLAR DISORDER, HENCE MOTIVATING THIS IDEA OF DOING THESE SEQUENCING STUDIES. SO I MADE THE POINT THAT BIPOLAR DISORDER SEEMS VERY SIMILAR IN TERMS OF THE GENETIC EPIDEMIOLOGY TO SCHIZOPHRENIA, BUT DESPITE THIS SIMILARITY, THERE'S THIS NAGGING CONCERN IN THE COMMUNITY OF RESEARCHERS IN BIPOLAR DISORDER THAT WERE SOMEHOW LAGGING BEHIND THE RESEARCHERS IN SCHIZOPHRENIA AND UNDERSTANDING THE GENETIC -- THIS SLIDE HIGHLIGHTS SOME OF THOSE FINDINGS THAT REALLY STARTING TO ILLUMINATE THE GENETIC STRUCTURE OF SCHIZOPHRENIA. WE HAVE ON THE X AXIS THE ALLELE FREQUENCY AND ON THE Y AXIS, WE HAVE THE EFFECT SIZE OF THE SNP. WHAT YOU CAN SEE ARE SOME OF THE FINDINGS PUBLISHED FOR SCHIZOPHRENIA THAT REALLY ARE STARTING TO I LEUM LATE THE RISK ALLELES CONTRIBUTING TO SCHIZOPHRENIA. WE'RE ALL AWARE OF ON THE COMMON END SPECTRUM OF THE GWAS FINDINGS THAT HAVE ILLUMINATED SOME OF THE COMMON VARIATION THAT CONTRIBUTES TO RISK OF SCHIZOPHRENIA BUT THERE'S ALSO BEEN A NUMBER OF FINDINGS POINTING TO DE NOVO VARIATION THAT'S CONTRIBUTING TO RISK, COPY NUMBER AND THIS RARE CODING DISRUPTIVE VARIATION THAT ALSO SEEMS TO BE CONTRIBUTING. SO WE'RE REALLY STARTING TO ILLUMINATE WHAT'S GOING ON WITH SCHIZOPHRENIA AND THE FINDINGS ACROSS THESE STUDIES ARE REALLY STARTING TO CONVERGE ON CERTAIN PATHWAYS THAT ARE BEGINNING TO GET US EXCITED ABOUT UNDERSTANDING WHAT'S GOING ON WITH THE RISK FOR SCHIZOPHRENIA. THE PICTURE FOR BIPOLAR DISORDER IS A LITTLE BIT MORE, IF YOU FORGIVE THE PUN, DEPRESSING. WE CERTAINLY HAVE MADE PROGRESS IN UNDERSTANDING THROUGH THE PGC IN IDENTIFYING SOME COMMON VARIATION THAT CONTRIBUTES TO BIPOLAR DISORDER. THE FINDINGS CERTAINLY AREN'T THE -- THE FINDINGS IN REGARD TO WHETHER DE NOVO CONTRIBUTION CONTRIBUTES OR COPY NUMBER VARIATION OR THIS RARE CODING VARIATION CONTRIBUTES TO BIPOLAR DISORDER, THE FINDINGS ARE MIXED AND IT'S JUST NOT CLEAR WHAT'S GOING ON WITH PIE BIPOLAR DISORDER. SO IT'S NOT CLEAR TO WHETHER THE SAMPLE SIZES AREN'T THE SAME OR IF THERE'S SOMETHING FUNDAMENTALLY DIFFERENT ABOUT THE GENETIC ARCHITECTURE OF PIE POLAR DISORDER, SO WE'RE STARTING TO INVESTIGATE THIS SPACE THROUGH SEQUENCING STUDIES TO TRY TO TEST THAT IDEA, WHAT'S GOING ON HERE WITH BIPOLAR DISORDER. SO THERE ARE A NUMBER OF STUDIES THAT HAVE BP DOING THIS ACROSS THE COUNTRY AND AROUND THE WORLD THROUGH EXOME AND GENOME SEQUENCING STUDIES. SO THE NIH THROUGH THEIR WISDOM DECIDED TO FUND JUST RECENTLY THIS EFFORT TO BRING TOGETHER ALL THESE SAMPLES, BECAUSE AS WE'VE HEARD, SAMPLE SIZE IS KING. AND EACH OF THESE INDIVIDUAL STUDIES WEREN'T GOING TO HAVE ENOUGH POWER BY THEMSELVES TO ILLUMINATE WHAT'S GOING ON IN TERMS OF THE RARE VARIATION. SO WE RECENTLY GOT FUNDED THROUGH A GRANT TO BRING THESE STUDIES TOGETHER, THE AIMS FOR THE PROJECT ARE VERY STRAIGHTFORWARD, VERY SIMPLE. THE FIRST IS TO COMBINE ALL THIS DATA FOR JOINT ANALYSIS TO INVESTIGATE THE ASSOCIATION OF THESE RARE VARIANTS WITH BIPOLAR DISORDER. WE'RE BRINGING TOGETHER DATA ON AROUND 5,000 CASES AND 9,000 CONTROLS FROM SEVEN DIFFERENT CASE CONTROLLED STUDIES. WE HAVE SEQUENCE DATA FROM OVER 300 FAMILIES WITH DATA ON OVER A THOUSAND AFFECTED RELATIVES, AROUND 700 UNAFFECTED RELATIVES FROM 13 DIFFERENT FAMILY STUDIES, AND THE GOAL IS TO BRING THIS DATA TOGETHER TO TEST FOR ASSOCIATIONS. THE SECOND AIM IS TO INTEGRATE THOSE FINDINGS WITH RESULTS FROM OTHER GENOMIC, PROTEOMIC AND TRANSCRIPTOMIC STUDIES TO TRY TO UNDERSTAND FOR ANY IDEA FIED RARE VARIANTS, WHAT ARE THE MECHANISMS BY WHICH THEY MIGHT CONTRIBUTE TO THE DISEASE. AND THE THIRD AIM IS TO FOLLOW UP THOSE FINDINGS IN SOME ADDITIONAL SAMPLES TO TRY TO CONFIRM THEM BY USING AND LEVERAGING THE WHOLE GENOME SEQUENCING CONSORTIUM WE JUST HEARD ABOUT AND ALSO TO LEVERAE WHERE WE CAN DATA AND RESULTS FROM THE PSYCHCHIP EFFORT. THESE ARE THE 13 FAMILY STUDIES INITIALLY BROUGHT TOGETHER FOR THE BIPOLAR SEQUENCING CONSORTIUM. THERE ARE 13 STUDY, AROUND 300 FAMILY, OVER A THOUSAND AFFECTED RELATIVES. THEY ARE MOSTLY EXOME SEQUENCING STUDIES BUT THERE ARE ALSO SEVERAL GENOME SEQUENCING STUDIES. I SHOULD ALSO MENTION THE FAMILIES THAT HAVE BEEN SEQUENCING THESE STUDYIES RANGE IN CONFIGURATION OF A VARIETY OF DIFFERENT TYPES OF RELATIVES AND PARENTS AFFECTED THAT HAVE BEEN SEQUENCED ACROSS THESE STUDIES BUT THERE'S A GREAT DEAL OF HETEROGENEITY. MOST OF THEM ARE SEQUENCED ON THE ILLUNINA PLATFORM, MOST ARE CAUCASIAN SAMPLES BUT SOME COME FROM AMISH POPULATION, FOCUSED STUDIES WITH BIPOLAR DISORDER. THESE ARE THE CASE CONTROLLED STUDIES THAT WE'RE BRINGING TOGETHER, SEVEN OF THEM WITH AROUND 5,000 CASES AND 9,000 CONTROLS. AGAIN YOU CAN SEE THE HETEROGENEITY ACROSS THE STUDIES. MOST OF THEM ARE EXOME SEQUENCING. THERE ARE TWO GENOME SCWEEPSING STUDY, THE BRIDGES STUDY HERE IS A LOW PASS WHOLE GENOME SEQUENCING STUDY WITH A MEDIAN COVERAGE AROUND 9X. THE REST OF THE STUDIES, EXOME, ARE HIGHER PASS COVERAGE. AGAIN MOST OF THEM ARE ALSO CAUCASIAN. THERE ARE FOUR CHALLENGES IN TRYING TO BRING THE DATA TOGETHER. THESE ARE SAMPLES THAT HAVE ALREADY BEEN SEQUENCED. SO WE'RE WORKING WITH EXISTING DATA THAT AS I MENTIONED HAVE BEEN SEQUENCED ON A VARIETY OF DIFFERENT PLATFORMS. SO THE FIRST CHALLENGE WE ENCOUNTER IS WHAT DATA DO WE SHARE ACROSS THIS CONSORTIUM? CAN WE GET AWAY WITH SHARING SUMMARY LEVEL STATISTICS, VARIANTS VAIR VARIANT LISTS AND SUMMARY STATISTICS. DO WE DO A HARMONIZED PIPELINE TO RECALL ALL THE DATA. THE SECOND CHAL INK IS HOW DO WE HANDLE THE HETEROGENEITY ACROSS THESE STUDIES. I THINK THERE ARE THREE MAIN SOURCES, THE DIAGNOSTIC HETEROGENEITY, HOW THESE CASES WERE ASSESSED AND ASCERTAINED, THE POPULATION HETEROGENEITY FROM A VARIETY OF POPULATIONS AS I DESCRIBED IN THE PREVIOUS TWO SLIDES. AND THEN THE SEQUENCING AND TECHNICAL HETEROGENEITY. HOW DO WE ADDRESS THOSE, AND MANAGE THAT HETEROGENEITY IN PULLING TOGETHER THIS DATA. THE THIRD IS HOW SHOULD WE ANALYZE THESE FAMILY SAMPLES AND THESE CASE CONTROL SAMPLES TOGETHER. THERE'S BP BEEN GREAT SUCCESS IN USING FAMILY SAMPLES IN SEQUENCING DATA FOR RARE MEN DALIAN DISORDERS, FOR IDENTIFYING DE NOVO VARIANTS IN AUTISM BUT HOW CAN WE COMBINE THAT WITH THE CASE CONTROL DATA THAT WE'RE WORKING WITH. THEN FINALLY, DO WE REALLY HAVE A LARGE ENOUGH SAMPLE. SAMPLE SIZE IS KING, THIS IS AN UNPRECEDENTED SAMPLE SIZE, BUT WILL WE HAVE ENOUGH WITH WHAT WE'RE WORKING WITH? WE STARTED TO DO SOME PRELIMINARY ANALYSIS WITH THE FAMILY DATA, WE'VE PULLED TOGETHER NINE STUDIES, SIX OF THEM WHOLE EXOME, THREE WHOLE GENOME. YOU CAN SEE AM THE WHOLE EXOME SEQUENCING STUDIES, THERE'S A LOT OF VARIABILITY IN THE PLATFORM, IN THE CAPTURE KITS USE AND THE VERSION OF THE CAPTURE KITS USED. SO WE QUICKLY LEARNED THAT WE ARE GOING TO HAVE TO SHARE INDIVIDUAL LEVEL BAND FILES, RECALL ALL THE DATA TO GET A MORE HARMONIZED DATA, AND THAT'S WHAT'S REFLECTED IN THIS SLIDE HERE. THIS IS SHOWING THE TS/TV RATIOS METRIC OF SEQUENCING QUALITY BEFORE DOING THIS HARMONIZED PIPELINE AND AFTERWARDS. WHERE EACH COLOR CODE REPRESENTS A DIFFERENT STUDY, THESE ARE THE KNOWN VARIANTS, NOVEL VARIANTS. YOU CAN SEE THE HETEROGENEITY ACROSS STUDY, BUT RECALLING EVERYTHING DID A MUCH BETTER JOB OF HARMONIZING THE DATA AND MADE US FEEL MUCH MORE COMFORTABLE ABOUT MOVING FORWARD SO THAT MOTIVATED US THINKING WE WERE GOING TO HAVE TO GO BACK AND RECALL ALL THIS DATA TO HARMONIZE IT. WE ALSO DID PRELIMINARY ANALYSIS ON THE CASE CONTROL DATA WITH THE FIRST THREE CASE CONTROL SAMPLES WE PULLED TOGETHER. WE DID TEST THE VARIANT LEVEL AND GENE LEVEL STATISTICS TO SEE IF WE FOUND ANY -- THESE ARE QQ PLOTS, ACROSS DIFFERENT CLASSES OF ALLELES TEST BID ALLELE FREQUENCY AND FUNCTIONALITY. AS YOU CAN SEE, THE QQ PLOTS ARE RATHER DEFLATED, ESPECIALLY FOR THE RARE VARIANTS, OR NULL, AND IT JUST MAKES THE POINT THAT EVEN THOUGH WE HAVE A LARGE SAMPLE SIZE TOGETHER, IT'S PROBABLY NOT GOING TO BE LARGE ENOUGH AND WE ARE GOING TO NEED LARGER SAMPLES TO HAVE THE POWER TO IDENTIFY RARE VARIATION IN ITS CONTRIBUTION TO BIPOLAR DISORDER. SO THE LAST SLIDE HERE IS THAT FORTUNATELY THERE HAVE BEEN A NUMBER OF SEQUENCING STUDYS THAT HAVE BEEN CARRIED OUT SINCE WE STARTED PULLING THIS CONSORTIUM TOGETHER. THE BROAD IS SEQUENCING 10,000 CASES AND 10,000 CONTROLS THROUGH PRIVATE FOUNDATION MONEY, THEY'VE GONE R. DONE 5,000 CASE, 5,000 CONTROLS. THE WHOLE GENOME SEQUENCING CONSORTIUM, WHAT WE'RE DOING NOW IS WORKING TOGETHER TO BRING ALL THOSE SAMPLES TOGETHER AND DO A JOINT ANALYSIS AND ULTIMATELY WE'LL HAVE AROUND 12,000 CASES AND 16,000 CONTROLS FOR COMBINED ANALYSIS, WHICH DOES REPRESENT AN UNPRECEDENTED SAMPLE SIZE AND GIVES US HOPE WE MAY BE ABLE TO START INVESTIGATE WHAT'S GOING ON IN THAT SPAI THE RARE GENETIC VAIR YAI AND HOW THAT CONTRIBUTES TO RISK FOR BIPOLAR DISORDER. SO WITH THAT, I'D LIKE TO THANK THE BIPOLAR SEQUENCING CONSORTIUM. IT'S BEEN A PRIVILEGE TO WORK WITH SOME OF THESE -- WITH THESE COLLEAGUES WHO REALLY THE LEADERS IN THE FIELD. I'D ESPECIALLY LIKE TO THANK THOMAS AND ANGIE AND GITA FOR THEIR ENCOURAGE AND SUPPORT IN PULLING THIS CONSORTIUM TOGETHER. [APPLAUSE] >> THANK YOU, PETER. NICE EXAMPLES OF HOW INTERCONNECTED TEAM SCIENCE -- PRESENTATION BUT REALLY ALL OF THESE CONSORTIA WORK IN ONE WAY OR THE OTHER WITH ONE OR MORE OF THE OTHERS. IT'S A PLEASURE FOR ME TO INTRODUCE DR. MATT STATE, WHO'S CHAIR OF PSYCHIATRY AT UCSF, WHO WILL TELL US ABOUT THE RESULTS IN TOURETTES. >> GREAT, THANKS SO MUCH TO BE HERE. 10 MINUTES. HERE WE GO. SO WHAT I WANT TO DO IS HOPEFULLY BY THE END OF THE TALK, GIVE YOU A GOOD SENSE THAT, ONE, WE'VE HAD EARLY SUCCESS WITH SEQUENCING TREE OWES TO LOOK FOR DE NOVO MUTATIONS IN TOUR RET'S SYNDROME, THE SAME TRAJECTORY WE HAVE IN OUGHT IN AUTISM. THE REASON I PUT THIS UP IS THAT IT'S OFTEN THE MOST PARITY -- WHAT LITTLE WE KNOW ABOUT SORT OF THE UNDERLYING BIOLOGY FOCUSES ON THE CIRCUITRY FROM A VARIETY OF CONVERSION DATA, INCLUDING NEUROIMAGING, LESIONING DATA, THERE'S A CHILDHOOD ONSET, ONE OF THE MOST INTERESTING THINGS ABOUT THE DISORDER IN THINKING ABOUT THE POTENTIAL TRACTION OF UNDERSTANDING BIOLOGY THROUGH GENETICS, ABOUT TWO THIRDS OF THE PATIENTS GET SIGNIFICANTLY BETTER INTO ADULTHOOD, SO WE ALREADY KNOW THERE'S SOME BUFFERING PROCESS GOING ON, SO UNDERSTANDING IN A DEEP WAY THE MOLECULAR, CELLULAR AND CIRCUIT LEVEL UNDER PINNINGS OF THE DISORDER MIGHT IMMEDIATELY TAKE TO YOU BEGIN TO UNDERSTAND SOMETHING ABOUT HOW TO EXTEND THE BENEFITS THAT NATURE HAS ALREADY PROVIDED. SO THE HISTORY OF TOURETTE GENETICS, GENETICALLY, PHENOTYPICALLY HETEROGENEOUS STARTED WITH THE MODEL OF A MONOGENIC DISOFFER, TO LINKAGE STUDY, TO UNSUCCESSFUL CANDIDATE GENE STUDY, ON TO GWAS. GWAS SO FAR IS PART OF THE PGC AND THEY'RE GETTING CLOSE BUT STILL HAVEN'T QUITE GOTTEN OVER THE FINISH LINE WITH REGARD TO COMMON VARIANTS, ALMOST UNDOUBTEDLY A CONSEQUENCE OF SAMPLE SIZE. IN 2010, J. TISHFIELD AND I GOT TOGETHER REALLY AT THE URGING OF THOMAS LAYNARD TO START A RARE VARIANT FOCUS GROUP CALLED TIC GENETICS, AND WE WERE LUCKY ENOUGH RIGHT ABOUT THAT TIME TO GET A COLLABORATIVE R01. UNFORTUNATELY IT ONLY SUPPORTED COLLECTION OF A SMALL NUMBER OF FAMILIES. SO WHAT I'M GOING TO SHOW YOU, THE REST OF THIS WORK IS PRETTY MUCH BUBBLE GUM AND A COUPLE STARTUP PACKAGES. WE WERE ABLE TO SHOW IN 2012 THAT THERE WAS OVERLAP IN -- SO FIRST THAT THERE WAS RISK ASSOCIATED WITH RARE CNVs, THAT THERE WAS A TREND TOWARDS SIGNIFICANCE FOR DE NOVO CNVs AND THAT THERE WAS SOME OVERLAP, NOT STATISTICALLY, BUT LOOKING AT THE CMVs THAT WE WERE HITTING, THAT THERE WAS CERTAINLY OVERLAP WITH WHAT WE WERE FINDING FOR OTHER NEURODEVELOPMENTAL DUS ORDERS INCLUDING AUTISM. SO WHAT I WANT TO TAKE YOU THROUGH IS THE EXOME SEQUENCING OF TREE OWES IN SEARCH OF DE NOVO MUTATIONS AGAIN IN MUCH THE SAME WAY WE WORKED ON AUTISM SPECTRUM DISORDERS THIS IS THE STRUCTURE OF ON KOWR OUR CONSORTIUM. WE DO HAVE A LOGO UP THERE. GARY HAS BEEN ROARLY REALLY SORT OF THE DRIVING FORCE IN KEEPING US ALL TOGETHER. WE HAVE OUR CLINICAL LEADS AS WELL, AND YOU'LL SEE IT'S A WELSEYETH ALL PAPER I'M GOING TO DESCRIBE FEWER. WE HAVE BEEN ABLE TO COLLECT NEARLY 3,000 SUBJECTS WITH DNA, ABOUT 650 TRIOS, AND ALL OF THESE ARE AVAILABLE PUBLICLY IN THE NIMH CENTER FOR COLLABORATIVE GENOMIC STUDIES AT -- HOUSED AT RUTGERS. SO WE DID EXOME SEQUENCING OF TREE OA AGAIN LOOKING FOR DE NOVO LIKELY GENE DISRUPTING MUTATIONS IN TOURETTE. OUR FIRST PAPER WAS JUST PUBLISHED IN "NEURON" AND WITH A TOTAL SAMPLE OF 484, WE SEE AN OVERREPRESENTATION OF DE NOVO LIKELY GENE DISRUPTING MUTATIONS BASED ON WHOLE EXOME SEQUENCING. THE CONTROL GROUP HERE IS THE SIMON SIMPLEX -- THE SAME CONTROL WE USED FOR THE AUTISM WORK. WE WERE REALLY FORTUNATE, THERE'S ANOTHER TURRETTE'S SYNDROME CONSORTIUM FOCUSED OP GWAS BUT THEY HAD ABOUT 173 -- ABOUT. ABOUT 173 TRIOS THAT THEY ARE SEQUENCED AND THEY CONTRIBUTED TO GIVE US A REPLICATION SAMPLE, SO WE SEE IN BOTH INDEPENDENT GROUPS AND THEN IN THE COMBINED ANALYSIS, WE SEE AN OVERREPRESENTATION OF BOTH -- LIKELY GENE DISRUPTING AS WELL AS MISSENSED MUTATIONS. WE BEGAN TO SEE SOME RECURRENTLY MUTATED GENES, THE TOP ONE IN THIS INITIAL STUDY IS WWC1, ALSO KNOWN AS KBRA, WHICH JUST KIND OF SNUCK UNDER OUR THRESHOLD USING AN FDR APPROACH FOR A HIGH CONFIDENCE GENE, AND WE'RE ABLE TO PREDICT, BASED ON OUR DATA, OUR BEST ESTIMATE OF THE NUMBER OF GENES THAT CONSTITUTE THE FULL TARGET OF DE NOVO MUTATIONS, ABOUT 420, SO LOOKING VERY MUCH LIKE AUTISM IN TERMS OF THE EFFECT SIZE, IN TERMS OF THE TOTAL TARGET SPACE. WE'VE NOW BEEN ABLE SINCE THE PUBLICATION JUST A COUPLE MONTHS AGO TO ADD 300 ADDITIONAL TREE TREE OWES THANKS TO PERRY PASCHOU. WE'VE USED THEM DO A DE NOVO CNV ANALYSIS. WE HAD A TREND PREVIOUSLY BUT WE DIDN'T HAVE ENOUGH TRIOS TO LOOK FOR -- TO ESTABLISH STATISTICAL SIGNIFICANCE FOR DE NOVO CNVs, AND THEN WE DID ADDITIONAL SNV ANALYSIS. THE SHORT STORY HERE IS THAT IT LOOKS VERY MUCH LIKE AUTISM. THE FIRST BAR GRAPH IS PHASE 1, 484 TRIOS. THE SECOND BAR GRAPH IS 291 PHASE 2, THAT'S USING WHOLE EXOME SEQUENCING TO DETECT CMVs AND WE SEE ODDS RATIOS ABOUT 2.5, 2.6. THE BAR GRAPH OVER ON YOUR RIGHT& IS AUTISM, THE AUTISM-AFFECTED INDIVIDUALS AND AUTISM CONTROLS ARE THE SSE SIBLINGS THERE. SO AGAIN, THE EFFECT SIZE IS SLIGHTLY LESS THAN WHAT WE SEE BUT QUITE SIGNIFICANT, AND SLIGHTLY DIFFERENT APPROACH, OUR GENOTYPING DATA WE ONLY HAD ON A SUBSET OF THE SAMPLE. WE ALSO SEE, AS WE HAVE IN PREVIOUS PUBLICATIONS, THAT THE CMVs THAT WE SEE ARE SORT OF THE STANDARD NEURODEVELOPMENTAL, 7Q11.23, AND THIS IS A STORY THAT'S ACTUALLY BEEN BREWING FOR QUITE A LONG TIME IN TOURETTE, AS EARLY AS 2003, THERE WAS A COMPLEX REARRANGEMENT, IN 2012, WE PUBLISHED CNVs THAT OVERLAP WITH -- YOU CAN READ THE GENES, THEN VERY RECENTLY, THE TSAICG WITH A LARGE ENOUGH SAMPLE TO SHOW STATISTICAL SIGNIFICANCE FOR NRXN1 AND CNTN6. WE'RE ALSO ABLE TO ADD TO OUR GENE DISCOVERY AND I'VE HIGHLIGHTED OUR TWO TOP GENES FROM THIS ANALYSIS. IT'S A GOOD SIGN THAT P VALUES AND THE Q VALUES HAVE GOTTEN BETTER WITH THE ADDITIONAL SEQUENCING, AND THE FIRST TWO GENES IN THE LIST HAVE NOW SWITCHED ORDER WITH CELSR3 BEING OUR TOP GENE. EVEN WHEN WE DO A QUICK ANALYSIS USING TOP GENE, WE'RE GETTING CELL POLARITY AS A CHARACTERISTIC, AND I WANTED TO JUST HIGHLIGHT FOR YOU CSLR3 FOR A COUPLE REASONS. SO IT'S OUR TOP GENE NOW. I HAVE TO SAY IT FEELS LIKE WE'RE REINVENTING THE WHEEL DOING WHAT WE DID IN ASD BUT I PUT A BET ON THIS, THE VERY FIRST GENES WE FOUND IN AUTISM THAT GOT BETTER WHEN WE EXPANDED THE SAMPLE A LITTLE TURNED OUT TO BE REALLY IMPORTANT GENES, SO I'LL PUT A MARKER DOWN RIGHT NOW THAT CSLR3 I THINK IS LIKELY TO BE IMPORTANT. IT'S RIGHT WHERE WE WANT IT TO BE, IT'S A PLAINER CELL POLARITY GENE BUT INVOLVED IN -- GUIDANCE AXON GUIDANCE AS WELL, PARTICULARLY IN THE CORTICAL STRIATAL CIRCUITRY. BUT IT ALSO HAS A NUMBER OF OTHER FUNCTIONS AND IT REALLY DEPENDS ON WHEN AND WHERE YOU LOOK. SO THIS IS A RECURRENT THEME. IF YOU LOOK IN BRAINSTEM, YOU CAN SEE THAT IT'S INVOLVED IN EARLY ANTERIOR POSTERIOR GUIDANCE OF MONO AM NER JIK NEURONS INCLUDING DOPAMINE AND SEROTONIN, IT'S INVOLVED IN EARLY CELL FATE, WHAT LOOKS LIKE A FEEDBACK LOOP BETWEEN IMMATURE NEURONS INFLUENCING THE FATE OF APICAL PROGENITORS, AND THEN IF YOU WAIT AND YOU DISRUPT IT A LITTLE BIT LATER IN DEVELOPMENT, WHAT YOU GET IS A DECREASE IN THE DEVELOPMENT OF GLUE TA MATTER JIK SNAPS. SO THE REASON I WANTED TO POINT THIS OUT IS IT'S A LOVELY GENE, BUT I THINK WHERE IT TAKES SUS SORT OF BACK TO THE KINDS OF THINGS WE'RE THINKING ABOUT IN AUTISM SPECTRUM DISORDERS, THAT WITH ADDITIONAL GENE DISCOVERY AND THE ABILITY TO BE ABLE TO THINK ABOUT CONVERGENCE, THAT WE HOPEFULLY WILL BE ABLE TO BEGIN TO SAY WHEN AND WHERE SHOULD WE REALLY BE LOOKING FOR THE FUNCTION OF CSLR3. THE LAST THING I WANT TO SAY IS THAT THE PATH FORWARD IS CLEAR. IN PHASE 1, WE USED THE DATA IN ORDER TO ESTABLISH THE PRA JEK TREES OF GENE DISCOVERY AND THE TAKE HOME FROM THIS SLIDE IS THAT IN PHASE 2, WE PREDICTED WE WOULD GET 7.3 PROBABLE GEEP, WE GOT 7, WE PREDICTED 1.88 HIGH CONFIDENCE, WE GOT 2. ONCE YOU KNOW YOUR EFFECT SIZE, BEING ABLE TO DETERMINE HOW MANY YOU NEED TO SEQUENCE TO DO BETTER IS PRETTY STRAIGHTFORWARD, AND SO THIS IS A COMMERCIAL MESSAGE, IF WE CAN SEQUENCE MORE TRIO, WE WILL FIND MORE GENES. CLEAR EXCESS, CLEAR PATH FORWARD, INITIAL SCIENCE TO OVERLAP WITH NEURODEVELOPMENTAL DISORDERS AND AGAIN PUT A MARKER DOWN FOR CSLR3 AND CELL POLARITY AND WE'LL SEE IN A YEAR OR SO, I HOPE, WHETHER I'M RIGHT. NEED TO THANK EVERYONE AT TIC GENETICS. YOU CAN SEE WE'RE SPREAD ACROSS THE WORLD, GERMANY, SOUTH KOREA, THE UNITED STATES, I WANT TO THANK EVERYONE, PARTICULARLY THOMAS LEHNER, TIC GENETICS WOULDN'T HAVE COME TOGETHER WITHOUT HIM. THANK YOU. [APPLAUSE] >> THANK YOU, MATT, AND FOR PAINTING THE WAY FORWARD FOR TOURETTE GENETICS. SO REALLY ALL THE RESULTS THAT YOU HEARD ABOUT UNTIL NOW ARE MOSTLY BASED ON CAUCASIAN INDIVIDUALS, AND WE'VE MADE STRIDES TO CHANGE THAT, WE TOLD YOU A LITTLE ABOUT DOING WHOLE GENOME SEQUENCING IN AVERAGES WE ALSO HAVE A FEW ACTIVE PROJECTS NOW AND WE'LL TELL YOU ABOUT THEM. SO THE FIRST SPEAKER IS DR. DAN STEIN FROM UNIVERSITY OF CAPE TOWN, DLIETDED TO HAVE HIM. HE'S ACTUALLY THE FURTHEST TRAVELER. GOOD TO SEE YOU. >> THANKS SO MUCH, TOMAS AND STEVE AND EVERYONE, GREAT TO BE HERE. PRARLYPARTICULARLY NICE TO MEET PEOPLE I'VE ONLY MET BY EMAIL PREVIOUSLY. SO I'LL BE TALKING ABOUT A STUDY OF SCHIZOPHRENIA, I'LL TALK A LITTLE ABOUT THE BACKGROUND, SIGNIFICANCE, SOME OF THE AIMS AND METHODS, AND A LITTLE ABOUT THE ETHICAL CONSIDERATIONS IN DOING THIS KIND OF WORK. SO WE'VE TALKED ABOUT EARLIER TODAY ABOUT THE IMPORTANCE OF HAVING DNA FROM AROUND THE WORLD, AND PARTICULARLY FROM AFRICA, AND WE HAVEN'T MENTIONED THIS PARTICULAR EFFORT BY NIH AND WELLCOME TRUST CALLED THE HUMAN HEREDITY AND HEALTH IN AFRICA PROJECT, WHICH AIMS TO FACILITATE THE STUDY OF GENOMICS OF COMMON DISEASES IN THE AFRICAN CONTEXT. AS PART OF THIS INITIATIVE BY NIH AND WELL COM, THERE'S ALSO A HUGE EFFORT FOCUSED ON CAPACITY BUILDING, ESTABLISHING INFRASTRUCTURE, NETWORK OF AFRICAN INVESTIGATORS. THERE'S A WHOLE RANGE OF GROUPS INVOLVED WORKING ON THINGS LIKE EDUCATION, ANALYSIS, DATABASING, DATA SHARING, BIOBANKING AND SO ON. SO I'M NOT GOING TO BE ABLE TO DESCRIBE ALL THE MANY PROJECTS IN H3 AFRICA, BUT THIS WILL GIVE YOU A SENSE, THIS WAS THE TENTH MEETING ACTUALLY OF THE INVESTIGATORS INVOLVED. NOW WITHIN H3 AFER CAR, THERE3 AFRICA THER E WAS EARLY ON VERY LITTLE -- HEALTH WHICH IS A PITY BECAUSE WE'VE DISCUSSED ALREADY THAT MOST OF THE WORK ON DISORDERS LIKE SCIZ HAS BEEN ON NON-AFRICAN POPULATION AND THIS REALLY HAS A NUMBER OF KEY IMPLICATIONS, WE'VE TALKED ABOUT THE SCIENTIFIC IMPLICATIONS THAT THESE LINEAGES MAY BE PARTICULARLY INFORMATIVE BUT I THINK THE THING I'D LIKE TO HIGHLIGHT TODAY IS MORTGAGE ISSUE OF MORE THE ISSUE OF HEALTH DISPARITIES, YOU ALREADY KNOW THIS, BUT JUST TO REALLY GIVE IT TIME, IS TO POINT OUT THAT THE LACK OF INCLUSION OF AFRICAN -- IN OUR WORK FURTHER EXACERBATES HEALTH DISPARITIES THAT ALREADY EXIST ACROSS THE WORLD. SO BEARING THOSE SCIENTIFIC AND MORAL CONSIDERATIONS IN MIND, THOMAS VERY KINDLY BROUGHT A GROUP OF US TOGETHER AND WE PUT FORWARD THIS GRANT. IN THIS APPLICATION, MY COLLEAGUE PRESENTED THIS PRELIMINARY STRUCTURE ANALYSIS. THE GREEN ON RIGHT WOULD BE JAPANESE AND CHAI CHINESE SAMPLE, THE GRAY WOULD BE EUROPEAN, AND THEN IN THE MIDDLE, THERE IS A WEST AFRICAN SAMPLE, AND AS WE KNOW FROM UNDERSTANDING OF THE WAY THINGS WORKED, THE INITIAL SAMPLES, THE ORIGINAL EARLY PEOPLES IN AFRICA WERE THE COISAN, WHICH IS ON YOUR LEFT. AND OUR GROUPS OF INTEREST ARE THEN SOME OF THE SOUTH AFRICAN POPULATIONS, STS, ZUL FOR ZULU AND XHS FOR XHOSA. THERE IS A GOOD DEAL ON THE ORIGINAL XHOSA LINEAGE AS WELL AS OF AFRICAN LINEAGES COMING THROUGH. SO THE AIMS OF THE PROJECT THEN WERE TO TAKE 1100 CASES OF SCHIZOPHRENIA, AGE AND GENDER MATCHED CONTROLS IN THIS -- IN ONE OF THESE LOCAL POPULATIONS, WE WORKED WITH THE XHOSA POPULATION BECAUSE THEY'RE IN OUR VICINITY TO IDENTIFY EXOMIC POINT MUTATION, INDELS AND CNVs IN CASES AND CONTROLS FOR POTENTIALLY DAMAGING MUTATIONS. SO THIS IS HOW SOUTH AFRICA LOOKS. AS I MENTIONED, CAPE TOWN IS ON THE BOTTOM LEFT. THIS REALLY IS WHERE WHEN OTHER AFRICAN GROUPS MOVED INTO AFER CAR, THE ORIGINAL XHOSA WERE IN CAPE TOWN. THE OLDEST HOME SAIP YENSSAPIANS -- IN CAVES NOT TOO FAR FROM WHERE I LIVED DATED ABOUT 100,000 YEARS WITH PEOPLE HAVINGS HAVING PIGMENTS -- DOWN AFRICA INTO EASTERN AND WESTERN CAPE, SO THOSE ARE THE PLACES THAT WE RECRUITED FROM. JUST TO THANK AND ACKNOWLEDGE ALL OUR COLLABORATIVES AT THE UNIVERSITY OF WASHINGTON, AND UNIVERSITY OF PANCREATIC ADENOCARCINOMA, UNIVERSITY OF CAPE TOWN, SIR THOMAS HAS BEEN ABSOLUTELY KEY IN THE CONCEPTUALIZATION AND IMPLEMENTATION OF THIS WORK. HERE IS RUBEN VISITING AN OUTLYING AREA IN THE EASTERN CAPE, COMMUNITY HEALTH CLINIC, AND THIS IS THE SORT OF THING THAT THIS PROJECT INVOLVED WHICH WAS GOING INTO EXTREMELY RURAL AREAS, GETTING TO KNOW PEOPLE AND CONVINCING OR TRYING TO PERSUADE THEM THAT THIS WAS A USEFUL THING TO DO. AS PART OF THE EQUITY OF THIS PROJECT, WE AGREED THAT GENOMIC ANALYSIS NEEDED TO BE UNDERTAKEN NOT ONLY IN THE STATES BUT ALSO IN SOUTH AFRICA, AND SO WE DID WHOLE GENOME EXOME SEQUENCING HERE IN THE U.S., BUT WE DID CNV WORK WITH THE AFFYMETRIX PLATFORM IN SOUTH AFRICA. SO I'VE ALREADY SORT OF GIVEN AWAW MY AWAY MY POSITION ON WHERE WE STAND ON THE ETHICS, BUT STEVE, IN HIS FIRST TALK EARLY THIS MORNING, SAID THAT WE NEED TO THINK A GREAT DEAL MORE ABOUT THIS. AND THERE ARE A NUMBER OF ISSUES THAT I'LL QUICKLY TAKE YOU THROUGH. THE FIRST IS, IS CONSENT FULLY INFORMED IN THESE POPULATIONS, HOW DO YOU EXPLAIN TO SOMEONE WHO HAS LOW HEALTH LITERACY IN GENERAL, LOW LITERACY IN GENERAL AND PARTICULARLY LOW MENTAL HEALTH LITERACY, THAT YOU'RE ABOUT TO DO A CELL CULTURE? IS THAT EVEN POSSIBLE? I NEED TO TELL YOU THAT OUR IRB GAVE US AN EXTREMELY DIFFICULT TIME ABOUT THIS. I THINK IN PART BECAUSE MOST OF THE WORK THEY DO IS NON-PSYCHIATRIC AND THEY VIEWED INDIVIDUALS WITH SCHIZOPHRENIA AS HIGHLY VULNERABLE. WHAT WE'VE DONE IS WE'VE USED THE UBAC, WHICH IS A PRETTY SHORT INSTRUMENT STUDY, AND WE GIVE IT ITTIVELY ITERATIVELY AS WE EXPLAIN THE PROJECT TO PEOPLE, AND WE HAVE A PAPER THAT'S BEEN SUBMITTED NOW SHOWING THAT REALLY AFTER A COUPLE OF ITERATIONS, EVEN PEOPLE WITH FAIRLY LOW MENTAL HEALTH LITERACY, NO HEALTH LITERACY, GET IT. IT'S NATURALLY EXPLAINED TO THEM IN THEIR HOME LANGUAGE, AND IT MAY TAKE MORE THAN ONE TIME AND IT IS TRUE THAT ALL THE PATIENTS AND THOSE WITH LOW EDUCATION MAY REQUIRE A LITTLE BIT OF EXTRA TIME, BUT I THINK THE BOTTOM LINE IS WE'RE FAIRLY CONFIDENT THAT CONSENT IS INFORMED. WE WERE TOLD BY THE IRB AT THE TIME THAT THIS PROJECT STARTED THAT WE REALLY NEED TO DO COMMUNITY ENGAGEMENT. I NEVER ACTUALLY HEARD OF COMMUNITY ENGAGEMENT AND WASN'T QUITE SURE WHAT TO DO ABOUT A COMMUNITY ADVISORY BOARD, BUT AFTER A FEW YEARS NOW OF WORKING WITH COMMUNITY ADVISORY BOARDS IN THE WESTERN CAPE, IN THE EASTERN CAPE, I'M ON THE SIDE OF THOSE THAT ARGUE THAT THESE CAN BE VERY VALUABLE. SO FOR EXAMPLE, OUR COMMUNITY ADVISORY BOARD, WHEN WE EXPLAINED TO THEM THAT THE IRB WAS CONCERNED ABOUT SOME OF THEIR DECISION-MAKING AND WHAT THEY SORT OF DIDN'T UNDERSTAND, THEY HELPED SUPPORT US IN GOING BACK TO THE IRB WITH SOMEWHAT BROADER CONSENTS, ARGUING THAT THEY UNDERSTOOD THAT THEY WERE A VULNERABLE POPULATION, WERE REPRESENTING A VULNERABLE POPULATION AND REALLY PRECISELY BECAUSE OF THAT, MORE WORK NEEDED TO BE DONE ABANDONS THAT A BROADER CONSENT WAS ENTIRELY FINE FROM THEIR POINT OF VIEW. SO THE COMMUNITY ADVISORY BOARD REALLY HELPED WITH A NUMBER OF ASPECTS OF INTERACTING WITH THE IRB. THEN EARLIER ON ALSO THIS ISSUE WAS RAISED WHICH IS WHAT IS THE RELATIONSHIP BETWEEN THESE GENETICS EXPLANATIONS AND STIGMATIZATION. I THINK THERE IS WORK FROM THE U.S. WHICH SUGGESTS THAT GENETIC CAUSAL MECHANISM, UNFORTUNATELY SOMETIMES INCREASED -- SO OUR IRB CON FRONTED US SAYING WHAT'S GOING TO HAPPEN WITH THIS, IS THIS GOING TO HAPPEN. BUT OUR CONTEXT IS VERY, VERY DIFFERENT FROM THE STATES. SO LOCAL -- THAT THIS IS A CURSE, SCHIZOPHRENIA, OR THAT -- VARIOUS OTHER KEUNDZ OF KINDS OF THINGS. SO WE'RE STILL WORKIN ON THIS, WE HAVE AN ONGOING PROJECT WHICH ADDRESSES THIS ISSUE IN SOME DETAIL USING FOCUS GROUPS AND OTHERS, SOME OF THE QUALITATIVE METHODOLOGIES, BUT I'M NOT CONVINCED AS WHAT I WANT TO SAY THAT THE FINDINGS THAT ARE FROM THE WEST WILL NECESSARILY HOLD UP WITHIN OUR LOCAL CONTEXT. MORE ON THAT AT ANOTHER POINT IN TIME IN THE FUTURE. SO IN CONCLUSION, IT'S BEEN A REMARKABLE EFFORT BY NIH AND WELLCOME, IT ADDRESSES AN IMPORTANT GAP, WE THINK, MARY KAY WILL MARY-CLAIRE WILL SAY MORE ABOUT THAT, PAYING MORE ATTENTION TO HEALTH EQUITY ISSUES AND CONSIDERATIONS. THANK YOU TO THOMAS AND TO THE NIH. [APPLAUSE] >> THANK YOU, DAN. SO I SHOULD HAVE MENTIONED THAT NIH AND THE WELLCOME TRUST ARE NOT THE ONLY ENTITIES THAT ARE INTERESTED IN EXPANDING GENETIC RESEARCH TO ANCESTRAL POPULATIONS. AS THE LEADER OF THE STANLEY FOUNDATION IS ALSO INVOLVED IN SETTING UP LARGE PROJECTS AND WE ACTUALLY ARE HOPING TO DEVELOP ANOTHER PUBLIC-PRIVATE PARTNERSHIP IN THIS RESPECT. NEXT SPEAKER DOES NOT NEED AN INTRODUCTION. IT'S DR. MARY-CLAIRE KING FROM THE UNIVERSITY OF SEATTLE. I USED TO SAY SHE'S KNOWN FOR HER CANCER WORK AND CIVIL RIGHTS WORK, NOT FOR PSYCHIATRIC -- IMPLIED THAT SHE'S NOT KNOWN FOR HER WORK IN PSYCHIATRIC GENOMICS, HOWEVER, THAT IS NOT TRUE ANYMORE, MARY-CLAIRE, SO YOU OWN THAT TOO. WELCOME, DR. MARY-CLAIRE KING. >> THANKS VERY MUCH, DAN, FOR INTRODUCING OUR PROJECT SO BEAUTIFULLY. IT'S A REMARKABLE HONOR TO SPEAK IN FRONT OF SUCH A REMARKABLE GROUP OF PEOPLE AS MY COLLEAGUES ON THIS PROJECT. WE'RE STUDYING THE GENETICS OF SCHIZOPHRENIA IN THE SOUTH AFRICAN XHOSA. 1100 CASES, 1100 CONTROLS, SAME AGE AND SEX DISTRIBUTION, ALL ARE XHOSA. INTERVIEWS CAN CARRIES OUT IN XHOSA BY NURSES WITH SUBSEQUENT TRANSLATION INTO ENGLISH, AND THEN CONSENSUS DIAGNOSIS BY PSYCHIATRIST INVESTIGATORS IN SOUTH AFRICA. DAN STEIN HIMSELF, GOODMAN, WHO I REALLY WANT TO MEET, THIS GUY IS AMAZING ON THE PHONE AND ON EMAIL. AND THEIR COLLEAGUES. SO WE HAVE CONSENSUS DIAGNOSIS, MORE IN A MOMENT. WE CARRY OUT EXOME SEQUENCE AND CNV SNP AL NAY SEES ON ALL OF OUR PARTICIPANTS AND WHOLE GENOME SEQUENCE ON A SUBSET OF THEM. SO WHY STUDY SCHIZOPHRENIA IN THE XHOSA? ACTUALLY BECAUSE MY EYES ARE BAD, I'M ACTUALLY GOING TO CHANGE SOMETHING ABOUT THE WAY I'M SEEING THIS. FORGIVE ME FOR ONE MOMENT. IT'S NOT BECAUSE THE XHOSA HAVE MORE SCHIZOPHRENIA. THEY DO NOT. THEY HAVE PRETTY MUCH THE SAME RATES AS EVERYONE ELSE ON THE PLANET. THE REASON FOR STUDYING SCHIZOPHRENIA IN THE XHOSA ARE THAT THEY HAVE UNIQUE FEATURES THAT MAKE THEM PARTICULARLY FASCINATING FOR US. FIRST AS I SHALL EXPLAIN MORE IN JUST A MOMENT, THEY OFFER US A UNIQUE WINDOW ON TO A CERTAIN PART OF HUMAN GENETIC VARIATION, A CERTAIN PART OF OUR SHARED EVOLUTION. SECOND, THEY'RE, THERE ARE UNIQUE CULTURAL FEATURES ON THE WAY THIS DIAGNOSIS IS PERCEIVED BY THOSE WHO HAVE IT IN THIS POPULATION. THERE'S A LOVELY PAPER BY MEGAN CAMPBELL ALREADY ON THIS POINT IN WHICH SHE EVALUATES THE WAY THAT OUR XHOSA PARTICIPANTS INTERPRET THEIR OWN ILLNESS. THIRD, IT'S A UNIQUE OPPORTUNITY TO STUDY THE INTERACTION OF THE ESTIMATES OF SCHIZOPHRENIA WITH HIV INFECTION WHICH AFFECTS A LARGE NUMBER OF OUR PARTICIPANTS, AND OF COURSE FOURTH AS DAN HAS ALREADY EXPLAINED, IT'S A FABULOUS OPPORTUNITY FOR CAPACITY BUILDING IN COLLABORATION WITH H3 AFRICA. SO TO RETURN TO THIS QUESTION ABOUT WHAT KIND KIEND OF UNIQUE INSIGHT DO WE HAVE WITH RESPECT TO HUMAN VARIATION BY STUDYING THE XHOSA, LET ME PUT THIS IN A BIT OF AN EVOLUTIONARY CONTEXT. HUMANS AND CHIMPANZEES DIVERGED 5 OR 6 MILLION YEARS AGO. THE OUT OF AFRICAN DIVERGENCE OF MODERN HUMANS WERE BETWEEN 100,000, 50,000 YEARS AGO AT THEIR ONSET. WHAT HAPPENED IN BETWEEN? NOT NOTHING. A GREAT DEAL OF VARIATION ACCUMULATE THE IN ACCUMULATED IN AFRICA, BUT AMONGST THE INDIVIDUALS WHO LEFT, THEY WERE A VERY SMALL FRACTION OF THE PEOPLE LIVING IN THE ENTIRE CONTINENT, THE ENTIRE HOME OF MODERN HUMANS. SO WHEN WE WISH TO STUDY RARE VARIANTS, WHAT WE CALL YOUNG ADULT VARIANTS IN THE WORLD, THE BEST PLACE TO LOOK FOR THOSE VARIANTS IS AFER KA. AFRICA. HOW DO I KNOW THIS? THERE IS ENRICHMENT IN AFRICAN POPULATIONS IN RARE VARIANTS, WHETHER WE LOOK AT THE XHOSA POPULATION OR THE EASE AFRICAN EAST OR WES T AFRICAN POPULATION EXAIFERRED COMPARED TO EUROPEANS AND EAST ASIANS IN THE THOUSAND GENOME PROJECT. SO WE HAVE AN OPPORTUNITY BY STUDYING AN AFRICAN POPULATION WITH DEEP ROOTS AND BY COMPARING CASES FROM THAT POPULATION WITH CONTROLS CAREFULLY SELECTED FROM THAT POPULATION TO UNDERSTAND IF THIS PARTICULAR SLICE OR TRANCH OF VARIATION CONTRIBUTES TO THE PHENOTYPE OF INTEREST. IT'S NOT THAT THIS CLASS OF VARIANTS DON'T EXIST IN EUROPE AND ASIA, CLEARLY THEY DO. BUT THEY ARE ENRICHED IN AFRICAN POPULATIONS. SO HOW ARE WE CARRYING THIS OUT? FIRST THE MATERIALS WE RECEIVE FROM OUR COLLEAGUES IN AFER CAR, WE CARRIED OUT WHOLE EXOME SEQUENCE, THIS WAS OUTSOURCED TO THE BAYLOR HUMAN GENOME SEQUENCING CENTER. AND THEN WE FILTERED IN SEATTLE AND WHEN I SAY WE, I MEAN -- WE FILTERED BY PRINCIPAL COMPONENTS ANALYSIS AND BY ALZHEIMER'S DISEASE MIXTURE BY AD MIXTURE ANALYSIS AND QUALITY CONTROL TOOLS IN ORDER TO BE ABLE TO EXCLUDE SOME SAMPLES FROM SOME ANAL ANALYSES. FIRST WE EXCLUDED 28 SAMPLES WHICH WILL BE RE-RUN BECAUSE THEY HAVE QUALITY ISSUES, THAT'S A SMALL SUBSE. THEN FOR PART OF THE ANALYSES, I THINK ACTUALLY FOR BOTH OF THE ANALYSES I'M GOING TO DISCUSS HERE, WE EXCLUDED INDIVIDUALS WHO WERE INTENTIONALLY DUPLICATED JUST IN ORDER TO CHECK US TO MAKE SURE WE GOT THE SAME THING TWICE, OR INDIVIDUALS WHO UNBEKNOWNST TO BOTH US AND OUR COLLEAGUES WERE EITHER PARENT-CHILD OR SIB-SIB OF EACH OTHER, AND THEN FOR PART OF THE ANALYSIS I'LL DISCUSS, WE ALSO EXCLUDED INDIVIDUALS WHO HAD SUBSTANTIAL CONTRIBUTIONS FROM OTHER POPULATIONS THAT WE DIDN'T HAVE EXPLICITLY IN OUR COLLECTION. SO FOR EXAMPLE HERE. AND WE THINK MOST OF THIS AD MIXTURE, WHICH US A SEE IT IS FROM A STRUCTURE ANALYSIS, MOST OF THIS ALZHEIMER'S DISEASE MIXTURE IS FROM THE -- SO WHAT IS OUR ANALYTIC DESIGN? OUR ANALYTIC DESIGN IS DESIGNED TO CAPTURE THIS LARGELY AFRICA-SPECIFIC CLASS OF YOUNG ADULT VARIATION. WE WILL, OF COURSE, ALSO CAPTURE IN A PRIVATE VARIANCE IN A DE NOVO VARIANCE, AND THOSE PRESUMABLY OCCUR IN AFRICAN POPULATIONS JUST LIKE THEY DO EVERYWHERE ELSE IN THE WORLD. THEY WILL BE INCLUDED IN THIS ANALYSIS, BUT WE AREN'T GOING AFTER THEM PARTICULARLY. IN THE NEXT GENERATION OF THIS PROJECT, WE WILL EVALUATE TREE OWES TRIOS, BUT THIS ANALYSIS WILL CAPTURE YOUNG ADULT VARIATION OF THE RARE BUT NOT ULTRA, ULTRA-RARE VARIETY, AS WELL AS THE PRIVATE VARIANTS. BY AND LARGE, WE WON'T BE ABLE TO TELL WHICH IS WHICH. THE ANALYSIS OF COURSE WILL BE CASE CONTROLLED AND AGAIN AND GENE BY GENE. OUR GOAL IS TO IDENTIFY GENES THAT ARE INTERESTING. AND HOW DO I DEFINE AN INTERESTING GENE? ONCE WE DEFINE THESE INTERESTING GENES, WE'RE GOING TO COUNT THE NUMBER OF CASES THAT HAVE A VARIANT IN AN INTERESTING GENE VERSUS THE NUMBER OF CONTROLS THAT HAVE A VARIANT IN AN INTERESTING GENE. WE'RE GOING TO DEFINE INTERESTING GENES GENETICALLY, NOT FUNCTIONAL LI AND WE'RE GOING TO DEFINE THEM AS GENES THAT HAVE INTERESTING VARIANTS. HOW DO I DEFINE AN INTERESTING VARIANT? AN INTERESTING VARIANT IS ONE THAT IS XHOSA HIGH PRESSURE SPECIFIC, THAT IS, SPECIFIC TO THE POPULATION WE CARE ABOUT SPECIFIC TO AFRICA, THAT IS& EITHER CASE-SPECIFIC OR CONTROL-SPECIFIC, SO IT'S FOUND IN ONE AND NOT THE OTHER. COULD BE ONE CASE IN ZERO CONTROLS OR FIVE CASES IN ZERO CONTROLS OR FOUR CONTROLS IN ZERO CASES, BUT IT'S SPECIFIC TO ONE GROUP OR THE OTHER IN THIS FIRST STEP OF THE ANALYSIS. AND IT IS PREDICTED TO BE DAMAGING TO GENE FUNCTION. HOLD THAT THOUGHT FOR A SECOND. I'LL TELL YOU WHAT THAT MEANS IN A MOMENT. AND WE NOW HAVE A TOTAL OF MORE THAN 1700 EXOMES COMPLETE, IN ORDER TO CARRY OUT THIS ANALYSIS, AS YOU WILL APPRECIATE THE WAY THAT I'VE DEFINED INTERESTING GENE, IS STEND DEPEND DENT ON SAMPLE SIZE SO WE NEED TO HAVE THE SAME NUMBER OF CASES AND -- FOR THIS ANALYSIS, FOR TODAY, BECAUSE WE HAVE 660& CONTROL EXOMES, I RANDOMLY SELECTED 660 CASES TO MATCH TO THEM. WE ACTUALLY HAVE MORE EXOMES THAN THIS BUT WE DON'T HAVE MORE CONTROL EXOMES THAN THIS. FAIR ENOUGH? STO WHATSO WHAT CONSTITUTES FOR TODAY AB INTERESTING VARIANT? AN INTERESTING VARIANT IS A VARIANT PREDICTED TO BE DAMAGING TO GENE FUNCTION. THE WAY THAT WE HAVE DEFINED THIS IS THAT THIS VARIANT MAY BE EITHER A STOP, THAT IS A NONSENSE OR A FRAME SHIFT OR A SPLICE WHERE IN A GENE THAT IS PREDICTED ITSELF AS A GENE TO BE INTOLERANT OF STOPS. AND FOR THE SAKE OF THAT PREDICTION, WE USED BOTH THE LOSS OF FUNCTION LIKELIHOOD CALCULATIONS OF THE EXACT CONSORTIUM DATABASE AND THAT SCORE HAD TO BE GREATER THAN .9 AND WE USED DAVE GOLDSTEIN'S RBIF SCORES, OH BOTH BOTH OF THOSE HAD TO BE TRUE FOR THE GENE TO QUALIFY AT ALL. SO A STOP OR A SPLICE IN SUCH A GENE COUNTED. FOR A SPLICE TO BE CONSIDERED, THE SPLICE HAD TO HAVE A SCORE THAT DROPPED AT LEAST 90%. THIS IS CLEARLY ONE OF THESE STATEMENTS FOR THE -- YOU WILL APPRECIATE A SCORE THAT DROPS 90% IS VERY, VERY LIKELY TO -- THEREFORE TO BE EITHER REMOVING AN IN FRAME PIECE OF THE GENE OR TO CREATE A STOP. SO THAT'S HOW WE DEFINED NONSENSE FRAME SHIFT OR SPLICE LOSS, OR THE GENE COULD HAVE A MISSENSE BUT WITH VERY NARROW CONSTRAINTS. IN THIS CASE, THE GENE HAD TO ALSO QUALIFY AS INTOLERANT OF MISSENSES BY HAVING AN COMPACT MISSENSE Z SCORE THAT'S GREATER THAN 2, SO 2 STANDARD DEVIATIONS OUT FROM WHAT WE WOULD EXPECT AS THE NUMBER OF OBSERVED MISSENSES TO BE TOLERATED SO IT'S 2 STANDARD DEVIATIONS ACTUALLY BELOW WHAT WOULD BE TOLERATED, AND THE MISSENSE HAD TO OCCUR IN OUR CASE OR CONTROL AT A SITE COMPLETELY CONSERVED IN TETRAPODS, NO EXCEPTIONS. SO LET ME GIVE YOU A COUPLE OF EXAMPLES. ONE EXAMPLE IS AQR, AQUARIUS, RNA SPLICING REGULATOR, WHICH HAS AN EVENT THAT QUALIFIES BY THE CRITERIA I JUST TOLD YOU ABOUT, IN FIVE CASES AND ZERO CONTROLS. THE PARTICULARS ARE THAT THERE WAS A FRAME SHIFT IN THREE OF THESE CASE, THE SAME FRAME SHIFT, A SUPPLIES SPLICE IN ONE CASE AND A MISSENSE IN ONE CASE. SO THIS IS A CASE GENE FOR THE PURPOSE OF THIS ANALYSIS. IN ONE OF THE ANKRDs, BY CHANCE ALL OF THESE WERE MISSENSES, FIVE DIFFERENT MISSENSES, SO THIS MAKES ANKRD17 A CONTROL GENE. YOU GET THE IDEA. MOST OF THE VARIANTS IN THE STUDY WERE PRESENT IN ONLY ONE CASE OR ONE CONTROL AS YOU MIGHT EXPECT. NOT SURE IF THAT WILL CONTINUE WHEN WE HAVE TWICE THE SAMPLE SIZE BUT THAT'S WHERE WE ARE NOW. SO OVERALL, OVERALL, WE FOUND 63 DIFFERENT GENES THAT HARBORED 154 DIFFERENT VARIANTS THAT QUALIFY IN CASES AND 42 DIFFERENT GENES THAT HARBOR 103 DIFFERENT VARIANTS THAT QUALIFY IN CONTROLS. SO HOW MANY CASES AND HOW MANY CONTROLS HAVE THESE VARIANTS? 139 OF OUR RANDOMLY SELECTED 660 CASES HAV E A VARIANT OF 1 OR MORE, AND 94 OF OUR 660 CONTROLS HAVE AT LEAST ONE QUALIFYING VARIANT, THE ODDS RATIO IS 1.6, IT'S SIGNIFICANT AT A .001 LEVEL. WILL OF COURSE BE INTERESTED TO FOLLOW WHEN WE HAVE LARGER SAMPLE SIZE. BACK IN OUR CELL PAPER IN 2013, WE DEFINE A NUMBER OF INTERESTING FEATURES OF SCHIZOPHRENIA GENES AND SHOW THAT IN THE DE NOVO CONTEXT, THESE FEATURES WERE ENRICHED AMONGST CASES COMPARED TO CONTROLS OR CASES THAT WERE ENRICHED IN CARRYING THESE CLASSES OF FEATURES COMPARED TO THEIR UNAFFECTED SIBLGHTS. SIBILITIES. AFFECTED SIBLINGS. A DE NOVO EVENT THAT'S BEEN REPORTED BY ANY COLLEAGUE IN SCHIZOPHRENIA OR AUTISM IN THE GENE, THAT THE GENE IS COEXPRESSED DURING DEVELOPMENT WITH GENES IN THE NETWORK OF SCHIZOPHRENIA GENES THAT WE DEVELOPED. THAT THE DPEEN IS GENE IS INVOLVED IN SYNAPTIC FUNCTION OR HAS ONE ISOFORM THAT HAS SYNAPTIC EXPRESSION. SOME WILL HAVE NONE OF THESE FEATURES BUT IT IS A CLASS OF INTERESTING FEATURES OF GENES THAT MAY BE INVOLVED IN SCHIZOPHRENIA. AT PRESENT, WITH 660 CASES AND 660 CONTROLS, CASES ARE ENRICHED FOR EACH OF THESE FEATURES COMPARED TO CONTROLS, BUT AT THIS MOMENT, NONE OF THOSE DIFFERENCES ARE SIGNIFICANT, THEY'RE ALL HOVERING ABOUT. SO WE SHALL SEE WHAT HAPPENS WHEN WE HAVE LARGER SAMPLE SIZE. NOW, WHAT IS STEP 2? STEP 2 IS TO TAKE OUR CANDIDATE GENES THAT WE'VE DEFINED BY STEP 1 AND EXTEND THEM A BIT. LET ME GIVE YOU ONE CAM, THE EXAMPLE OF CHD9. THIS IS A CHROMO DOMAIN DNA BINDING GENE, WHICH WITH RESPECT TO EXONS 2 THROUGH THE THREE PRIME -- 2 THROUGH 38 IS MORE THAN 90% IDENTICAL IN A HUMAN SEQUENCE TO SEE CHD8, KNOWN TO MANY OF YOU AS A CENTRAL GENE IN AUTISM AND BRAIN DEVELOPMENT. SO CHD9 QUALIFIED FOR OUR INTERESTING GENE GROUP BY VIRTUE OF HAVEING FOUR OR I GUESS FIVE DIFFERENT EVENTS, FIVE XHOSA CASES SPREAD ACROSS FOUR DIFFERENT MISSENSE MUTATIONS AS SHOWN HERE, AT LEAST MISSENSE AS QUALIFIED UNDER OUR CRITERIA FOR MISS MISSENSES. CHD9 HAS ALSO BEEN REPORT TODAY HARBOR DE NOVO EVENTS, MUTATIONS THAT ARE MISSENSES, IN CASES OF AUTISM, INTELLECTUAL DISABILITY AND DEVELOPMENTAL DELAY AS SHOWN HERE. IN INDIVIDUALS REGARDLESS OF ANCESTRY. SO WHAT WE DECIDED TO DO NEXT WAS NOW GO BACK TO THIS GENE AND SAY, WHAT ELSE IS GOING ON IN THIS GENE IN OUR SERIES? AND HERE'S WHAT YOU SEE. SO WE DID A COUPLE OF THINGS NOW. WE RELAXED OUR DEFINITION OF WHAT WOULD COBS TEUT CONSTITUTE AN INTERESTING VARIANT IN A COUPLE OF WAYS. MOST IMPORTANTLY WE SAID THE VARIANT WOULD STILL BE XHOSA SPECIFIC, BUT IT NEED NOT BE DIAGNOSIS-SPECIFIC. NOW WE'RE GOING TO LOOK FOR VARIANTS THAT ARE XHOSA VARIANTS, SO THEY'RE IN THIS MIDDLE AGED OR YOUNG ADULT GENE GROUP, BUT THEY MAY BE PRESENT IN CASES OR THEY MAY BE PRESENT IN CONTROLS, THEY MUST STILL BE DAMAGING, BUT I'M GOING TO RELAX THE DEFINITION OF WHAT CONSTITUTES A DAMAGING MISSENSE A LITTLE BIT TO ALLOW DURING TETRA POD EVOLUTION FOR THERE TO BE ONE OR TWO TETRA PODS THAT HAVE DEVELOPED SOME OTHER RESIDUE AT THIS SITE. SO LONG AS IT IS NOT THE VARIANT RESIDUE THAT WE SEE IN OUR CASE OR OUR CONTROL. A LITTLE CONVOLUTED BUT I WANTED TO HAVE A SLIGHTLY RELAXED CONSERVATION CRITERION. SO WE NOW COUNT ALL OF THE VARIANTS IN CHD9 THAT FIT THIS NEW -- THES RELAXED DEFINITIONS, AND AS YOU CAN SEE, THERE ARE NOW FOUR NEW VARIANTS THAT COME ON THE LINE AND THREE OF THEM APPEAR EACH IN ONE CASE AND ONE OF THEM APPEARS IN THREE CASES AND ONE CONTROL. I LEFT OWN ONE CRITERION, LESS THAN 1 IN 200 CASES, THEY DO NOT NEED TO BE DIAGNOSIS-SPECIFIC. SO IN ALL, FOR THIS -- EXCUSE ME FOR THIS PARTICULAR GENE, WE HAVE 10 XHOSA CASES AND ONE XHOSA CONTROL THAT HAVE A DAMAGING VARIANT. CHD9 IS GENOME-WIDE ONE OF THE GENES THAT IS THE MOST INTOLL RAPT TO INTOLERANT TO STOPS, SO IT DOES NOT SURPRISE US AS ALL NOR OTHERS WHO WORKED IN SCHIZOPHRENIA AND AUTISM TO SEE NO STOPS IN THIS GENE. IT IS LIKELY THAT EVEN -- INSUFFICIENCY IS DEVASTATING WITH RESPECT TO THIS GENE. SO THIS IS AN EXAMPLE OF THE KIND OF ANALYSIS WE CAN CARRY OUT IN STEP 2 ACROSS ALL OF THE GENES THAT WE ENCOUNTER OF COURSE WITH A SOMEWHAT LARGER SAMPLE SIZE, AND WE HOPE AS WE JUST HEARD FROM MATT STATE, WE HOPE THAT WE CAN IDENTIFY A FEW GENES THAT ARE VERY GOOD CANDIDATES IN AND OF THEMSELVES BY THIS STEP WISE MANEUVER, EVEN THOUGH WE WON'T KNOW CASE BY CASE WHETHER THE EVENT IS DE NOVO OR NOT. I'LL TURN JUST VERY BRIEFLY TO THE QUESTION OF COPY NUMBER VARIANTS. THE CMV ARRAY ANALYSES WILL INCLUDE BOTH DUPLICATIONS AND DELETIONS IN THE ANALYSIS, AND WE'LL HAVE THOSE LATER PRIMARILY FROM OUR FRIEND FROM CAPE TOWN, BUT I'M GOING TO TALK JUST FOR A MOMENT ABOUT DELETION ANALYSIS, ALSO WORK CARRIED OUT BY GULSONAR. SO THEY USED OUR EXOME DATA TO IDENTIFY DELETIONS, SO IT'S NOT NECESSARY CASES AND CONTROLS BE DEFINED IN THE SAME WAY, THERE'S NO BIAS WITH RESPECT TO THE WAY THE EVENT IS DEFINED AND OF COURSE ONE USES THE APPROPRIATE DENOMINATOR. SO WE HAVE FOR THIS EVENT, FOR THIS ANALYSIS, WE HAVE 823 CASES AND 677 CONTROLS THAT PASSED QC FOR CNVs, WHICH YOU WILL APPRECIATE IS TOUGH QC FOR OBJECTION EXOME DATA. WE USE CONFER TO IDENTIFY DELETIONS THAT HAVE AT LEAST THREE ADJACENT EX-ONCE, WE USED PLINK FOR SEGMENTATION AND OVERLAP. THEN THE MINIMAL LENGTH FOR EACH OF THESE EVENTS, BEARING IN MIND EACH MUST DELETE AT LEAST THREE ADJACENT EXONS, WAS IDENTIFIED USING MANUAL CURATION OF THE Z-SCORES. AND THE CONCLUSION THUS FAR FROM THIS ANALYSIS IS THAT 54 OF 823 CASES, SO 6.5% AND 25 OUT OF 677 CONTROLS OR 3.7% CARRY A RARE DELETION, DELETION THAT FIT ALL THE PREVIOUS QUALIFICATIONS. OF COURSE NOT ALL OF THESE WILL BE DEVASTATING EVENTS BUT IT GIVES US A SENSE THAT THERE IS ENRICHMENT IN THE CASES AND THIS IS SOMETHING TO PURSUE. IN THIS ANALYSIS, THE MEDIAN EVENT SIZES IS -- ALL OF THIS MAKES THE EVENT AS SMALL AS POSSIBLE AND DOES NOT INCLUDE THE INTRONS THAT FLANK THE THREE EXON OR GREATER DELETION. WE'LL OF COURSE HAVE MUCH BET ARE DATA ON THIS FOR THOSE INDIVIDUALS WHO HAVE WHOLE GENOME SEQUENCE AND INDEED WE'LL HAVE MUCH BETTER INFORMATION FROM THE SNP OR ACNVs. AGAIN, THIS IS SIGNIFICANT THIS TIME AT P.01. SO WHAT ARE THE NEXT STEPS IN THIS ANALYSIS? WE'LL COMPLETE ENROLLMENT AND SEQUENCING OF ALL OF OUR CASES AND CONTROLS. WE'LL EVALUATE POINT MUTATIONS AND CNVs AS I HAVE EXPLAINED, WE'LL SCREEN INDIVIDUALLY INTERESTING GENES IN OTHER COHORTS, WE'LL SEE IF ANY OF THE OUT OF AFRICA POPULATIONS CAN OFFER THE RICHNESS OF THE XHOSA. PLACE YOUR BET. AND WE'LL GENERATE AND CHARACTERIZE SOME OF THESE MUTATIONS IN MODEL SYSTEMS, BOTH IN MODEL ORGANISMS THAT WE WORK WITH IN OUR GROUP AND IN HUMAN EMBRYONIC STEM CELL LINEAGES THAT WE CAN DIFFERENTIATE INTO NEURONAL LINEAGES. SO I WANT TO CONCLUDE WITH A VOTE OF ENORMOUS GRATITUDE TO THOMAS FOR PUTTING US TOGETHER. THIS PROJECT BEGAN BECAUSE BASICALLY OF TWO PAIR WISE FRIENDSHIPS, ONE BETWEEN RAJ AND ME THAT GOES BACK TO 1994 AND THE OTHER BETWEEN DAN STEIN AND EZ RAH WHO ARE, WE DECIDED LAST NIGHT, THIRD COUSINS THROUGH ZENA STEIN. AND A REMARKABLE SET OF SOUTH AFRICAN COLLABORATORS WHO HAVE CARRIED OUT ALL OF THE ASCERTAINMENT, DIAGNOSIS, AND SAMPLING AND ENROLLMENT AND CONSENTING OF THIS VERY FUN GROUP OF CASES AND CONTROLS. A MEAN YON THANKS. [APPLAUSE] A MILLION THANKS. [APPLAUSE] >> THANK YOU, MARY CLAIRE. >> NO DISCUSSION? >> UNFORTUNATELY NO DISCUSSION. WE HAVE PROBABLY A LITTLE TIME A LITTLE LATER, BUT NOT NOW. I ACTUALLY HAVE TO ANNOUNCE A CHANGE OF THE PROGRAM BECAUSE OF SCHEDULE CONFLICTS. SO WE WANT TO STAY A LITTLE BIT LONGER IN THIS SPACE OF -- HOWEVER, WE'LL TAKE A LITTLE BREAK AND MOVE TO RARE DISORDERS WITH KNOWN GENETIC LESIONS. THE PARTICULAR ONE WE'RE GOING TO DISCUSS IS -- STILL HAS 20 PLUS GENES IN THE DELETION REGION, SO MANY UNANSWERED QUESTIONS AND MORE QUESTIONS TO ASK. BUT I AM DELIGHTED TO WELCOME DR. RAQUEL GUR FROM UNIVERSITY OF PANCREATIC ADENOCARCINOMA TO TELL US ABOUT THE CONSORTIUM EFFORTS THAT WE HAVE PUT INTO THIS SPACE. >> I CAN START EVEN BEFORE THE SLIDES, THANKING YOU FOR ACCOMMODATING US. WE HAVE INVESTIGATIVE MEETINGS IN GENEVA, SO THOSE OF YOU WHO FIRST WERE ATTENDING, BEVERLY WILL TAKE OVER RIGHT AFTER NIGH TALK TO GO TO THE INVESTIGATOR MEETING THAT WAS DECIDED A YEAR AGO, BEFORE WE KNEW ABOUT THIS. SO THE STUDY OF PSYCHOSIS AS MANY OF YOU KNOW, SCHIZOPHRENIA IS PART OF IT, IN RECENT YEARS, PROBABLY IN THE PAST DECADE, HAVE SHIFTED TO THE LEFT, MAINLY AN EFFORT TO IDENTIFY PEOPLE BEFORE FULL BLOWN PSYCHOTIC ILLNESS. THIS PROVIDES AN OPPORTUNITY TO LOOK AT THE EMERGENCE OF PSYCHOSUS AND HOPEFULLY, AS WE MOVE FORWARD, TO EARLY INTERVENTION. SO THESE POPULATIONS HAVE BEEN COMMONLY BEEN THE HEALTH SEEKING INDIVIDUALS WHO PRESENT WITH PSYCHOTIC SYMPTOMS. EPIDEMIOLOGIC STUDIES LOOKING AT THE COMMUNITY AND SCREENING, AND THE STORY THAT I WILL TELL YOU TODAY ABOUT IS ON 22Q11DS DELETION SYNDROME, THAT IS A NEUROGENIC DISORDER, HIGHLY INFORMATIVE, WHERE WE TAKE THE POSITION AT THE IBBC THAT THE RARE CAN INFORM THE COMMON. AND I WILL SHOW YOU DATA, ESPECIALLY ON THE PHENOTYPIC SIDE, THAT WILL BOLSTER THIS POINT AND ALSO EARLY GENOMIC DATA. SO THROUGHOUT THIS EFFORT, WE CAN TRANSVERSE IN GOING FROM BASIC RESEARCH TO TRANSLATIONAL AND CLINICAL POPULATION AND IN THIS STUDY OF RARE DISORDERS, WE KEEP IN MIND THE PATIENT AND THE FAMILIES WHO WOULD LIKE TO PROVIDE CAREFUL AS WE'RE DOG THE STUDY OF NON-DELETED INDIVIDUAL WITH SCHIZOPHRENIA. SO THIS NEUROGENIC DISORDER PREVIOUSLY CALLED STILL CALLED -- SYNDROME IS RELATIVELY RARE, 1 IN 3 THOWR IT 4,000 LIVE BIRTH, THOUGH -- SOME OF THE FETUSES ARE NOT COMPATIBLE WITH LIFE. MICRO DELETION OF 3 MEGABYTE ON THE LONG ARM OF CHROMOSOME 22, THE PRESENTATION LIKE EVERYTHING ELSE IN GENETICS IS VERY HETEROGENEOUS. IMPORTANTLY IS THE TWIN, THE CO-MORBIDITY THAT PRESENTED VARIOUS ORGAN SYSTEMS ARE CARED FOR AND NOW DAYS ARE CARED FOR MUCH BETTER THAN THEY USED TO THROUGH SURGICAL AND MEDICAL INTERVENTIONS. THE PROMINENT NEUROPSYCHIATRIC FEATURES MAINTAIN, AND THEY RELATE TO DEVELOPMENTAL DELAYS, COGNITIVE IMPAIRMENT WITH A RANGE, SOME OF THEM BEING WITHIN NORMAL IQ, ADHD IN YOUNG PEOPLE, ANXIETY DISORDERS AND, REMARKABLY, 25 TO 30% OF THEM PRESENT WITH SCHIZOPHRENIA. THEREFORE, THIS WORD "DISORDER" IS WORTH TO PURSUE. AND THIS IS A SLIDE THAT WAS TAKEN FROM DONALD -- COLLABORATOR IN THE GROUP THAT SHOWS THE PRESENTATION, THE DELETION REGION, AND MOST OF THE PRESENTATIONS ARE FROM A TO D, ALTHOUGH 90%, A SMALL PERCENTAGE ARE ALSO FROM BA TO D AND C AND THIS PROVIDES A UNIQUE OPPORTUNITY TO LOOK AT THE EFFECT OF THE DOSE OF THE DELETION. HOW DID THE IBBC DO INTERNATIONAL BRAIN BEHAVIOR CONSORTIUM WAS BORN, IT WAS BASED ON INTENSIVE COLLABORATION AMONG INVESTIGATORS IN THE FIELD -- IN TORONTO WAS LEADING THE EFFORT ON SCHIZOPHRENIA WELL BEFORE I JOINED THE EFFORT WITH SUPPORT OF NIMH. SO IT WAS BASED ON -- THAT THE DELIBERATIONS ARE CRITICAL FOR PROGRESS IN THE FIELD. EVERY SMALL SAMPLES ACROSS THE GLOBE IS NOT GOING TO ADVANCE THE UNDERSTANDING OF THE DELETION. WE NEED LARGE SCALE STUDIES AND WE NEED COMPLEMENTARY EXPERTISE IN ORDER TO ADVANCE UNDERSTANDING OF COMPLEX ORDERS, AND THIS WILL PROVIDE THE PLATFORM OF MULTIPLE LEVEL OF ANALYSIS, ANIMAL SPECIFIC MODEL TO ADVANCE MECHANISTIC UNDERSTANDING IN IPSC. ALL OF THIS IS HAPPENING. THIS IS A MAP, I DIDN'T KNOW BEFORE THAT OTHERS WILL BE USING THIS SAME MAP AS EVIDENCE OF COLLABORATION, BUT YOU CAN SEE THE NORTH AMERICA, TORONTO, CHILE, WORKING WITH TORONTO, AND WE SEE MULTIPLE SIDES -- THAT ALSO INCLUDE ISRAEL AND AUSTRALIA, SO THE 22 SIDES WORLDWIDE THAT THEY'RE WORKING TOGETHER, NOW FOR ABOUT FOUR YEARS. THE GOAL BETWEEN ESTABLISHED -- HOW TO LOOK DIMENSIONALLY -- FORMATIVE PHENOTYPES ASSOCIATED WITH THE DEVELOPMENT OF SCHIZOPHRENIA TO PERFORM ALL GENOME SEQUENCING IN 600 AND GENERAL -- IN 1,000 INDIVIDUALS, AND TO DEVELOP THIS ALL HAS BEEN RETROSPECTIVE. IN OTHER WORDS, WE RELIED ON AVAILABLE SAMPLES, PUT THEM TOGETHER TO COMMON PIPE LIKE PIPELINE AND TO ESTABLISH -- HOWEVER, WE RECOGNIZE THE NEED TO DEVELOP PILOT -- MEASURES IN ORDER TO OPTIMIZE THE ASSESSMENT IN A RICHER DATASET FOR PROSPECTIVE STUDIES AND, OF COURSE, WE PLAY BY THE RULES, ALL OF IT IS GOING TO BE IN THE PUBLIC DOMAIN SOON. SO THIS IS BRIEFLY WHERE WE ARE NOW. SO THESE ARE THE SITES THAT HAVE PARTICIPATED. THERE HAVE BEEN SEVERAL GENOMIC SITES THAT HAVE BEEN COLLABORATING BEFORE THE PROJECT WAS FUNDED. WE'VE BEEN DOCUMENTING THE COLLECTION OF DATA, CLOSE TO 2,000 SAMPLES. THESE SAMPLES DID NOT UNDERGO QC. SO WHEN WE GO FOR QC, THE NUMBER IS SMALLER BUT STILL IT WILL PROVIDE A SAMPLE OF CLOSE TO 1,500 THAT ALLOWS A SAMPLE OF THIS DISORDER AVAILABLE, AND AS YOU CAN SEE, THEY PROPOSED A GENOMIC APPROACH IS COMPLEX. BECAUSE THE DISORDER IS COMPLEX AND EXPERTISE IS REQUIRED DID NOT EXIST IN ONE PLACE. FIRST AND FOREMOST, WE START WITH VERY -- DELETION, SOME PEOPLE WILL SAY DELETED, THE EXTENT OF THE DELETION, THIS IS DONE BY BEVERLY -- AT CHILDREN'S HOSPITAL AND THEN GENOME WIDE SEQUENCING DATA, RELYING ON -- AT EINSTEIN -- AND GOING THROUGH THIS PIPELINE, THE GOAL IS TO ACHIEVE AN AN INTEGRATED GENOME WIDE ANALYSIS, CONDUCTED IN TORONTO. TO PROVIDE THE WHOLE GENOME SEQUENCING. I STARTED BY SAYING THIS IS A GOOD MODEL FOR SCHIZOPHRENIA. WHILE WAITING FOR THE GENOMIC ANALYSIS THAT IS LABOR INTENSIVE, AS YOU KNOW. WE HAVE CORRELATION THAT WILL SHOW IN THE NEXT FEW SLIDES, IS THAT -- PROVES THE POINT THAT THE PATTERN OF PRESENTATION OF PSYCHOPATHOLOGY AND PSYCHOSIS IS SIMILAR IN 22Q AS IT IS IN OTHER POPULATIONS. NONE DELETED. THEREFORE, IT IS A GOOD MODEL OF GOING FROM THE RARE TO THE COMMON. IF YOU LOOK HERE, CHILDREN IN GREEN, LESSONS IN YELLOW, YOUNG ADULTS IN ORANGE AND ADULTS IN THE POPULATION, YOU CAN SEE THAT -- MORE COMMON AMONG THEM IS ADHD, AND YOU SEE SOME THRESHOLD SIGH TO KICK FEATURES IN YOUNG KIDS. WE START WITH AGE 8. SO THESE ARE YOUNG CHILDREN. MOOD IS MORE EVIDENT IN OTHER LESSONS, SOME ESTIMATES ARE NOT COMMON IN ADULT BUT PSYCHOSIS, MEETING CRITERIA FOR SCHIZOPHRENIA SPECTRUM DISORDER, IS EVIDENT MORE IN ADULTS. IF YOU LOOK AT THE STRUCTURE OF PSYCHOPATHOLOGY, A SEMI STRUCTURED INTERVIEW IN 22Q IS EXACTLY THE SAME AS AGE-EQUIVALENT NON-DELETED YOUNG PEOPLE, THERE'S A GENERAL -- A FACTOR OF PSYCHOSIS SYMPTOMS MOOD AND ANXIETY JUST AS IT IS IN THE GENERAL POPULATION. IF WE LOOK AT THE AGE OF ONSET, IT'S A COMPLEX SLIDE, LITERATURE REVIEW, BUT SUFFICE IT TO SAY, THE LONGITUDINAL STUDIES OF 22Q, WE COULD SEE THE PSYCHOSIS UNFOLDING, EMERGING BEFORE YOUR EYES, ADOLESCENCE, EARLY ADULTHOOD IS THE TIME THAT IT EMERGES, AND THE MEAN AGE, THE EMERGENCE, 16.5, 2017 IS SIMILAR, MOVING TO SCHIZOPHRENIA SPECTRUM DISORDER, IT'S -- EXACTLY THE SAME AS IN THE NON-GENERAL -- IN THE GENERAL POPULATION. SEVERAL U.S. STUDIES PUBLISHED FROM THE GROUP ALREADY SHOW THESE ARE ALL SCHIZOPHRENIA SPECTRUM DISORDERS, WHEN WE MOVE TO AN OLDER AGE, MID 20s, YOU CAN SEE SCHIZOPHRENIA IS THE PREDOMINANT DIAGNOSIS FROM PSYCHOTIC OTHERWISE NOT SPECIFIED YOUNG PEOPLE, YOU CAN SEE THE GENERAL POPULATION. IF WE NOW LOOK AT THE IMPACT OF COGNITION IN THE POPULATION, THIS IS FROM A -- WHAT I'D LIKE TO HIGHLIGHT IS LONGITUDINAL STU CAN STUDIES SHOW THE GROUP OF PEOPLE THAT NOW HAVE A PSYCHOTIC ILLNESS, THE DISORDER IS A STEEPER SLOPE THAN THOSE WITH THE DELETION WITHOUT. THIS IS ATABLE BY ACCOUNTABLE BY THE DECREASE IN VERBAL IQ. THIS HAS BEEN REPORTED IN SCHIZOPHRENIA NON-DELETED INDIVIDUALS AS WELL. WITH A RECENT STUDY NOW IN PRESS, BY UNPUSSING IT -- VARIATION AND EXPRESSION IN THOSE WITH SCHIZOPHRENIA IN THE DELETION, AND THE DELETION WITHOUT SCHIZOPHRENIA. AND SOME OF THE FINDINGS SHOW THAT KNOWN GENES HAVE LOST SIGHT AND HAVE BEEN IMPLIMENTED PF BUT ALSO -- FINDINGS IN A NETWORK, ANALYSIS, SO DID SOME OF THE DELETION GENES IN THE DELETION REGION AND OTHER GENES LED BY WORLD GENOME WIDE CNVs, SO IT IS CONSISTENT JUST LIKE IN THE GENERAL POPULATION WITH A -- HYPOTHESIS OF SCHIZOPHRENIA IN 22Q. ONE OF THE QUESTIONS THAT DELETION CAN HELP US ASSESS IS INVERSION POLYMORPHISMS BETWEEN LCRs, -- THAT CAUSE THE GENOMIC DISORDERS, ONE OF THE GENETIC SITES FOR THE COLLABORATION HAS EXAMINED IT, LOOKING FOR DUPLICATION AND SMALLER SIZE OF THE DELETION AND CARRY INVERSIONS BETWEEN THEM. THEY'RE FINDING THE INVERSION PREDISPOSES TO MY TOMMY TO MEIOTIC REARRANGEMENTS. WHAT HAPPENS DURING THE INVERSION PROCESS. SO IN SUMMARY, IT IS CRITICALLY IMPORTANT TO ESTABLISH COLLABORATIONS OVERALL, BUT DEFINITELY FOR -- CM Vs -- ARE HIGHLY INFORMATIVE, PERSONAL EXPERIENCE OF SEEING THE KIDS TO BE CON VIPSED, YOU WANT SCHIZOPHRENIA EMERGING TRULY BEFORE YOU ARE -- IT'S UNBELIEVABLE. IT'S VERY POWERFUL, SORT OF GIVES US THE STRENGTH AND THE WILL TO DEAL WITH REALLY COMPLEX DISORDERS, SO IT'S A STRONG MODEL, WE NEED COMMON MEASURES AND DATABASE, THE IDBC WAS THE FIRST STEP BUT YOU CAN'T PLAY IN THE -- BY YOURSELF, YOU HAVE TO PLAY WITH OTHERS AND ESTABLISH COMMON TOOLS AND THEN YOU CAN DO WHATEVER YOU WANT TO AND DECIDE, BUT COMMON TOOLS IN ORDER TO BENEFIT THE PATIENTS, SCIENCE AND THE PATIENTS ARE CRITICAL IN THE ANALYSIS UNDER WAY, INCLUDING WOULD BE COMPLETED SHORTLY, RISK SCORE THAT IS HIGHLY PROMISING, THERE WILL BE A PAPER SOON. EE ANYTHING MA, CAN WE BUILD THEM, 500 TIMAGEING, VERY INTERESTING FINDING THAT SHOWS THAT THE DEGREE OF DELETION IMPACTS THE IMAGING FINDINGS AND LOOKING AT THE EXTREME PHENOTYPES BECAUSE OF THE HETEROGENEITY, AND WITH THIS, I'D LIKE TO THANK THE STRONG COLLABORATORS OF THE PART OF THE EXECUTIVE COMMITTEE. I WAS AN OUTSIDE IR. I FEEL WELCOME NOW. THOMAS AND GITA HAVE BEEN SUPPORTING FOR REGULAR PHONE CALLS, INDIVIDUALS IN THE FAMILIES, AND THIS IS OUR LOGO. THIS IS ALL SUPPORTED BY NIMH. THANK YOU. [APPLAUSE] >> THANK YOU, RAQUEL. WE ACTUALLY WILL HAVE -- WE'LL NEED TIME FOR SOME QUESTIONS AFT NEXT TALK. SO NOW WE'RE RETURNING TO AFRICAN-AMERICAN AD MIXTURE MAPPING, AND I WELCOME JOHN CONSTANCE TEEN FROM MY FAVORITE MEDICAL SCHOOL, UNIVERSITY OF WASHINGTON ST. LOUIS. SO IF YOU EVER NEED A BONE MARROW TRANSPLANT, I'D HIGHLY RECOMMEND THAT PLACE. >> THANK YOU. >> GLAD TO HEAR WASHINGTON U IS THOMAS' FAVORITE ME CAL SCHOOL, THAT'S FOR SURE. I'M A CHILD PSYCHIATRIST AT WASHINGTON U. WHILE THEY'RE GETTING UP THE SLIDES, I'LL JUST SAY THAT GIVING THIS TALK ON BEHALF OF A GENE DISCOVERY NETWORK THAT STARTED 10 YEARS AGO AS ONE OF THE AUTISM CENTERS OF EXCELLENCE, IT WAS FUNDED BY NIMH AND LED BY DAN -- OTHER CLLEAGUES ALL OVER THE COUNTRY, SO TALKING ON BEHALF OF THE ENTIRE GROUP, GOING BACK TO WHAT WAS SAID EARLIER ABOUT NEITHER TAKING RESPONSIBILITY OR ACCEPTING BLAME FOR ANY OF THE THINGS THAT I'LL PRESENT TODAY. BUT VERY HAPPY TO BE HERE AND GIVING THIS TALK ON BEHALF OF BOTH OF US. SO ONE OF THE ISSUES AT HAND FOR THE STUDY OF AUTISM IS DEPICTED ON THIS SLIDE FROM THE CENTERS FOR DISEASE CONTROL THAT SHOWS A SOMEWHAT OF A LOWER COMMUNITY PREVALENCE ESTIMATE FOR BLACK NON-HISPANIC INDIVIDUALS IN THE POPULATION FOR THE PREVALENCE OF AUTISM. THIS IS A MINOR DIFFERENCE IN COMMUNITY DIAGNOSIS WHEN IT'S KNOWN THAT THE REAL PREVALENCE OF THE CONDITION IS ACTUALLY THE SAME ACROSS ETHNICITIES AND RACE. BUT WHAT'S INTERESTING IS THAT THE PROPORTION OF CHILDREN WITH AUTISM WHO HAVE INTELLECTUAL DISABILITY AMONG BLACK AND NON-HISPANIC IS MARKEDLY ELEVATED IN COMPARISON TO WHITE AND YOU'RE PINE EUROPEAN SAMPLES. SO THINKING ABOUT WHAT THE POTENTIAL CAUSES OF THAT ARE, BOTH FROM A STANDPOINT OF GENETIC POSSIBILITIES AS WELL AS ALL OF THE SOCIAL FACTORS AND HEALTH DISPARITIES ACCESS ISSUES ARE IMPORTANT SORT OF FOCUS FOR THIS PARTICULAR BRANCH OF THE GENE DISCOVERY EFFORT WHICH FOCUSED ON AFRICAN-AMERICAN FAMILIES AFFECTED BY AUTISM. JUST TO SAY ONE WORD IN THE ACCUMULATING DATASET TO DATE ABOUT THAT ISSUE IS THIS SLIDE SORT OF DEPICTS THE SITUATION FOR AFRICAN-AMERICAN CHILDREN AFFECTED BY AUTISM IN THE U.S. THEIR PARENTS CITE FIRST CONCERNS AROUND 21 MONTHS, SHARE THOSE CONCERNS ON AVERAGE AT 27 MONTHS, THE CHILDREN ON AVERAGE ARE NOT DIAGNOSED UNTIL 54 MONTHS, AND DON'T GET ANY DEVELOPMENTAL SERVICES ON AVERAGE UNTIL 39 MONTHS DESPITE AT LEAST, AT LEAST FOR NOW, MOST OF THEM HAVING HEALTH INSURANCE COVERAGE. SO THIS PARTICULAR STUDY AND PROJECT ADDRESSES THE ISSUES THAT MIGHT RELATE TO GENETIC DISPARITIES AND DIFFERENCES IN BOTH CAUSATION AND HOW THAT RELATES TO THESE SORT OF SOCIAL DIFFERENCES. FLEUR BACK WHEN THE STUDY WAS STARTED, ABOUT 2% OF THE INDIVIDUALS IN THE AUTISM GENE -- MULTIPLEX FAMILY COLLECTION WERE AFRICAN-AMERICAN IN CONTRAST TO A REPRESENTATIVE SAMPLE WHICH WOULD HAVE BEEN 13%. THIS IS TRUE, THIS SAMPLING, BIAS HAS BEEN REFERRED TO BEFORE IN SOME OF THE OTHER TALKS THIS MORNING, IN TERMS OF THE NECESSITY OF INCREASING THE PROPORTION OF AFRICAN-AMERICANS AND OTHER MINORITY SUBJECTS INCLUDED IN THESE PARTICULAR STUDIES. AND OVER THE LAST 10 YEARS NOW, BEGINNING IN THE FIRST FIVE-YEAR SEGMENT WITH A SORT OF TEST OF PRINCIPLE, THEN FOLLOWED IN THE NEXT FIVE-YEAR SEGMENT WITH A VERY ERNEST MULTISITE EFFORT TO ENROLL PREFERENTIALLY AND ENRICH THE EXISTING GENE BANK WITH AFRICAN-AMERICAN FAMILIES, WE'RE GETTING CLOSE TO ABOUT 10% WITHIN -- AND HOPE TO CONTINUE TO IMPROVE OP THAT. THE ISSUES ABOUT POPULATION DIFFERENCES IN OUTCOMES AS EFFECTIVE ANCESTRAL ORIGINS HAS BEEN BROUGHT UP ALREADY. THIS IS SORT OF A LISTING OF SOME PRECEDENCE IN THE LITERATURE FOR INHERITED HEMOGLOBIN DISORDERS, CANCERS, VARIOUS DIFFERENCES IN PHARMACOGENETICS, AND DRIFT AND POPULATION ISOLATION FOR RARE GENETIC DISORDERS THAT REALLY EMPHASIZE THE NEED TO INCLUDE AND EXPAND STUDIES TO NON-EUROPEAN POPULATIONS IN OUR PSYCHIATRIC GENETIC RESEARCH. THIS IS FROM THE THOUSAND GENOMES PROJECT SHOWING THAT THE VARIANT SITES PER GENOME IS SUBSTANTIALLY ELEVATED IN FAMILIES WITH AFRICAN ANCESTRAL ORIGIN, AND THAT CREATES ALL KINDS OF OPPORTUNITIES FOR GENETIC MISDIAGNOSIS IN THE SETTING OF CLINICAL GENOMIC CHARACTERIZATION WHEN RARE VARIANTS ARE IDENTIFIED. SO GETTING SOME STANDARDS AND GETTING A BACK BEING FOR HOW WE UNDERSTAND RISK FOR AUTISM AND RELATING THAT TO ANCESTRY IS AN IMPORTANT OBJECTIVE. SO AS THIS PARTICULAR PROJECT HAS STARTED TO ACCRUE ITS SAMPLE, THE PLAN FOR DATA ANALYSIS, AND AGAIN, WE'RE I THINK IN GENERAL ABOUT HALFWAY THROUGH THE TOTAL EFFORT TO BALANCE REPRESENTATION OF AFRICAN-AMERICANS WITHIN THE U.S. REGISTRIES, INCLUDE THE PLAN TO INCLUDE EXOME ANALYSIS, SNP ANALYSIS, USING VARIOUS DIMENSIONS OF PHENOTYPE, WHICH I'LL TALK ABOUT IN A MINUTE, BOTH ANCESTRAL CALLING AND CNVs, COMBINING NOT ONLY THE ANALYSIS BASED ON AD MIXTURE BUT ALSO SPECIFIC ASSOCIATION CONTROLLING FOR ANCESTRAL VAIR VARIATION AND I'LL SHOW THAT IN A SECOND, AS WELL AS CNV ANALYSIS, SEARCH FOR RARE VARIANTS AND POLYGENIC RISK SCORES. AS A SORT OF SET OF PRINCIPLES OF WHY IT'S IMPORTANT TO DO THIS, IN THE EARLY COLLECTION, WE WERE ABLE TO LOOK AT THE INITIAL FIRST JUST SHORT OF 100 AFRICAN-AMERICAN FAMILIES ENROLLED IN THE STUDY AND JUST TESTED THE HYPOTHESIS OF WHETHER POLYGENIC RISK SCORES THAT WOULD DIFFERENTIATE AFFECTED FROM UNAFFECTED INDIVIDUALS IN A SAMPLE OF EUROPEAN FAMILIES AFFECTED WITH AUTISM WOULD SIMILARLY RESULT IN THAT DIFFERENTIATION AM THE AFRICAN-AMERICAN FAMILIES AND AS YOU CAN SEE, WHEN APPLYING THAT AGAIN TO A SMALL SAMPLE, BUT THE SORT OF SOASHES BREAKS DOWN WHEN SHIFTING OVER TO AN AFRICAN-AMERICAN POPULATION. AGAIN, UNDERSCORING THE NEED TO REALLY SPECIFY THESE ANALYSIS WITH ANCESTRAL ORIGIN IN MIND. IN THINKING ABOUT THE ANALYSIS, ONE OF THE ISSUES IS THERE HAVE BEEN DISCOVERED NUMEROUS RECOMBINATION HOT SPOTS IN THE AFRICAN GENOME, AND SO INCLUDING NOT ONLY THE ISSUE OF WHAT IS THE PROPORTION OF AFRICAN ANCESTRY REPRESENTED BY A SAMPLE, BUT ALSO USING ANALYSES THAT CONDITION ON ANCESTRAL BACKGROUND BASED ON SOME OF THESE RECOMBINATION EVENTS THAT RESULT IN LOCAL VARIATIONS IN WHAT ONE CAN INFER FROM AN ASSOCIATION, PUTTING THOSE TWO ASPECTS TOGETHER RATHER THAN EITHER ONE ALONE ACTUALLY INCREASES STATISTICAL POWER DPENTD DEPENDING ON -- BUT COMBINING ADD MIXTURE AND FREQUENCY SIGNAL, WE'RE HOPING THAT THAT'S GOING TO INCREASE STATISTICAL POWER TO IDENTIFY ASSOCIATION. SO IN THE SAMPLE THAT WE'VE ACCRUED TO DATE, AGAIN, HERE THAT SLIDE IS SHOWING FROM THE THOUSAND GENOMES, THIS EXCESS IN NOVEL VARIANTS, HERE IN THE DATA WE'VE COLLECTED FROM AFRICAN-AMERICAN FAMILIES AFFECTED BY AUTISM IN THE PROJECT SO FAR, THIS IS THE PLOTTED PROPORTION OF GENOME THAT'S AFRICAN, THERE'S STRONG CORRELATIONS BETWEEN THESE DEMONSTRATED BOTH IN LYMPHOBLASTIC -- LYMPHOBLASTOID CELL LINES AND WHOLE BLOOD. SIMILARLY OVERALL NUMBER OF CM Vs -- AFRICAN AND ANCESTRAL ORIGIN AND THEN PROPORTION OF AFRICAN ANCESTRY AND RARE MISSENSE VARIANTS, AGAIN SHOWING THESE IMPORTANT RELATIONSHIPS AND SORT OF DEMONSTRATING THIS PRINCIPLE. THOSE VARIATIONS OCCUR NOT ONLY ALL OVER THE GENOME BUT IN SPECIFIC REGIONS THAT ARE GENES OF REGIONS THAT RIN ARE INTOLERANT TO THE SPECIFIC KINDS OF VARIANTS, THESE ARE TWO SAMPLES OF INTERNALLY TRANSMITTED VARIANTS WITHIN THE PROGRAM OF CHILDREN COAFFECTED BY AUTISM, INHERITING THESE PARTICULAR TYPES OF EVENTS IN AUTISM FAMILIES. AMONG THE FAMILIES, THE RATE FOR SYNDROMIC CNVs WE'VE IDENTIFIED IS APPROXIMATELY EQUAL TO WHAT'S BEEN IDENTIFIED IN OTHER CNV ANALYSES OF AUTISM-RELATED STUDIES. THE LAST POINT I WANT TO MAKE, EVEN AS WE ARE WORKING ON THESE SORT OF GENETIC ANALYSES WITHIN THE SAMPLE, WE ARE USING THIS GENE DISCOVERY NETWORK TO TRY TOO ADVANCE PHENOMIC ADVANCE, TO THE PHENOTYPING. ONE OF THE THINGS WE'VE LEARNED FROM LOOKING EARLIER THAN WHEN AUTISM DEVELOP, IN OTHER WORDS, LOOKING A EITHER DEVELOPMENTAL PHENOTYPES OR PARENTAL PHENOTYPES, IS THAT THERE ARE HIGHLY INHERITED TRAITS THAT DON'T NECESSARILY RELATE TO ONE ANOTHER AND THAT ACTUALLY PREDICT WITH STRONG R SQUARES THE DEVELOPMENT OR THE OCCURRENCE OF AUTISM. GETTING BACK TO THE EXAMPLE TALKED ABOUT BEFORE ABOUT BMI, REFERRING TO THE ISSUE THAT RISK FOR LUNG CANCER CAN BE INDEXED BY RISK FOR SMOKING, THAT THESE PARTICULAR TYPES OF HIGHLY INHERITED PHENOTYPES GIVE US A NEW OPPORTUNITY TO SORT OF CONTROL FOR THESE DEVELOPMENTAL VARIATIONS AND LOOK AT WHAT THE EFFECTS OF GENETIC LIABILITY FACTORS ARE. SO IT'S POSSIBLE TO CONTROL FOR THESE DEVELOPMENTAL FACTORS, AND IF POLYGENIC VARIANCE IS ASSOCIATED WITH THESE TRAITS BUT UNMEASURED AMONG CONTROLS, THEN THIS CAN CONTRIBUTE TO MISSING HERITABILITY AND POLYGENIC RISK THAT'S ASSOCIATED WITH CO-INHERITED LIABILITIES. SO IMPAIRMENT IN MOTOR COORDINATION IS SOMETHING THAT'S CAN BE HIGHLY PLEA OWE TROPIC, IT HAS ITS OWN UNIQUE CONTRIBUTION IN PREDICTING AUTISM THAT'S UNRELATED TO PREDICTION FROM A VERY INTERESTING EARLY INFANCY PHENOTYPE AND VARIATION IN SOCIAL VISUAL ENGAGEMENT, WHICH IN TURN IS ALSO UNRELATED TO INDEX OF LIABILITY FOR AUTISM THAT CAN BE DERIVED JUST FROM SUBCLINICAL AUTISTIC TRAITS OF PARENTS. SO THINKING ABOUT HOW THESE GENETIC LIABILITIES MIGHT LIE ALONG THESE AXIS OF EARLY DEVELOPMENT AND CONTRIBUTE TO AUTISM IS AN IMPORTANT ASPECT OF OUR PROGRAM. SO IN THE NEXT PHASE, WE WANT TO EXPAND THE GENE DISCOVERY EFFORT, WE WANT TO ENSURE THAT EARLY ASD DIAGNOSIS IS AVAILABLE TO THE CHILDREN TO COUNTERACT THIS ISSUE OF LATE DIAGNOSIS AND SEE WHETHER WE CAN MOVE THE NEEDLE ON THE OUTCOMES OF THOSE CHILDREN TO PHENOTYPE IN RESPECT TO INHERITED DEVELOPMENTAL RISK AND TO SECURE ACCESS TO APPROPRIATE COMMUNITY INTERVENTIONS FOR THE CHILDREN AND FAMILIES AND TEST WHETHER POLYGENIC RISK INDESEES WILL BE RELATED TO QUANTITATIVE DEVELOPMENTAL PHENOTYPES. SO I WANT TO RECOGNIZE THE WHOLE TEAM ASSOCIATED WITH THIS, THIS IS INDEED TEAM SCIENCE, AND AGAIN, TRYING TO GROUND THIS UNDER AN APPROACH THAT INCLUDES BOTH PHENOMICS AND GENETICS AMONG MANY, MANY COLLEAGUES AND MANY PEOPLE WHO HAVE WORKED VERY HARD IN THE FIELD TO RECRUIT THE FAMILIES AND MAKE THEM COMFORTABLE WITH AUTISM GENETIC RESEARCH AND COUNTLESS NUMBERS OF RESEARCH ASSISTANTS THAT HAVE DONE THAT KIND CH WORK. SO THANK YOU. [APPLAUSE] >> THANK YOU, JOHN. SO TIME IS AN UNFOR GIVING MASTER AND UNFORTUNATELY WE WON'T HAVE TIME TO TAKE QUESTIONS NOW. HOWEVER, WE WILL HAVE A BREAK AFTER THE NEXT SESSION. A SHORT BREAK. SO AS UP TO NOW, WE WERE LOOKING AT THE GERMLINE DISCOVERING RISK VARIANTS FOR DISEASE, BUT TO REALLY UNDERSTAND THE FUNCTIONAL CONSEQUENCES OF THE CAUSAL RELATIONSHIP BETWEEN A VARIANT THAT CAUSES DISEASE AND THE PHENOTYPE, NEED TO UNDERSTAND PERTURBATION THAT HAPPENED IN THE MOLECULAR MACHINERY IN THE TISSUE OF INTEREST, NAMELY THE BRAIN, SO HENCE WE ARE ENTERING NOW THE INTEGRATIVE GENOMICS ERA AND SESSION. I'M GOING TO START OUT WITH THE COMMON -- WE HAVE TO TRY SOMETHING OUT HERE BECAUSE PAMELA SKLAR IS SPEAKING NOW, SHE'S FROM MOUNT SINAI. SHE COULD NOT BE HERE TODAY SO PAMELA, IF YOU HEAR THIS, SEE THIS, WE MISS YOU. BUT SHE HAD SENT SLIDES AND ANDY HAS BROUGHT AN AUDIO RECORDING. SO WE'LL TRY TO FIGURE THIS OUT. s >> I'M DELIGHTED FOR THIS OPPORTUNITY TO PRESENT THE WORK BY AUDIO RECORDING. EVEN THOUGH MY HEALTH KEEPS ME FROM BEING THERE IN PERSON. SINCE THEIR INCEPTION, I'VE HAD LEADERSHIP ROLES IN MANY OF THE TEAM SCIENCE EFFORTS YOU'RE HEARING ABOUT AT THIS WORKSHOP. I'VE BEEN A MEMBER OF THE PSYCHIATRIC GENOMICS CONSORTIUM MANAGEMENT GROUP FROM THE BEGINNING, AND ALSO HAVE CO-LED THE BIPOLAR -- ITS BIPOLAR DISORDER WORKING GROUP. MORE RECENTLY, MY LEADERSHIP ROLES HAVE BROADENED TO INCLUDE THE COMMON MIND CONSORTIUM, OR CMC FOR SHORT, PSYCHENCODE, AND THE BIPOLAR STEERING CONSORTIUM. IN MOST OF THESE GROUPS, I'VE PERFORMED A NUMBER OF MANAGEMENT TASKS, BUT ALSO HAVE LED EXPERIMENTAL DESIGN AND DATA GENERATION AND DATA ANALYSIS. AS AN ASIDE FOR THE JUNIOR SCIENTISTS, FOR MOST OF MY CAREER, I'VE BEEN PURSUING THESE KINDS OF TEAM SCIENCE EFFORTS BECAUSE OF THEIR INHERENT NATURE TO PRODUCE INTERESTING AND UNEXPECTED DISCOVERIES. I FIND IT STRAIGHTFORWARD TO ASSURE YOU THAT YOU CAN ACHIEVE SUCCESS IN THIS ARENA. THERE ARE KEY POINTS FROM THE TALKS WE HEARD THIS MORNING OR FROM THE TALKS THAT YOU HEARD THIS MORNING THAT BEAR REITERATING. ESPECIALLY SINCE MY TALK WILL FOCUS ON SCHIZOPHRENIA. THERE ARE MANY GENES THAT INFLUENCE ITS GENETIC RISK AND CURRENTLY THESE ARE FOUND ACROSS ALL VARIANT CLASSES FROM RARE TO COMMON, FROM COMMON TO RARE. WHILE THE RARE CODING VARIANTS MAY HAVE SOME INCREASED VALUE IN INTERPRETABILITY ANDESE OF BIOLOGIC FOLLOW-UP, THEY CAN ALSO INTRODUCE POTENTIAL BIASES TOWARDS GENES AND PATHWAYS FOUND IN INTELLECTUAL DISABILITIES AND AUTISM. THIS MAY OR MAY NOT BE DESIRABLE. NEEDLESS TO SAY, TEAMS WORKING TOGETHER ARE EQUALLY IMPORTANT TO COMMON AND RARE VARIANT STUDIES TO MODEL ADEQUATE SAMPLE SIZES OR TO HAVE ADEQUATE POWER TO MODEL SAMPLE SIZES AND TO DEVELOP POWERFUL ANALYTIC TOOLS FOR FUNCTIONAL NETWORK ANALYSIS AT ALL LEVELS. SO THIS MEANS NOT JUST GENE EXPRESSION NETWORKS, BUT NETWORKS THAT EXIST ACROSS AND BETWEEN THE EPIGENOME AND THE GENOM AND BETWEEN A VARIETY OF LAYERS OF THE KINDS OF FUNCTIONAL INFORMATION THAT WE'RE NOW GIPPING TO COLLECT. IT HAS BEEN DIFFICULT TO NOMINATE SINGLE GENES FROM GWAS FOR SEVERAL MOLECULAR REASONS. A PROPORTION OF THESE VARIANTS LIE WITHIN INTERGENIC REGIONS AND ALSO THE ASSOCIATED LOCUS MAY HAVE MULTIPLE GENES, ALL OF WHICH ARE IMPOSSIBLE TO DISTINGUISH ON THE BASIS OF LD FROM EACH OTHER. DESPITE THIS, THERE ARE ALREADY GENES FROM GWAS LOCI THAT HAVE BEEN IDENTIFIED WITH STRONG EVIDENCE FOR BEING THE LIKELY CAUSAL DOREEN GENE IN A PARTICULAR RACE ASSOCIATED REGION. THE FIRST WAS IDENTIFIED IN A COLLABORATION BETWEEN THE STANLEY CENTER, ABOUT A YEAR AGO, WHERE THEY DEMONSTRATED COMPELLING STATISTICAL CORRELATION BETWEEN COMPLEMENT FACTOR C4A SCHIZOPHRENIA ALLELES AND GENE -- EXPRESSION. ANOTHER WAS FOUND IN OUR OWN CMC WORK WHERE WE RECENTLY REPORTED THAT OF AMONG THE 108 RISK LOCI, THERE WERE ABOUT 20% WHERE A SINGLE GENE OR COUPLE OF GENES COULD BE SEEN AS THE LIKELY CAUSAL GENE, LIKELY ACCOUNTING FOR THE SCHIZOPHRENIA RISK SIGNATURE OBSERVED. WE FOUND AS FOR THE C4A ALLELES FIVE LOCI WHERE ONLY A SINGLE GENE WAS IMPLICATED. IN ONE OF THOSE FIVE LOCI, THE GENE THAT WE FOUND WAS ADDITIONALLY FOUND FROM A COUPLE ZEBRAFISH ASSAYS THAT WE HAD TESTED. THESE ARE TWO SIMPLE EXAMPLES FOR CONNECTING SCHIZOPHRENIA LIST I.R.S. BEING ALLELES TO THE GENES THEY REGULATE, THAT IS THROUGH GENE EXPRESSION AND ALSO OTHER TYPES OF FUNCTIONAL CHANGES IN THE EPIGENOME. IT SHOULD BE RATHER STRAIGHTFORWARD OVER THE NEXT FEW YEARS OR MAYBE OVER THE NEXT DECADE BY ENLARGING THE SAMPLE SIZES FOR GENOTYPING SEQUENCINGS AND FUNCTIONAL ASSAYS UNTIL WE HAVE A REASONABLY REFINED VIEW OF THE OVERALL ARCHITECTURE. CURRENTLY IT'S LIKE WE HAVE A JIGSAW PUZZLE WITH ABOUT HALF THE PIECES, BUT NOTELY WITHOUT THE PICTURE ON THE BOX FOR REFERENCE. WHILE THERE ARE MANY TRADITIONAL REDUCTIONIST STRATEGIES LIKE THE ONE I'VE JUST KE SCRIBED, THEY CAN AND SHOULD BE EXPLORED, WE ALSO NEED TO MAKE USE OF NEWER SYSTEMS-BASED APPROACHES TO REDEFINE SCHIZOPHRENIA AT THE MOLECULAR LEVEL. HOWEVER, TO DO THIS SUCCESSFULLY, MUCH LARGER DATASETS ARE NEEDED FOR THE SOPHISTICATED NETWORK MODELING POWER TO REALLY MAKE SENSE OF THE GENETIC OBSERVATIONS. AGAIN, FOR ME, THIS ECHOS WHAT WE LEARNED THIS MORNING IN THAT ANALYSIS TOOLS AND NEW GENETIC MODELS TO USE GWAS ANALYSIS REQUIRE MUCH LARGER DATA THAN WE EVER THOUGHT. HOPEFULLY IT CAN BE SWIFTER PROJECT THIS TIME BUT THERE' NO WAY AROUND GENERATING AND EXPLORING ANALYSES IN LARGE SCALE DATA. THESE MODELS HOPEFULLY WILL BE ABLE TO DISCOVER MECHANISMS OF DISEASE AND SPECIFIC GENES THAT WE CAN MODULATE WITH DRUGS. ONE, YOU KNOW, IMPORTANT GOAL WOULD BE TO TRANSFORM A DISEASE NETWORK AND HERE I USE THE TERM NETWORK VERY BROADLY, INCLUDING BOTH GENE EXPRESSION, EPIGENOMIC NETWORKS AND THEIR INTEGRATED -- AND NETWORKS WHERE THEY'RE INTEGRATED BACK TO THE HEALTHY STATE. FOR EXAMPLE, WE MIGHT BE ABLE TO DEFINE A STABLE NETWORK SUCH THAT IF WE SUPPRESS A PARTICULAR NODE, ONLY ONE OR A FEW NEIGHBORING NODES MIGHT BE AFFECTED. HOWEVER, IF WE COULD FIND AND UNDERSTAND HOW TO MODULATE NODES THAT AFFECT LARGE SWATHS OF A NETWORK, MAKING IT LOOK MUCH MORE LIKE A CONTROL THAN A SCHIZOPHRENIA CASE, THAT WOULD CLEARLY BE ONE WE WOULD WANT TO GO AFTER IN TERMS OF DRUG DISCOVERY. SO BY FINDING THESE SCHIZOPHRENIA-RELATED NETWORKS WE CAN ULTIMATELY GET TO THE MECHANISM OF DISEASE AND WHAT GENES ARE BEST TO BE MANIPULATED AS DRUG TARGETS FOR AN INDIVIDUAL. IN 2012, I BEGAN WORKING ON A PUBLIC-PRIVATE PARTNERSHIP THAT WE TITLED THE COMMON MIND CONSORTIUM. THIS WAS ALONG WITH MY COLLEAGUES AT MOUNT SINAI, AT SAGE BIONETWORKS, AT THE UNIVERSITY OF PITTSBURGH, AT THE UNIVERSITY OF PANCREATIC ADENOCARCINOMA, AT ROCHE, AND ALSO ENCOLONELLED AND STIMULATED BY THE NIMH PROGRAM OFFICIALS, THOMAS AND GITA. WE WERE JOINED BY BARBARA LIPSKA IN 2014 IN HER GROUP AT THE NIMH. OUR GOAL WAS TO BUILD A LARGE DATA GENERATING PROJECT SO THAT WE COULD ENHANCE AND FILL IN THE GAPS IN THE AVAILABLE HUMAN BRAIN DATA SO THAT WE WOULD HAVE THE MAPS AVAILABLE THAT WE WOULD NEED TO BE ABLE TO UNDERSTAND MECHANISTICALLY HOW SCHIZOPHRENIA RISK ALLELES PRODUCE THEIR EASKTS. IN LATE 2012, I BEGAN WORKING ON A PUBLIC-PRIVATE PARTNERSHIP. >> WE ARE A SHORT REPEATING PART. SORRY. >> WE WERE ALSO STIMULATED BY THE NIMH PROGRAM OFFICERS. AND OUR GOAL, WE WERE JOINED BY BARBARA LIPSKA IN 2014. OUR GOAL WAS TO BUILD THE SORTS OF LARGE DATA GENERATING PROJECTS THAT WE KNEW WE WERE GOING TO NEED TO BE ABLE TO FILL THE GAPS IN THE AVAILABLE HUMAN BRAIN DATA SO THAT WE WOULD BE ABLE TO UNDERSTAND AND MAKE SENSE OF THE MANY AND DIVERSE GENETIC FINDINGS THAT WE WERE UP AGAINST. FOR EXAMPLE, AT THAT TIME, THERE WERE FEWER THAN A HUNDRED BRAIN SAMPLES AVAILABLE THROUGH GTEX AND VERY LITTLE CHIP SEQ DATA FROM THE BRAIN. IN TOTAL IN THE CMC PROJECT, WE NOW HAVE 1400 BRAIN SAMPLES, ON WHICH THE VARIETY OF ANALYSES ON& THE SLIDE WILL BE DONE. WE WILL HAVE BASELINE GENETIC DATA, WHICH IS GENOTYPING AND RNA SEQUENCING. WHILE WE WILL NOT HAVE DATA TYPES ON ALL SAMPLE, WE WILL HAVE A SUBGROUP OF BRANDS WHERE WE WILL HAVE THE LARGEST NUMBER THAT WE CAN POSSIBLY AFFORD. CURRENTLY WE HAVE 1600 SAMPLES THAT HAVE RNA SEQUENCING. WE HAVE ABOUT 700 WITH CHIP SEQ AND THAT NUMBER WILL ENLARGE QUITE A BIT OVER THE NEXT YEAR. WE HAVE WHOLE GENOME SEQUENCING ON 650, AND THERE ARE NEWER PROJECTS BEING LED BY ACTIVE CMC MEMBERS AND SOME OF THESE INCLUDING THE ANALYSES OF RNAs AND ISOFORM IDENTIFICATION, RNA EDITING USING MULTIPLEX MICRO FLUIDICS -- LEAD THE CMC SCHIZOPHRENIA FOCUS MOUNT SINAI PORTION OF THE PSYCHENCODE PROJECT. YOU'LL HEAR MUCH MORE ABOUT THAT IN THE TALKS THAT FOLLOW. WE ARE PART OFOUR PART OF THE PROMG IS PRO JECT IS FOCUSED ON SCHIZOPHRENIA, AND THEY'RE WORKING ON A NUMBER OF ASPECTS OF THE EPIGENOME AND PARTICULARLY SOME SINGLE CELL OR CELL TYPE-SPECIFIC ANALYSES. THERE ARE INTERESTING SECONDARY GENETICS ANALYSIS THAT HAVE BEEN LED BY SALLY AND ELI INCLUDING ATITIONAL EQTL WORK, AS WELL AS SOME ANALYSIS OF PREDICT SCAN THAT ARE UNDER REVIEW, AND I'LL TALK A LITTLE ABOUT SOME OF THESE TOWARDS THE END OF THE TALK. FINALLY, THERE'S ALSO A CMC PROJECT EXPLORING SOMATIC MOSAC ICM. THIS IS BEING LED AT MOUNT SINAI. AND IN COLLABORATION WITH CHRIS WALSH AND PETER PARK AT HARVARD. WE HAVE A WORK SPACE AT SYNAPSE FOR AGGREGATING, DESCRIBING AND SHARING ALL OF OUR DATA. AND THIS IS ALL PUBLICLY RELEASED THROUGH THE SAME WEBSITE. WE TRY TO RELEASE THE DATA AS QUICKLY AS POSSIBLE. SO GENE EXPRESSION DATA WAS RELEASED ABOUT A YEAR BEFORE WE PUBLISHED IT AND WE TRY TO RELEASE AS SOON AS WE'RE CONTENTED THAT BASIC QC CAN ENSURE A GOOD DATASET. WE'VE DEVELOPED A TYPICAL ANALYTIC PIPELINE FOR OUR RNA SEQUENCING, THE DETAILS NOT AT ALL IMPORTANT, EXCEPT WE SPENT CONSIDERABLE TIME -- KNOWN AND UNKNOWN COVAIR YATS USING CIRCUIT VARIABLE ANALYSIS. OUR PIPELINE IS ALL AVAILABLE IN GIT HUB AS WELL AS THE DATA, YOU'LL NOTICE WHAT WE HAVE IS MOSTLY AN OLDER POPULATION, MOSTLY CAUCASIAN AND ABOUT 60% MALE. THE FIRST IMPORTANT STEP IS CREATING AND EVALUATING THE QUALITY OF THE EXPRESSION QUANTITATIVE LOCI. WE FIND AS OTHERS HAVE THAT THE EQTLs WERE ENRICHED IN -- REGIONS AS WELL AS IN -- ENHANCERS FROM THE BRAIN BUT NOT FROM OTHER TISSUES. WHEN WE COMPARE TO THE ROAD MAP TO EPIGENOMICS PROJECT. IN FACT, WE FOUND THAT EVEN USING LARGE DATABASES THAT ARE AVAILABLE, THAT THERE ARE STILL A SIGNIFICANT NUMBER OF CMCE GENES THAT WERE NOT FOUND IN LARGE BLOOD DATABASES, BLOOD DATABASES THAT WERE FOUR OR FIVE FOLD LARGER, AND THUS WE CONCLUDE THAT TO STUDY THE BRAIN, IT REALLY IS NECESSARY TO STUDY THE BRAIN RATHER THAN LARGER, EASIER TO OBTAIN TISSUES. AS WE NOTED BEFORE, A GWAS LOCUS TYPICALLY CONTAINS MANY GENES AND SNPS THEREIN ARE OFTEN HIGHLY CORRELATED VIA LINK WAJ DISEQUILIBRIUM. SO WE LEVERAGED THE SAME -- RISK VARIANTS TO EXPRESSION OF SPECIFIC GENES. THE WAY WE DID THIS WAS USING A BAYESIAN APPROACH THAT PRIORITIZES CONSISTENCY BETWEEN DISEASE ASSOCIATION AND EQTL SIGNATURES IN GWAS. THUS WE USED THE RISK SNPS FROM THE 108 PGC LOCI AND INTERCEPTED THEM USING SHERLOCK AND SUBSEQUENTLY FOLLOWED UP IN PHENOTYPIC ASSAYS A SMALL SUBSET OF GENES THAT WERE IDENTIFIED THIS WAY. WE HAVE MORE RECENTLY EXTENDED THIS USING -- SCAN IN A LARGER SAMPLE THAT WILL BE DESCRIBED IN A MINUTE OR TWO. WE FOUND 37 GWAS LOCI MET OUR STRINGENT -- CORRECTED CRITERIA FOR SIGNIFICANCE FOR ONE OR MORE GENES IN EACH OF THE LOCI. OR SAID ANOTHER WAY, APPROXIMATELY 5% OF THE GWAS SIGNIFICANT LOCI HAD ONE OR MORE GENES WHOSE GENE EXPRESSION CHANGES COULD BE EXPLAINED BY THE RISK ALLELE AND ITS RELATED SNPS. AMONG THE LOCI, THERE WAS ONLY A SINGLE GENE IMPLICATED. OF PARTICULAR INTEREST IS FURIN BECAUSE IT PROCESSES -- INCLUDING A KEY MOLECULE IN BRAIN DEVELOPMENT WHOSE DOWN MODULATION HAS BEEN HYPOTHESIZED AS RELATED TO SCHIZOPHRENIA, AND BDNF AND FURIN ARE UPREGULATED IN ASTROCYTES IN RELATIONSHIP TO STRESS. WHAT I NEED TO TELL YOU IS THAT GENE EXPRESS IS SUBTLELY DISRUPTED ACROSS MANY JEAN GENES WITH GENES, THUS WE WITH AN AVERAGE FOLD CHANGE OF ABOUT 1.09, THUS WE WOULD NEED 11,000 SAMPLES TO DETECT A SIGNIFICANT GENE EXPRESSION DIFFERENCE FOR FURIN. I WOULD FOCUS THIS IS THE WRONG -- TO HAVE CREDIBILITY AT AVOIDING FALSE POSITIVES SUCH AS THOSE WE EXPECT -- AND IN FACT IN DETAILED SIMULATIONS BURIED IN THE SUPPLEMENT OF THE PAPER, WE SHOWED THAT SMALLER SAMPLE SIZES, SAY THOSE IN THE RANGE OF 25 OR SO, LIKE THOSE THAT ARE OFTEN IN TYPICAL GENE EXPRESSION STUDIES WERE MUCH MORE LIKELY TO RESULT IN GENE EXPRESSION CHANGES THAT WERE SIGNIFICANT OR OVER TWOFOLD. AS A RESULT, WE FIRST STARTED BY CHOOSING TO MODEL NEUROANATOMICAL CHANGES AT THE PHENOTYPIC LEVEL. WHILE WE COULD HAVE CHOSEN ANY OF THE 20 GENES THAT WE THOUGHT WERE CAUSAL TO FOLLOW UP ON, WE FOCUSED ON THE FIVE WHERE THERE WAS ONLY ONE GENE IMPLICATED BY SHERLOCK AND AT THAT PARTICULAR GWAS LOCUS. WE DID THIS TO AVOID ANY ISSUES OF CONDITIONAL EQTLs SUCH AS THOSE OBSERVED IN THE -- REGION IN THE -- REPORT. WE LOOKED FOR A MODEL SYSTEM WHERE WE COULD HAVE A SIMPLE READOUT THAT SOMETHING WAS CHANGING IN THE BRAIN, AND WE CHOSE TO TEST WHETHER SUPPRESSION OR OVEREXPRESSION OF THE -- GENES WITHIN EACH OF THE FIVE SCHIZOPHRENIA LOCI COULD IDENTIFY KEY PROTEINS THAT REGULATE BRAIN DEVELOPMENT AS MEASURED BY HEAD SIZE IN ZEBRAFISH. WE FOUND FOUR OF THE FIVE GENES CAUSED A DECREASE IN HEAD SIZE. THIS WAS CORROBORATED BY CHANGES IN CELL PROLIFERATION AND/OR APOPTOSIS. TO DRILL INTO THE MODEL WITH THE LARGEST IMPACT ON HEAD SIZE, WE TESTED FURIN KNOCK DOWN IN HUMAN NEUROGENERAL TORE CELLS NEUROGENERAL NEUROGENETOR CELLS, AFTER INTUBATION FOR 48 HOURS IN THREE CONSTRUCTS PER CONDITION, WE FOUND MARKED DECREASE IN NEWER I'D OUTGROWTH. TO INCREASE OUR POWER TO DETICKET A GENOTYPE DRIVEN GENE EXPRESSION CHANGE, IN COLLABORATION WITH NANCY AND -- WE'VE APPLIED THEIR TRANSCRIPTOMIC IMPUTATION& PACKAGE CALLED PREET DICTIONARY SCAN, THE PREDICT SCAN. PREDICT SCAN CAN IDENTIFY AND CODIFY THESE RELATIONSHIPS IN MATCH PANELS OF INDIVIDUALS AND THEN CAN IMPUTE THE PORTION OF THE TRANSCRIPTOME ATTRIBUTABLE TO GENETICS INTO LARGE SCALE GENOTYPE-ONLY DATASETS, SUCH AS A CASE CONTROL GWAS COHORT, THUS ENABLING INVESTIGATION OF DISEASES SO THE DISEASE ASSOCIATED GENE EXPRESSION CHANGES. THIS ALLOWS US TO STUDY GENES OF MODEST EFFECT SIZES LIKELY REPRESENTING RISK FOR SCHIZOPHRENIA. SO IN THIS CASE, WE USED OUR CMC GENE EXPRESSION DATASET, CURRENTLY THE LARGEST EQTL REFERENCE PATTERN FOR THE DORSO LATERAL PFC. WE CREATED A SET OF PREDICTORS AND DEMONSTRATED THEIR UTILITY. THESE PREDICTERS WERE THEN APPLIED TO THE 35,000 SCHIZOPHRENIC CASES AND 45,000 MATCH CONTROLS FROM THE PGC SCHIZOPHRENIA STUDY AND MORE RECENT -- GENE EXPRESSION LEVELS ARE NOW SIGNIFICANTLY DIFFERENT BETWEEN THE CASES AND CONTROLS FOR 65 GENES. OF THESE, 19 ARE NOVEL, BUT AMONG THE NOVEL ONES, THE MOST WERE NEARLY SIGNIFICANT. SO NOW WHAT WE HAVE IS A SET OF GENES THAT ARE LIKELY CAUSALLY RELATED TO CHANGES IN GENE EXPRESSION. WE WERE ABLE TO ALSO INVESTIGATE 21 OF THOSE THAT HAD BEEN IN OUR PRIOR SHERLOCK ANALYSES, AND IN FACT, THE VAST MAJORITY OF THEM WERE IN THE SAME DIRECTION OF EFFECT WITH A P VALUE OF 10 TO THE MINUS 55, HIGHLY UNLIKELY TO HAVE OCCURRED BY CHANCE. IN THE LAST MINUTE OR TWO, I'LL BEGIN TALKING ABOUT OUR PSYCHENCODE PROJECT AND SOME OF OUR EFFORTS MAPPING PROMOTERS AND ENHANCERS. THEY'VE DEVELOPED A ROBUST METHOD FOR USING NEURONAL NUCLEI SPECIFIC ANTIBODY AND FACT SHORTING TO PRODUCE CELL TYPE-SPECIFIC FRACTION FOR CHIP SEQ. OUR FIRST PRIORITY IS TO MAKE A MAP OF SEVERAL HISTONE MARKS IN TWO BRAIN REGIONS IN 17 INDIVIDUALS IN A NEURONAL AND NON-NEURONAL FRACTION. FITTING A LINEAR MIXED MODEL, WE FIND THAT USING CELL TYPE SUBJECT REGION AND SEX,, THAT CELL TYPE AND SUBJECT HAVE THE STRONGEST SOURCES OF VARIANTS IN THE NUMBER OF PEAKS THAT WE OBTAINED. FUNCTIONAL ENRICHMENTS IN THE NEURONAL PEAKS FOR BOTH H3K H3K4ME3 -- ACETYLATION WERE IN RECEPTORS, GA PA RECEPTORS, POTASSIUM CHANNELS, AND OTHER MARKERS THAT WE WOULD HAVE EXPECTED FOR THE NON-NEURONAL MARKS, WE FIND THAT THERE APPEAR TO BE MULTIPLE PATHWAYS, IMPLYING THERE ARE MORE CELL TYPES IN THIS. FINALLY WE USED LD SCORE REGRESSION TO LOOK AT ENRICHMENT -- WHAT WE FOUND WAS THAT THE STRONGEST ASSOCIATION WAS FOR SCHIZOPHRENIA-ASSOCIATED LOCI, WITH NEURONAL HISTONE MARKS, FOLLOWED BY EDUCATION IN YEAR, AND ALMOST EXCLUSIVELY THOSE ARE FOUND IN THE NEURONAL HISTONE LANDSCAPE. FOR NON-BRAIN RELATED TRAITS, THERE WAS LITTLE ENRICHMENT AT ALL. IN CONCLUSION, I'D LIKE TO SAY THAT COMMON MIND IS BUILDING A LARGE RESOURCE OF MOLECULAR DATA FROM HUMAN BRAINS, AWFUL WHICH WILL BE, ALL OF WHICH WILL BE PUBLICLY AVAILABLE. SCHIZOPHRENIA RELATED GENE EXPRESSION CHANGES CAN BE MEASURED IN WHOLE HOMO GENATE TISSUES, TRANSCRIPTOMIC IMPUTATION IS NOW INCREASING OUR POWER TO MOVE THAT FORWARD IN THE GWAS REALM. I'D LIKE TO THANK MY MANY COLLEAGUES, BOTH AT SINAI AND IN THE VARIOUS CONSORTIA, FOR THE WONDERFUL WORK WE'VE BEEN ABLE TO DO TOGETHER. >> THANK YOU, PAMELA. SO I AND THE FIELD, WE OWE A BIG CREDIT TO PAMELA. WHAT MANY OF YOU WON'T KNOW, SHE'S ONE OF THE FEW PEOPLE THAT ARE MY GO-TO FOLKS. LIKE STEVE I GO TO FOR MONEY. PAMELA, I GO TO -- >> [INAUDIBLE] [LAUGHTER] I CALLED PAMELA AND I ASK HER, YOU KNOW, WOULD YOU BE INTERESTED IN DOING SUCH AND SUCH, AND SHE ALWAYS SAYS YES. SHE'S ALWAYS GRACIOUS AND SHE ALWAYS WORKS WONDERFULLY TOGETHER IN THIS LARGE CONSORTIA. SO THANK YOU, PAMELA, FOR THAT AS WELL. OUR NEXT SPEAKER IS PANAGIOTIS ROUSSOS FROM MOWBT SINAI MOUNT SINAI. >> THANKS SO MUCH. I WOULD LIKE TO THANK THE PROGRAM COMMITTEE FORT INVITATION, AND I WOULD LIKE TO PRESENT SOME OF THE EARLIER ANALYSIS AND DATA GENERATION WORKING AT SINAI IN TERMS OF FURTHER UNDERSTANDING THE NEUROEPIGENOME. I WOULD LIKE TO GIVE YOU SOME EARLY RESULTS. SO WE'VE RECENTLY FINISHED -- TARGETING TWO DIFFERENT POPULATION OF CELLS, NEURONAL, NON-NEURONAL, AND WE ACTUALLY DID THAT ACROSS 14 DIFFERENT BRAIN REGIONS. SO WHAT WE HAVE LEARNED FROM THIS ANALYSIS? SO FIRST OF ALL, BILL DOING MDS PLOT BUT ALSO BY TESTING UNSUPERVISED -- USING THE -- SIMILARITY, WE ACTUALLY SEE THERE IS A CLEAR SEPARATION OF NEWER ONLY VERSUS -- YOU CAN SEE GREEN BOX IN BOTH FIGURES. AT THE SAME TIME, JUST BY LOOKING IN THE NEURONAL PORTION, YOU CAN SEE THAT ACROSS THE SECOND AXIS, WE HAVE FURTHER SEPARATION THAT MARKS THE DIFFERENT KIND OF REGIONAL DISTRIBUTION OF THE EPIGENOME, MEANING THAT AS YOU CAN SEE HERE, WE HAVE A -- INCLUDING THE NEOCORTEX, AMYGDALA, THE SECOND CLUSTER INCLUDES -- JUST TO REMIND YOU THERE WAS A STUDY BACK IN 2011 THIS, IS FROM THE BRAIN -- USING THE EPIGENOME, THEY USE -- WE ACTUALLY FIND A SIMILAR KIND OF SPECIFICITY BY JUST -- THE NEUROEPEPIGENOME. WE HAVE NEOCORTEX -- SEPARATION WHEN YOU LOOK AT THE THALAMUS. FURTHERMORE IF WE START LOOKING WITHIN THE NEW CORTICAL REGION, YOU CAN SEE WE HAVE A CLEAR SEPARATION AND THIS AGAIN IS SOMETHING, PREVIOUS EFFORT -- OVERALL OUR DATA INDICATING THE EPIGENOME IS VERY SPECIFIC, BUT AT THE SAME TIME, THIS DIFFERENCE IS MOSTLY COMING FROM NEURONAL AND NOT NON-NEURONAL PORTION. NOW STUDYING USING THIS KIND OF SPECIFIC ANNOTATION AND TRY TO SEE HOW WE CAN USE IT TO FURTHER UNDERSTAND THE COMMON VARIANTS, SO HERE WE USE -- REGRESSION AND ACTUALLY RUN ACROSS 14 DIFFERENT BRAIN REGION AND TWO CELL TYPES, WHAT YOU CAN SEE IS YOU SEE A VERY CLEAR ENRICHMENT ONLY FOR THE NEURONAL EPIGENOME AS PAMELA DESCRIBED EARLIER, BOTH SCHIZOPHRENIA AND BIPOLAR, MINIMAL -- WHEN YOU START LOOKING AT NON-NEURONAL POPULATION. AT THE SAME TIME, WE SEE THAT IN SCHIZOPHRENIA BUT BIPOLAR, WE SEE MAXIMUM ENRICHMENT IS ACTUALLY COMING FROM REGIONS THAT USUALLY WE ACTUALLY IGNORE AND FOCUS MORE ON THE CORTICAL REGIONS. JUST TO ALSO SHOW YOU HERE, THIS RECENT STUDY, THEY ACTUALLY USE FUNCTIONAL KAY TA COMING DATA COMING FROM CNA R S.E.C., THE SEQ, COMING FROM HIPPOCAMPUS , AS WELL AS -- FINDING NEURONS WHICH IS SIMILAR TO WHAT WE SEE IN OUR STUDY. NOW AT THE SAME TIME, WHEN WE HAVE THIS KIND OF CROW MA TIN ACCESSIBILITY DATA WE CAN DO MORE SORT OF -- DIGITAL FOOTPRINT THEN WE'RE GOING TO ASSIGN SPECIFIC FACTOR TO SPECIFIC GENES AND OVERALL, WE CAN ACTUALLY AGGREGATE AND FIND RELATIONSHIP OF THE -- TO THE SPECIFIC GENE FOR EACH BRAIN REGION AND CELL TIME AND NETWORK APPROACHES. SO UP WITH EXAMPLE THAT WE HAVE DONE HERE IS ACTUALLY WE USE THIS KIND OF DIVERGENCE OF THE EFFECT OF THE TRANSCRIPTION FACTOR ON PROTEIN CODING CHANGE AND WE CAN ACTUALLY RUN FROM THE ONE THAT SEEMS TO BE MOSTLY NEURONAL ALL THE WAY DOWN TO MOSTLY NON-NEURONAL. DOING THIS PATHWAY ANALYSIS, WE ACTUALLY CONFIRMED THE TOP WHAT WE CALL NEURONAL GENES, THEY PARTICIPATE IN NEURONAL FUNCTION, AND VICE VERSA, WHEN WE LOOK AT A NON-NEURONAL, WE SEE PATHWAYS RELATED TO GLIA AND ALL THE WAY TO THE OTHER SIDE OF FUNCTION. NOW WE DO EXACTLY THE SAWM THING BY LOOKING AT THE CORTICAL AND SUBCORTICAL REGIONS. WE GET KIND OF A SIMILAR FIGURE. IF YOU SEE HERE -- A GENE THAT'S WELL-KNOWN TO BE EXPRESSED MOSTLY IN THE COURT RAL CORTICAL REGIONS. MORE IMPORTANTLY IN THE SUBCORTICAL REGIONS, WE SEE THE DOAN DOAP MEAN RECEPTOR -- OR THIS GENE THAT HAS HIGHLIGHT OVER HERE, WHICH IS THE PPP1 -- MANY PEOPLE -- SO FOR THAT SPECIFIC GENE, YOU CAN ACTUALLY SEE HERE THE CHROMATIN ACCESSIBILITY PROFILE, AND AS YOU SEE THE PROMOTER REGION AND -- ENHANCER IS ONLY PRESENT IN THE STRIATUM OF THE NEURONS. THAT'S A UNIQUE OPPORTUNITY BECAUSE NOW WE ACTUALLY USE HUMAN NON-CODING -- WE CAN ACTUALLY USE THIS ELEMENT TO BE ABLE TO DRIVE EXPRESSION IN A VERY SPECIFIC SPECIAL WAY IN MOUSE BRAINS. SO WHAT WE DID, WE ACTUALLY -- IT CAN DRIVE EXPRESSION USING MOUSE BRAIN, AND AS YOU CAN SEE, THE GFP, THE HUMAN ENHANCER ELEMENT LOCALIZED PERFECTLY -- NEURONS AND WHAT THIS SPECIFIC -- IT'S NOT CONSERVED IN THE MOUSE BRAIN. AS YOU CAN SEE ALSO HERE, WE CAN SEE PROJECTION FOR THE STRIATUM AND THE -- IN THE MOUSE, SO THIS IS ANOTHER INDICATION OF HOW WE CAN USE THAT KIND OF DATA, THAT WE CAN ACTUALLY LIBERATE IN DIFFERENT ANIMAL MODELS. OKAY. SO MOVING FORWARD, I WOULD LIKE TO SHOW YOU SOME KIND OF EARLY PRELIMINARY RESULTS FROM SOME OF THE DIFFERENT DIRECTIONS WE'RE GETTING IN MY LAB AND ALSO COLLABORATION WITH OTHER INVESTIGATOR. THIS COLLABORATION AT MOUNT SINAI, DOING EPIGENOME STUDIES IN NEURONAL AND NON-NEURONAL IS SOMETHING THAT MOVE FORWARD IN THE FIELD BUT AT THE SAME TIME, WE DON'T CAPTURE THE CELLULAR HETEROGENEITY THAT EXISTS. SO WHAT WE HAVE DONE OVER THE LAST FEW YEARS, WE ACTUALLY SPENT TIME, WE ACTUALLY DEVELOP NEW -- PROTOCOL SO WE CAN SORT AND ISOLATE FROM FROZEN TISSUE DIFFERENT NEURONAL SUBPOPULATION INCLUDING -- GLUTAMATE AS WELL AS NON-NEURONAL -- AT THE SAME TIME, ESTABLISHED A PIPELINE THAT WHETHER WE GET THE WORM AUTOPSY, WE HAVE A NEW BRAIN THAT IS GOING, SO BEFORE WE -- AT THE SAME TIME USE CELL SURFACE -- AT THE SAME TIME, WITH THE NEW TECHNOLOGY, WE'RE NOT IN A POSITION THAT WE CAN STUDY THE TRANSCRIPT OF AN EPIGENOME JUST BY USING A VERY LIMITED INPUT AAMOUNT SO FOR ALL THESE USE BE CHIP SEQ, RNA SEQ, WE NOW HAVE PROTOCOLS WE CAN USE A FEW THOUSAND OF CELLS THAT WE CAN ACTUALLY USE AND MARKET -- DO THE DIFFERENT FUNCTIONAL STUDIES. SO THIS IS AN EXAMPLE, THIS IS -- HERE YOU'RE LOOKING AT THE GABA FRACTION, YOU CAN SEE RNA SEQ, YOU ONLY SEE SIGNAL COMING ONLY FROM THE GABA NER JIK PORTION, INDICATING CELL TYPES -- AT THE SAME TIME, THIS AGAIN PRELIMINARY ZONES, BUT USING THIS ANNOTATION ACROSS THREE DIFFERENT BRAIN REGIONS, TRYING TO SEE WHAT IS THE ENRICHMENT, AGAIN WE SEE FOR SCHIZOPHRENIA, WE DON'T SEE MUCH FOR NON-NEURONAL PORTION, THEN WE SEE LITTLE ENRICHMENT WHEN YOU SAW THE -- THE VISUAL CORTEX. BIPOLAR WE START SEEING -- SIGNIFICANCE ONLY FOR GABA NEUROEPIGENOME, AS A CONTROL WE'RE USING CROHN'S DISEASE, THROAFER ENRICHMENT FOR -- AS WELL AS MICRO GLIA. ANOTHER DIRECTION WE'RE GETTING IS WE TRY DOING THE FOX, IT'S GRADE BUT STILL YOU ARE LIMITED IN THE REPERTOIRE OF -- DIFFERENT NUCLEI, SO HERE WE HAVE SINGLE CELL ATAC SEQ, FORM ATAC S.E.C. REACTION, HERE WE ACTUALLY MAKE 50% COMING FROM MOUSE, ANOTHER 50% FROM HUMAN, AND AS YOU CAN SEE HERE, THE RATES THAT WE SEQUENCE EXCLUSIVELY TO THE HUMAN GENOME OR MOUSE GENOME INDICATING WE HAVE DONE A VERY GOOD JOB IN SEPARATING THE DIFFERENT NUCLEI. KEEP IN MIND THIS IS A PILOT STUDY, WE HAVE ONLY 2,000 NUCLEI AND BECAUSE IT'S MUCH MORE SPARSE THAP THE SINGLE RNA SEQ, WE ESTIMATE THAT WE WILL NEED ABOUT 20,000 NUCLEI TO BE ABLE TOE DO A MORE ROBUST KEANU DE NOVO TAXONOMY, BUT WHAT IS INTERESTING IS THE WAY WE MAKE THE LIBRARIES WITH KNOW WHICH SPECIFIC NUCLEI, AS YOU CAN SEE HERE, ALL THE NON-NEURONAL AND ALL THE NEURONAL I HAVE SEPARATED PERFECTLY USING THIS KIND OF [INAUDIBLE]. MAYBELY SKIP THIS FOR THE SAKE OF TIME. ALSO PART OF TEAM SCIENCE, ANOTHER KIND OF -- THAT WOULD HE HAVE ACTUALLY DEVELOPED IN MY LAB IS WE ARE PART OF LABS CONSORTIA, INCLUDING THE ONE FOE KUTION ON ALZHEIMER'S DISEASE, WHAT WE DO WITH THIS CONSORTIA, WE ARE GENERATING -- ON TOP OF THAT, MY LAB HAS ACTUALLY GENERATED CELL TYPE-SPECIFIC REGIONAL -- WE USE DIFFERENT BRAIN REGIONS TO ACTUALLY GENERATE -- FILING FOR THE OPEN CROW MA TIN, AND WE ACTUALLY HAVE IN A VERY AMBITIOUS GOAL TO GENERATE -- LIBRARY, BUT OVERALL AT LEAST FOR THE ALZHEIMER'S AND THE BIPOLAR, WE HAVE TO -- MORE THAN 50% OF THOSE LIBRARIES. AS WE HAVE BEEN DOING, THEY'RE GOING TO BE PUBLICLY AVAILABLE FOR EVERY OTHER PERSON TO USE FOR THEIR OWN STUDY. FINALLY, WE WOULD LIKE TO USE THIS KIND OF BIG DATA TO DO IN THE SAME WAY WE DO FOR EQTL, AS YOU KNOW THE METHODS THAT WE CURRENTLY HAVE TO INTEGRATE GWAS WITH MULTIPLE DIFFERENT TRAIT IS WIDE LIMITED SO THIS IS A COLLABORATION WITH BOB AT UCLA, SO WE DEVELOPED A NEW METHOD WHICH ACTUALLY ALLOWS TO GET SIGNAL COMING FROM EXPRESS, EPIGENOME AND OTHER TRAITS. HERE -- ANDREW PROVIDED THE DATA. AS YOU CAN SEE THIS, IS AN EXAMPLE FOR FURIN, WHAT PAMELA DESCRIBED EARLIER. HERE WHAT WE FOUND IS THE SAME WAY THAT PAMELA WAS DESCRIBING, WE FOUND LOABLIZATION OF GWAS, NOW WE HAVE A THIRD TRADE, WHICH IS THE EPIGENOME. I WOULD LIKE TO THANK MANY, MANY COLLABORATION, INCLUDING PEOPLE FROM THE PGC -- FROM MY LAB I WOULD LIKE TO THANK JOHN FULLARD WHO HAS BEEN DOING MOST OF THE DATA GENERAL JEN RANGE OF MOTION, MADZ HAUBERG, AND CLAUDIA IS THE PERSON WHO DEVELOPED -- [APPLAUSE] >> THANK YOU. SO BEFORE WE GO ON BREAK WE HAVE ANOTHER PROGRAM CHANGE. AGAIN, TIME MASTER IS NOT COOPERATING. THIS CASE IT'S -- SO ANYWAY, FROM UCLA, DAN, PLEASE. >> I REALLY HAVE TO THANK THOMAS THOMAS. AND THE REST OF THE ORGANIZERS FOR GIVING ME THE OPPORTUNITY TO TALK HERE. THIS HAS REALLY BEEN AN EXTRAORDINARY MEETING SO FAR THAT REALLY HIGHLIGHTS A LOT OF THE PROGRESS IN PSYCHIATRIC GENETICS. I'M JUST GOING TO TALK ABOUT A SMALL KIND OF TIP OF THE ICEBERG THAT WE'VE BEEN WORKING ON. THERE ARE A LOT OF PEOPLE WHO ARE WORKING ON THIS AND IT'S REALLY, AGAIN, PART OF TEAM SCIENCE AND LARGE CONSORTIUM, AND IT'S THE COLLABORATIVE EFFORTS OF MANY THAT HAVE LED TO THESE DATA. THIS IS MY CONFLICT OF INTEREST STATEMENT. I'M NOT GOING TO BE TALKING ABOUT THESE ISSUES TODAY. HERE'S THE SIMPLE PARADIGM. FIND LOCI AND FIND THE GENES THAT UNCOVER MECHANISMS. OF COURSE AS WE'VE HEARD TODAY, NONE OF THESE STEPS ARE WITHOUT SUBSTANTIAL CHALLENGES. YOU HEARD REALLY SPECTACULAR TALKS ABOUT FINDING LOCI IN GENES. I'M GOING TO SPEAK A LITTLE BIT TO THAT BUT MOSTLY TALK ABOUT HOW WE CAN USE NETWORKS TO UNCOVER MECHANISMS. WITH REGARD TO FINDING GENES, ONE OF THE CRITICAL THINGS IS THAT SINCE MOST -- NOFT OF THE COMMON VAIR YATS THAT LIES IN NON-CODING REGION, A CRITICAL THING IS TO ASSIGN FUNCTIONALITY. YOU'VE BEE BEGUN TO HEAR SOME OF THE TALKS -- I'LL DISCUSS SOME OF OUR WORK THAT I THINK MIC ODEN VAN SHARED BRIEFLY. SECONDLY, I'LL DISCUSS THE WORK USING GENE NETWORKS TO IDENTIFY POINDPOINT OF CONVERGENCE AND VERY SIMILAR TO WHAT JEREMY SHOWED, WE REALLY FIND MOLECULAR CONVERGENCE OF RISK GENES BUT THEN WE CAN USE THESE NETWORK BASED MODELS TO CONDUCT DIFFERENT LEVELS OF ANALYSES. GOING FROM GENOMIC VARIANT TO GENE IS EXTREMELY COMPLEX AND A RAPIDLY EVOLVING AREA, BUT IT'S DEPENDENT ON TEAM SCIENCE EFFORTS SUCH AS PSYCHENCODE ROAD MAP. YOU CAN SEE HERE THE FUNCTIONAL GENOMIC ANNOTATIONS THAT ONE WANLTS TO IDENTIFY THE NON-CODING FUNCTIONAL REGIONS FROM HISTONE MODIFICATIONS, METHYLATION, TRANSCRIPTION FACTOR BINDING SITES, CONSERVATION, EQTL OR SPLICING EQTL. ONE OF THE BUY BIOLOGIES -- PAM HIGHLIGHTED THIS VERY WELL AT THE CELL TYPE AND TISSUE LEVEL, AND WE NEED TO HAVE IN THE CELL AT THE STAGES OF INTEREST. AND THE TISSUE OF INTEREST. THE SECOND POINT IS THAT GENE REGULATION IS NOT A LINEAR TWO-DIMENSIONAL ISSUE. THE TARGET GENES ARE ACTED UPON OFTEN BY CHROMATIN REGIONS THAT ARE HUNDREDS OF KILOBASES AWAY. WE HAVE TO UNDERSTAND THAT TOWPPED STAND GENE REGULATION. SO WE USE HIGH C, WHICH IS A BIOCHEMICAL TECHNIQUE THAT PULLS DOWN INTERACTING REGIONS OF THE GENOME NO MATTER HOW FAR THEY ARE APART. BECAUSE GENE REGULATION OCCURS VIA CHROMOSOME FOLDING THAT BRINGS -- INTO CONTACT WITH THEIR TARGET GENES IN A TISSUE-SPECIFIC MATTER, WE CAN USE THIS HIGH C TECHNIQUE -- NOT EXCLUSIVELY IN DEVELOPING CEREBRAL CORTEX DIVIDED INTO THE CORTICAL PLATE AND THE -- CONTAINING PRO GENERAL TORS. WE CAN USE IT TO IDENTIFY CHROMATIN STRUCTURE AT MULTIPLE DIMENSIONS AND DOMAINS AND SCALES, ONE OF THE CRITICAL SCALES IS THE GENE LOOP, WHICH IS SUPPOSEDLY WHAT IS BRINGING ENHANCERS AND PROMOTERS INTO CONTACT. SO WE'RE IDENTIFYING LONG-RANGE CHROMOSOME INTERACTIONS, SPECIFYING GENE EXPRESSION REGULATION AND BRAIN THROUGHOUT THE ENTIRE GENOME AT 10K RESOLUTION. AND THIS WORK WAS DONE BY -- WHO'S IN THE AUDIENCE, WHO WAS A POSTDOC IN THE LAB. SO ONE OF THE APPLICATIONS OF THIS WAS TO HUMAN-SPECIFIC INVOLVED ENHANCERS. THE OTHER WAS TO SHOW HOW ONE COULD USE THIS KIND OF INFORMATION TO IDENTIFY THE LIKELY MECHANISM, AGAIN IT'S JUST A METHOD TO PRIORITIZE AND IDENTIFY LIKELY CANDIDATES. SO WHAT'S SHOWN HERE IS A GWAS SNP SHOWING THE HIGH C INTERACTION MAP LOG 10P VALUE ON THE Y AXIS SHOWING A VERY SUBSTANTIAL SIGNIFICANT INTERACTION IN THE FOX G1 PROMOTER WHICH IS OVER 700KB AWAY. WE DON'T JUST RELY ON THAT, WE CAN -- WE TOOK PRIMARY HUMAN NEUROPROGENATORS BETWEEN 15 AND 18 WEEKS, CUT OUT THE PUTATIVE ENHANCER REGION AND SHOW THAT IT AFFECTS FOX G1 EXPRESSION AND NOT ANY OF THE CLOSER GENES, WHICH IS OVER 700KB AWAY. AND IN FACT, ONLY HALF OF THE GWAS LOCI ARE ACTING ON THE CLOAEST GENE AS IDENTIFIED BY THESE HIGH C INTERACTION, AND IN FACT, THAT'S VERY CONSISTENT WITH KNOWN EQTL DATA THAT SHOWS QUITE A FEW IN FACT PREDOMINANTLY ABOUT 50% DISTAL INTERACTION, AND MANY ARE ACTING MORE THAN 700 BASE PAIRS AWAY, AND THIS IS NOT IN LINE WITH LINKAGE DISEQUILIBRIUM OR OTHER GENETIC MEASURES SORK THIS KIND OF FUNCTIONAL CHROMATIN MAP IS ACTUALLY CRITICAL FOR ANNOTATING GENOME SEQUENCING AND GENETIC ASSOCIATION. HERE WE'RE ABLE TO IDENTIFY DOZENS OF NEW RISK GENES THAT WERE YOU IS PORTED BY EQTL AND GENE EXPRESSION FROM THE COMMON MIND CONSORTIUM IMPLICATING NEW PATHWAYS FROM THE ORIGINAL INCLUDING CORTICAL NEUROGENESIS AND -- CHOLINE RECEPTOR ACTIVITY. SO THAT'S JUST ONE I WOULD SAY SNIPPET OF HOW ONE COULD USE THIS DATA TO BEGIN TO GAIN FUNCTIONAL INSIGHT INTO HOW GENOME WIDE, HOW A GWAS OR A RARE VARIANT, HOW A PARTICULAR VARIANT ASSOCIATED WITH DISEASE IN A NON-CODING REGION MIGHT BE ACTING AND ASSIGN IT TO GENES OF INTEREST. BUT OF COURSE AS WE DO THAT, WE'RE JUST GOING TO BE IDENTIFYING MORE AND MORE GENES. YOU CAN SEE ONE OF THE GREAT THINGS THAT'S HAPPENED, WE'RE SITTING HERE WITH REALLY A BASKET THAT'S ALMOST -- IT'S NOT OVERFLOWING, IT'S ONLY ABOUT A QUARTER FULL, BUT IT'S MAYBE MORE FULL THAN WE THOUGHT IT MIGHT HAVE BEEN 10 YEARS AGO. SO WE HAVE QUITE A FEW GENES THAT HAVE BEEN IDENTIFIED. SO NOW THE ISSUE IS, HOW DO WE IDENTIFY POINTS OF CONVERGENCE? ARE WE GOING TO HAVE TO DEAL WITH HUNDREDS OF DIFFERENT ACTUAL GENETICALLY DEFINED DISORDERS OR IS THERE POINTS OF CONVERGENCE AND PATHWAYS THAT WE CAN USE TO INFORM DISEASE MECHANISMS? AND I'LL BRIEFLY DISCUSS TWO EXAMPLES OF THIS MOLECULAR CONVERGENCE OF AUTISM RISK GENES. JEREMY WILLSLEY GAVE A GREAT PRESENTATION OF THAT, THEN I'LL TALK ABOUT CONNECTING DIFFERENT LEVELS OF ANALYSES. SO THIS IS ALL BASED ON THE OBVIOUS. EXTREME HETEROGENEITY AT THE POPULATION AND INDIVIDUAL LEVELS. GENES DON'T ACT IN ISOLATION, THEY'RE COMPONENTS OF BIOLOGICAL SIGNALING PATHWAYS AND COMPLEX NEURAL NETWORKS AS WELL. CAN WE IDENTIFY AREAS OF CONVERGENCE WITH THESE LARGE DIMENSIONAL DATA SETS RELATED TO KEY ASPECTS OF BIOLOGICAL FUNCTION OR DISEASE PATHOPHYSIOLOGY. AND WHEN WE IDENTIFY THESE PATTERNS, CAN WE USE THEM TO INFORM UNDERSTANDING OF DISEASE MECHANISMS OR TREATMENT DEVELOPMENT. I USE THE TERM SYSTEMS BIOLOGY IN QUOTES, EVERYBODY HAS A DIFFERENT DEFINITION OF WHAT THIS IS BUT IN ESSENCE, RATHER THAN JUST LOOKING AT ONE LAYER OF DATA, WE'RE TRYING TO -- IN ALMOST A SIMULTANEOUS WAY BUT AT THE VERY LEAST IN A SERIAL WAY LOOK AT ALL OF THESE DIFFERENT DATA FROM SEQUENCE TO COPY NUMBER TO EPIGENETIC TO GENE EXPRESSION DATA TO PROTEOMICS DATA AND INTERGREAT INTEGRATE THAT WITH PHENOPOINT DAY ATTACHMENT YOU CAN SEE WITH ALL THESE RESOURCES WHERE THIS IS NEARLY ACHIEVABLE. SO AGAIN, WE HAVE TO COPE WITH THIS ISSUE OF A SYSTEMS LEVEL HIERARCHY. I SHOW THIS DIAGRAM THAT'S VERY SIMILAR TO THE ONE THAT THOMAS SHOWED EARLIER TODAY, WHERE WE HAVE TO GO ACROSS ALL OF THESE MOLECULAR HIERARCHIES FROM GENETIC VARIATION TO MESSAGE RNA AND ALL THE EPIGENETIC CHANGES THAT LEAD TO EXPRESSION, TO PROTEIN, TO SNAPS, TO CIRCUIT, TO BEHAVIOR. AND THAT IS A DAUNTING TASK. EVEN CONNECTING TWO OR THREE LEVELS WILL BE AN ENORMOUS ADVANCE. ONE OF THE THINGS THAT HELPS US, GENES DON'T ACT INDEPENDENTLY. I'M NOT GOING TO GO THROUGH THE NETWORK THEORY, A COEXPRESS NETWORK, WE CAN SEE GENES GOING UP AND DOWN TOGETHER, WE HAVE THESE DIFFERENT GROUPS OR MODULES IN THE NETWORK, THEN WE CAN ANNOTATE THOSE MODULES USING VARIOUS METHODS. WE TOOK PUBLICLY AVAILABLE DATA FROM MULTIPLE MICRO ARRAY STUDIES OF HUMAN BRAIN CELL TYPE WITHIN THAT TISSUE, AND THAT WE DO A DISSECTION THAT THE MODULES WHEN IDENTIFIED ACTUALLY CORE SPOPPEDDED TO THE MAIN CELLULAR COMPONENTS OF THE TISSUE. AND WITHIN THE MODULES, ONE CAN IDENTIFY THE HUB GENES OR CENTRAL GENES THAT MIGHT BE DRIVERS. SO WE'RE USING THIS METHOD THROUGHOUT. SO NOW WHAT I'M GOING TO TALK ABOUT IS HOW WE USE THIS TO QUERY NEUROTYPICAL BRAINNET WORKS. JEREMY'S WORK WITH MATT STATE SHOWED THAT THEY TOOK AUTISM GENES AND CREATED A SEEDED NETWORK. WHAT WE'RE DOING IS TAKING THE BRAIN SPAN DATA AND CREATING A NEUROTYPICAL BRAINNET WORK. SO WE TAKE THE PAIR WISE FUNCTIONAL RELATIONSHIPS, IN THIS CASE, THERE'S SOMETHING CALLED TOPLOGIC OVERLAP, ASSOCIATION BETWEEN THE GENES GOING UP AND DOWN TOGETHER. WE IDENTIFY NETWORKS AND THE GENES THAT ARE COEXPRESSED TOGETHER THAT CHANGE TOGETHER TO FIND MODULES, THESE SHARE FUNCTION AND WE COLOR CODE THEM FOR FUN AND FOR VISUALIZATION. SO WE GO TO TENS OF MODULES THAT RELATE TO CELL TYPES OR ASPECTS OF INDIVIDUAL BIOLOGICAL FUNCTION. AND THE CRITICAL QUESTIONS TO ASK ARE, WHERE ARE THE GENES OF INTEREST? WHERE DO AUTISM OR SCHIZOPHRENIA OR BIPOLAR GENES LIE WITHIN THIS NORMAL BRAIN DEVELOPMENTAL NETWORK, UNDERSTAND MODULE BIOLOGY, IF THEY LIE IN A PARTICULAR MODULE LIKE SHOWN HERE IN THE YELLOW MODULE, WHAT DOES THAT MEAN? IN SPACE, TIME, PHYSICAL INTERACTION, AS JEREMY MENTIONED. SO WE CAN GO ACROSS TIME, BRAIN REGIONS AND CELL TYPES. SO THE QUESTION WE ASKED WITH THAT WAS DO AUTISM RISK -- CONVERGE ON ANY BIOLOGICAL PROCESS AND THE ANSWER IS, OF COURSE, YES, JEREMY SHOWED THAT, AND USING THIS KIND OF ORTHOGONAL METHOD, WE ESSENTIALLY FIND VERY SIMILAR DATA. SO IF WE TAKE DE NOVO RIS FACTORS THAT ARE JUST OCCURRING IN THE SIMPLEX COLLECTION, WE SEE ENRICHMENT IN TWO MODULES, M2 AND M3, SHOWING PEAK DURING CORTICAL NEUROGENESIS. THIS IS EVEN EARLIER, THIS IS TELLING US THAT THESE GENES ARE ENRICHED AND LIKELY ACTIVE DURING NEURONAL CORTICO GENESIS. WE ALSO DON'T JUST RELY ON THAT ONE DATASET, WE REPLICATE AND SHOW THESE MODULES EXIST IN OTHER DATASETS, AND WE ALSO SHOW THAT IF WE HAVE AN MRNA MODULE, THAT SHOULD ALSO BE ENRICHED IN PROTEIN-PROTEIN INTERACTIONS SH SHOWN HERE. SO BOTH OF THESE MODULES ARE ENRICHED AND VERY INTERESTINGLY, THIS M3 IS HIGHLY ENRICHED IN THE -- COMPLEX SHOWN TO BE INVOLVED IN NEUROGENESIS. IT HAS SIX OUT OF THE 28 MEMBERS OF THE BAV COMPLEX IN THIS MODULE WERE HIT BY DE NOVO MUTATIONS. IF WE LOOK, THESE INHERITED RISK FACTORS WERE MUCH LESS CONVINCING, WE STILL USE THE DATA THAT'S OUT THERE, INHERITED RISK IS NOT ACTING VERY EARLY, IT'S ACTING A LITTLE LATER BUT STILL BEGINNING WITH MID FETAL DEVELOPMENT. WHAT'S INTERESTING IS THOSE GENES ARE ALSO DOWNREGULATED IN AUTISM CEREBRAL CORTEX. SO THESE NEURONAL SYNAPTIC AND SIGNALING GENES -- AUTISM DOWN REGULATED BRAIN. SO JUST TO SUMMARIZE, AUTISM GENES MAP ON TO SPECIFIC PATHWAYS AND DEVELOPMENTAL NETWORKS AND INCLUDING GLUE TA MATTER JIK NEURONS. SO WE CAN SEE TRANSCRIPTIONAL REGULATION EARLY AND SYNAPTIC DEVELOPMENT SUBSEQUENTLY. THIS MAKES REALLY TOTAL SENSE IN SOME WAYS BECAUSE WE HAVE THESE RARE DE NOVO MUTATIONS OCCURRING IN THESE VERY TOP TIER TRANSCRIPTION FACTOR AND CHROMATIN REGULATORS, IF WE LOOK AT THIS FIGURE FROM THE ENCODE PROJECT SHOWING THREE LEVELS OF REGULATION, WE CAN SEE THAT THESE WOULD BE EXPECTED TO HAVE THE LARGEST EFFECTS, THAT THE GENETIC ARCHITECTURE IS PARALLELLING A MOLECULAR ARCHITECTURE, IT HIGHLIGHTS THE NEED FOR REFINED UNDERSTANDING OF THE REGULATORY NETWORKS, THAT ARE GOING ON WITH SIEK PSYCHENCODE AND OTHERS. NETWORK MODULES REPLICATE AND HAVE GOOD PREDICTIVE POWER. THESE THREE STU DIES IN THE MIDDLE ARE NOT TOTALLY INDEPENDENT, THE STUDY AT THE TOP SUN PUN IS UNPUBLISHED AND THIS IS IN THE AGREE MULTIPLEX FAMILY COHORT. ONE CAN SEE IN ALL OF THESE AUTISM OR THIS DEVELOPMENTAL DELAY CONSORTIUM M2 AND M3 STILL COME OUT AND ARE HIGHLY ENRICHED. THIS IS HIGHLY SIGNIFICANT ENRICHMENT IN RISK GENES. ALSO BECAUSE WE'RE LOOKING AT A FAMILY SAMPLE AND WE HAVE SOME POWER TO IDENTIFY INHERITED VARIATION, WE SEE THIS PREVIOUS DEVELOPMENTAL M16 USING IDENTIFYING GENES USING THE TA DA TEST. SO WE CAN USE THESE MODULES NOT ONLY TO BEGIN TO SET A FRAMEWORK FOR STUDYING TECHNOLOGY, BUT TO IMPROVE POWER IN STATISTICAL ANALYSES OF OUR GENETIC ASSOCIATION STUDIES. SO IN THE LAST COUPLE MINUTES, I'LL TALK ABOUT HOW WE CAN USE THIS NOWP TO BUILD MECHANISTIC MODELS TO CONNECT DIFFERENT LEVELS OF ANALYSIS. WE ALL KNOW AND EVERYBODY'S USE USING IPS DERIVED NEURONS, LOUISE AND JASON STEIN IN THE LAB, WHEN THEY WERE IN THE LAB, DECIDED THAT IT WOULD BE CRITICAL TO UNDERSTAND HOW THESE IN VITRO MODELS RECAPITULATE HUMAN IN VIVO CORTICAL DEVELOPMENT. WE HAVE THE BRAIN SPAN RESOURCE, WE CAN USE TRANSCRIPT TOMORROWS TO -- AS WELL AS OUR NETWORK ANALYSIS TO DETERMINE WHICH BIOLOGICAL PROCESSES ARE PRESERVED. REMARKABLY IN THESE 3D CULTURES THAT CAME ALONG LATER, WE SAW FAR MORE RECAPITULATION OF IP VEE VO NEURAL DEVELOPMENT THAN IN ANY OF THE 2D CULTURE, AS WELL AS FAR MORE MATURATION. IN FACT THE CHROMATIN AND TRANSCRIPTION FACTOR ENRICHED MODULES THAT ARE ASD RISK GENES ARE ALSO HIGHLY PRESERVED IN THESE CULTURES, VALIDATING WE CAN LIKELY USE THEM TO STUDY THE BIOLOGY. SO WE'RE DEVELOPING REGULATORY MAPS OF HUMAN CORTICAL DEVELOPMENT THROUGH BRAIN SPAN LAPS LIKE -- OURS AND OTHERS AND MARK GER STEEN GERSTEIN, GENE EXPRESSION DATA, USING SINGLE CELL SEQUENCING, FROM IN VIVO CEREBRAL CORTEX AT MID GESTATION FETAL BRAIN USING THE FANTASTIC DROP SEQ METHOD YOU'RE GOING TO HEAR ABOUT LATER. WHEN WE TAKE THE ORGAN OID DATA FROM IP VITRO, WE SEE A RECAPITULATION OF ALL THE MAJOR CELL TYPES USING SINGLE CELL SEQUENCING. SO IT'S TELLING US THAT THESE HUMAN -- ACHIEVE MATURATION COMPARABLE WITH POSTNATAL HUMAN BRAIN DEVELOPMENT WHEN CULTURED LONG ENOUGH. SO IN THE END, WE THINK WE CAN USE THESE AND OTHER METHODS TO HAVE A COMPREHENSIVE EXPERIMENTAL APPROACH TO DEFINE CONVERGENCE. ONE CAN TAKE HIGHLY PENETRANT RISK LOCI, MANY OF THEM CULTURE IN PARALLEL, AND ASK DO WE HAVE CELLULAR, MOLECULAR, ANATOMIC OR PHYSIOLOGIC CONVERGENCE? WE CAN ALSO ADD AND DO THIS ON THE CONTEXT OF THE TOP 10% OF POLYGENIC COMMON GENETIC RISK SCORE AND BOTTOM TO ASK HOW COMMON VARIATION MEDIATES THE EFFECT -- AT WHAT PHENOTYPES AT A MOLECULAR AND CELLULAR LEVEL IT'S ACTUALLY MEDIATING IN VITRO VITRO. OF COURSE AS A FURTHER JUMP, WE CAN COMPARE THIS WITH MOUSE MODELS AT ALL OF THESE DIFFERENT LEVELS, USING A NETWORK CONVERGENCE APPROACH. THIS IS NOT JUST PIE IN THE SKY FOR SCIENCE FICTION, BECAUSE& BRAIN ACTIVITY, IN ADDITIO TO GENE EXPRESSION, DISEE FINED IS DEFINED BY NETWORK PROPERTIES. HERE IS A PAPER THAT SHOWS CORRESPONDENCE BETWEEN RESTING STATE ACTIVITY AND BRAIN-GENE EXPRESSION THAT WE DID. THIS SHOWS THAT THE DEFAULT NETWORK ACTIVITY -- DOWN REGULATED IN AUTISM BRAIN. BUT ALSO AT A MICRO SCALE, WHEN WE TOOK SASHA NELSON'S MICRO ARRAY DATA FROM 12 DIFFERENT NEURONAL CELL TYPES, WE IDENTIFIED TWO MITOCHONDRIAL MODULES, THIS IS THE AMINO CYTOCHEMISTRY WITH ONE OF THE HUBS OF THE CELL BODY, MITOCHONDRIA, ONE CAN SEE THAT IT'S CORRELATED WITH THE -- ARRAY OF THAT MODULE, OF ALL OF THE 12 CELL TYPES THAT HAD BEEN PROFILED IN THAT STUDY, AND WHEN MIKE GANDEL LOOKED ACROSS MULTIPLE DISORDERS, I THOUGHT I WAS GOING TO SPEAK AFTER HIM SO THIS WOULD MAKE MORE SENSE BUT HOPEFULLY IT WILL MAKE SENSE WHEN YOU HEAR HIS TALK LATER, IS THAT WHEN HE LOOKS ACROSS DISORDERS, HE SEES, AGAIN, A VERY SIMILAR PATTERN AS SEEN WITH THE DEFAULT MODE NETWORK DATA, WE SEE DOWN REGULATION IN AUTISM BY POLAR AND SCHIZOPHRENIA BUT NOT IN ALCOHOLISM OR MAJOR DEPRESSION. SO IN SUMMARY, WAY TONIGHT TO SAY LARGE SCALE DATA GENERATION PROVIDES AN UNBIASED CONTEXT FOR UNDERSTANDING THE FUNCTION OF GENETIC VARIATION, THAT NETWORK APPROACHES PROVIDED ORGANIZE ORGANIZING -- PHENOTYPES IN THE BRAIN. TO BEGIN TO ELUCIDATE THE NEURAL SYSTEMS BASIS OF AUTISM AND OF COURSE OTHER PSYCHIATRIC DISORDERS. SO AUTISM RISK GENES CONVERGE DURING EARLY FETAL BRAIN DEVELOPMENT GLUE TA MATTER JIK NEURONS, HOWEVER WHEN WE DO -- WE SEE SUBSTANTIAL EFFECTS ON INTERNEURONS AS WELL, SO CLEARLY CLEARLY, WE HAVE TO CONNECT -- TO ACROSS DEVELOPMENT TO WHAT'S HAPPENING IN THE CORTEX, AND THAT'S THE CRITICAL CHALLENGE TO CONNECT THESE DIFFERENT LEVELS AND ONE OF THE METHODS THAT WE'RE USING IS NETWORK AND INTEGRATIVE GENOMICS. SO THANKS FOR YOUR ATTENTION, AND IN BLUE ARE ALL THE FOLKS IN MY LAB WHO WORKED ON THIS, RED ARE OUR CURRENT COLLABORATORS WHO EITHER CONTRIBUTED OR WORKED WITH US ON NETWORK ANALLAL CYST. WE'RE GRATEFUL TO PSYCHENCODE AND OTHER PROJECTS WHO HAVE DEVELOPED THESE RESOURCES THAT WE CAN USE AS WELL AS THE PGC THAT'S PROVIDED SO MUCH RICH GENETIC DATA. THANK YOU. [APPLAUSE] >> THANKS FOR THIS WONDERFUL PERSPECTIVE. SO WE ACTUALLY NOW ARE GOING TO HAVE BREAK, AND I WILL ASK THE IMPOSSIBLE FROM YOU, TO BE BACK HERE WITHIN 10 MINUTES. WE HAVE TWO HOURS LEFT OF REALLY EXCITING STUFF, SO PLEASE COME BACK. DON'T GIVE UP ON US. SO THOMAS IS A SMART GUY, BECAUSE NOW THAT IT'S VIRTUALLY IMPOSSIBLE TO MAKE IT THROUGH, HE'S GIVEN ME BACK THE PODIUM. IF EVERYBODY COULD GET TO THEIR SEATS, WE'RE GOING TO TRY TO GET THROUGH -- I'LL HAVE TO BE RELATIVELY RUTHLESS ON TIME. THE NEXT TWO TALKS HAVE ALREADY BEEN ADVERTISED TO YOU.& THESE ARE ON PSYCHENCODE AND THEN ON MOLECULAR PATHOLOGY FROM MICHAEL GENDAL. WE HAD TO HAVE DAN'S TALK FIRST SO HE COULD RUSH TO THE AIRPORT AND MISS HIS PLANE. SO NENAD. >> THANK YOU, STEF. THANK YOU ALSO FOR PRONOUNCING MY LAST NAME VERY WELL. YOU ARE REALLY -- SO I'LL START BY THANKING THOMAS, DAVID, ANDREA, FOR SUPPORT ORGANIZING ALL OF THIS, AND SO OF COURSE I'LL HAVE THE PRIVILEGE OF PROBABLY GIVING MOTORS BORING TALK BECAUSE I THINK BY NOW, EVERYBODY KNOWS WHAT PSYCHENCODE IS. SO WHAT I DID, YOU ALL PROBABLY SAW ME WORKING ON MY LAPTOP UNTIL THE LAST MINUTE, I REMOVED ALL DATA SLIDES, AND I REALLY -- WHY THIS WAS ALL SET SET UP AND REALLY TALK ABOUT TWO THINGS. ONE IS WHY PSYCHENCODE WILL HOPEFULLY HELP US UNDERSTAND MORE NORMAL HUMAN BRAIN DEVELOPMENT, THROUGH USE OF FUNCTIONAL GENOMICS AND EVERYTHING ELSE ON BEHALF OF PSYCHENCODE DATA FROM OUR LAB. SO THE REASON I THINK THIS IS AN IMPORTANT CONSORTIUM IS FOR SEVERAL REASONS, I'LL START REALLY WITH THE VERY BASIC UNDERSTANDING WHICH I THINK IS IT CONVEYED TO YOU ALREADY, ONE IS WE ARE NOT RODENTS, WE ARE NOT BIG RODENTS AND REALLY TRY TO UNDERSTAND HUMAN DEVELOPMENT OF THE HUMAN BRAIN, I REALLY THINK IT'S IMPORTANT TO STUDY IN THE CONTEXT OF HUMAN TISSUE AND HUMAN BRAIN DEVELOPMENT, WE HAVE UNIQUE FEATURES AND THE SECOND POINT I WANT TO CONVEY WITH THIS SLIDE IS ALSO WHEN WE TALK ABOUT THE DEVELOPMENT, IF YOU LOOK HERE, MAJOR PHASES, BIRTH, CHILDHOOD, YOU CAN SEE THE BRAIN IS UNDERGOING TREMENDOUS CHANGES WHEN YOU ARE BORN, THEN IT TAKES -- ESPECIALLY WHEN YOU ARE BORN, YOU ARE ONLY 25% OF YOUR SIZE AND IT TAKES NEXT THREE YEARS TO BASICALLY COME TO THE FULL ADULT SIZE AND EVEN THEN, YOU WOULD NOT CALL SOMEBODY ADULT. AND IN ESSENCE, IT ACTUALLY TAKES UP TO THE SECOND HALF OF THE THIRD DECADE OF POSTNATAL LIFE TO REALLY CALL SOMEBODY AN ADULT, AND THE REAP FOR THAT IS THAT MANY OF THE KEY DEVELOPMENTAL PROCESSES, HERE I PLOTTED THOSE RELATED TO PREFRONTAL CORTEX SUCH AS NEUROGENESIS TAKE MANY MONTHS, IN THE CASE OF NEUROGENESIS -- IT TAKES ALMOST SEVEN MONTHS WHILE ENEMAS ALL OF THE -- ACTUALLY SEVEN DAYS, AND THEN WHEN YOU THINK ABOUT NEURONAL MIGRATION OF -- CELLS PROBABLY UP TO BIRTH -- TO SHOWN EVEN NEXT COUPLE OF YEARS OF EARLY POSTNATAL LIFE, YOU HAVE TREMENDOUS PROLIFERATION OF OS TROA SITES, OLIGODEN TROA SITES -- ACTUALLY EXPLODES IN FIRST TWO YEARS, AND THEN IT'S ACTUALLY -- FINALLY ALSO IN ONE OF THE ME MECHANISMS LEAST UNDERSTOOD BUT REALLY ONE OF THE MAJOR PROPERTIES POSTNATAL DEVELOPMENTS TAKES SO LONG IS MYELINATION, AND HAS SHOAP IN MANY EVIDENCE, IT LASTS ALL THE WAY TO THE SECOND DECADE OF THE -- THE THIRD DECADE. SO REALLY BASICALLY THE POINT IS THAT WHEN WE THINK ABOUT THE DEVELOPMENT, WE SHOULD NOT THINK JUST ABOUT PRENATAL, IT'S ACTUALLY REALLY PROTRACTED IN HUMANS. THE THIRD REASON I THINK THIS IS IMPORTANT TO UNDERSTAND AND PUT IN CONTEXT IS IF YOU LOOK AT. SO MAJOR NEUROLOGICAL PSYCHIATRIC DISORDERS, YOU CAN SEE ACTUALLY WHEN YOU LOOK AT THE AGE OF DIAGNOSIS, IT'S ACTUALLY INTELLECTUAL DISABILITIES THAT THEY'RE USUALLY DIAGNOSED IN CHILDHOOD, AUTISM USUALLY END OF INFANCY AND EARLY CHILDHOOD. ON THE OTHER HAND, IF YOU LOOK AT NEURODEGENERATIVE DISORDER SUCH AS ALZHEIMER'S, PARKINSON'S, THEY'RE USUALLY IN THE MID OR LATE ADULT. HOWEVER, AS WE ALL KNOW FROM PREVIOUS TALKS THE RISK GENES FOR ALL DISORDERS SEEMS TO BE EXPRESSED AT LEAST DURING SOME PERIODS AS WELL AS THE CELLS AND UNDERSTANDING WHETHER -- WHAT ARE THE DEFECTS AND ALTERATIONS IN NORMAL DEVELOPMENT OF PROCESSES AND IS THAT THE REASON WHY WE ARE SEEING THIS DIFFERENT TRAJECTORIES IN THE AGE OF DIAGNOSIS AND ONSET OF CLINICAL PICTURES OF THESE DISORDERS. I THINK TO REALLY TRY TO UNDERSTAND THE NORMAL DEVELOPMENT IS REALLY IMPORTANT FOR TEASING OUT WHO WE ARE AS SPECIES, AS WELL TRY TO UNDERSTAND COMPLEX DISORDERS SUCH AS THESE LISTED HERE. AS I MENTIONED TO YOU, THE WORK FROM MANY, MANY GROUPS INCLUDING OURS AND I'M GOING TO SHOW YOU SOME DATA OVER THE LAST DECADE TO REALLY TRY TO UNDERSTAND HOW DYNAMICS OF GENE EXPRESSION, MANY OTHER GENOMIC FEATURES ARE CHANGING DURING DEVELOPMENT AND JUST ILLUSTRATE TO YOU HOW COMPLEX THE PROBLEM IS AND HOW LITTLE WE KNOW, I'M GOING TO SHOW YOU THIS SLIDE THAT WAS DONE USING MICRO ARRAYS AND ALSO SIMILAR WORK WAS DONE BY -- INSTITUTE AND MANY OTHER GROUPS, SOME OF WHO ARE PRESENT HERE, AND HERE ARE 1700 SAMPLES DISSECTED, CLOSE TO 60 ARE HUMAN BRAINS RANGING IN AGE FROM FIVE WEEKS POST CONCEPTION ALL THE WAY TO 80 YEARS DIVIDED INTO 15 PERIODS, EACH PERIOD PAINTED IN DIFFERENT COLORS, THIS IS MVS, YOU CAN SEE BASICALLY FIRST COMPONENT DOES APPEAR -- THIS IS PERIOD 1, PERIOD 2. IF YOU LOOK CAREFUL, YOU CAN DIVIDE IT INTO EMBRYONIC, FETAL, THIS IS BIRTH, EARLY POSTNATAL. WHAT'S REALLY INTERESTING ABOUT THIS PICTURE WHAT IT ILLUSTRATES IS THAT ACTUALLY ALMOST TWO THIRDS OF VARIANTS IN GENE EXPRESSION CAN BE EXPLAINED BY PRENATAL EVENTS. THIS IS NOT JUST CHANGES IN GENE EXPRESSION CELLS, IT'S CHANGES IN -- WHEN YOU LOOK AT THE BIRTH TO ACTUALLY 20 YEARS OF AGE, YOU CAN SEE -- THIS IS 20 TO 80 SHOWING HOW U.S. CHANGES COMPARING SHOW HOW YOU HAVE CHANGES -- SPATIAL DIMENSION, HERE ARE THE REGIONS ANALYZING THE STUDY -- CORTICAL AREAS, THAT ALL PAINTED IN IN BLUE, AND YOU CAN SEE, FOR EXAMPLE, IF YOU LOOK AT JUST MID FETAL DEVELOPMENT, YOU CAN ALSO SEE THAT THERE IS A DIFFERENCE œNOT JUST IN TIME BUT ALSO WITHIN THE SAME PERIOD, THIS IS THE SAME GRAPH JUST PAINTED NOW USING COLORS TO REPRESENT DIFFERENT REGIONS, AND YOU CAN SEE THE CORTICAL AREAS, THIS IS CEREBELLUM, AND THE DIFFERENCE IS AS YOU WOULD EXPECT BECAUSE OF ALL ANALYZED NEOCORTEX -- MOST DIFFERENT CELLULAR TYPE. SO BASICALLY THE POINT IS THAT HUMAN BRAIN TRANSCRIPT HIGHLY DYNAMIC SPATIAL DIMENSION ALL OF THIS WAS GENERATED AND MANY OF THE STUDIES DONE AT THE SAME TIME USING -- TISSUE SEQUENCING. WE REALLY KNOW VERY LITTLE HOW CHANGES AT THE LEVEL SPECIFIC CELL TYPES ARE HAPPENING, NOT JUST DPEEN EXPRESSION, GENE EXPRESSION, I THINK THIS SUNDAY LYING -- SO THAT IS WHY NIH WAS REALLY GENEROUS ENOUGH AND HAD THE WISDOM TO IN 2014 SET UP WHAT'S CALLED NOW PSYCHENCODE PROGRAM, AND THE THERE ARE THREE MAIN GOALS, THE FIRST WAS TO SYSTEMATICALLY IDENTIFY GEE GENOMIC ELEMENTS, EPIGENETIC MODIFICATIONS IN DEVELOPING AND ADULT HUMAN BRAIN. TYPICAL BRAIN, FUNCTIONALLY CHARACTERIZE -- TO MICE, MEMBERS OF THE CONSORTIUM SHOWED YOU EXAMPLES OF BOTH OF THEM, AND THIRD, THIS ALSO -- SEVERAL OF MY COLLEAGUES HAVE ALREADY SHOWN YOU EXAMPLES, SO THE IDEA IS BASICALLY REALLY TO TRY TO HELP US TO UNDERSTAND HOW DO WE GO FROM DNA TO BEHAVIOR, IN PARTICULAR UNDERSTANDING HOW THE VARIATIONS IN GENOMIC DNA LEADS TO CHANGES IN PROTEINS, CELL CIRCUITS AND BEHAVIOR. AND SO TELL YOU A LITTLE ABOUT CONSORTIUM,, WE HAVE DATA COORDINATION ANALYSIS CENTER, WE ARE -- SCHIZOPHRENIA, BIPOLAR DISORDERS AND ALSO MY GROUP AND SEVERAL OTHER GROUPS ARE ALSO LOOKING AT TYPICAL DEVELOPMENT ACROSS -- TYPICAL DEVELOPMENT INCLUDING ENTIRE PRENATAL EARLY POSTNATAL ADULT PERIODS, WE ARE USING MULTIPLE TECHNIQUES, I'M NOT GOING TO MENTION THEM BECAUSE I THINK ALL METRICS -- FINAL LIMOS OF THE STUDIES HAVE BEEN INITIALLY DONE WITH THE -- SCWEEPSING, WE HAVE MOVED TO CELL TYPE, I THINK -- ALSO -- FINALLY WE HAVE -- PREVIOUS PRESENTATIONS. SO THESE ARE THE I DON'T GO TO ALL OF THEM, I JUST PUT THEM SO I DON'T FORGET EVERYBODY AND I'LL TELL YOU WHERE CAN YOU FIND ALL THIS INFORMATION AND DATA. ALSO I THANK MEMBERS OF THE NIMH WHO HAVE BEEN SUPPORTIVE IN REALLY MANAGING THIS PROJECT AND WORKING WITH US. YOU CAN SEE THERE ARE MULTIPLE GROUPS, MULTIPLE UNIVERSITIES AND INSTITUTIONS. OUR FIRST MARK WAS -- PUBLICATION WAS A MARKET PAPER THAT CAME IN TWEP 14, I JUST HIGHLIGHTED THE KEY PAPERS, SORRY IF I MISSED ANY OF YOUR PAPERS, FROM SEVERAL GROUPS REALLY SHOWING THAT THE FLAVORS AND TYPE OF DATA COMING OUT. THESE TYPE OF PAPERS COME IN TWO FLAVORS, ONE IS THE DATA GENERATION -- THREE, SORRY. DATA GENERATION, NEW PIPELINES D AND THE THIRD IS A LITTLE MORE MECHANISTIC STUDIES USING THE AVAILABLE DATA. SO THERE ARE THREE TYPES OF EFFORT WE ARE TRYING TO, DO TYPE OF STUDIES. FIRST IS A PROJECT SPECIFIC OR INDIVIDUAL GROUPS, I'M GOING TO TELL YOU A LITTLE MORE ABOUT IT, THEN ALSO WE HAVE A -- THE FIRST ONE IS INDIVIDUAL TEAM GO -- THEY INTERACT WITH THE DATA COORDINATION CORE OR CENTER THAT IS BASICALLY RUN BY -- USING SYNAPSE PLATFORM. THE IDEA IS TO SHARE AND STORE AND PROCESS DATA, DATA ANALYSIS CORE, THE DOLE OF THE DATA ANALYSIS CENTER IS TO PROCESS ALL OF THE DATA, AND REALLY FACILITATE META INTEGRATIVE ANALYSIS. THE SECOND TYPE OF EFFORT WE HAVE DONE, MAKE SURE I CAN TELL YOU WHEN IT'S FINISHED I THINK IT WILL BE REALLY HELPFUL TO ALL OF YOU WHICH IS THE REFERENCE BRAIN PROJECT, WE HAVE ALSO DECIDED TO GENERATE A MULTIPLE LEVEL OF THE DATA FROM THE SAME SET OF BRAIN AND TISSUE SAMPLE, AND BECAUSE WE ALL APPLY USING DIFFERENT APPROACHES, BASICALLY WE HAVE FOUR BRAINS AND WE HAVE TISSUE -- SO WE CAN ALL -- PARTICIPATE IN ALL OF THESE DATASET AND ALL OF THIS IS AVAILABLE NOW IN THE SYNAPSE, SO WE ALSO HAVE FIVE -- PAPERS, I WILL LIST THEM FOR THE SAKE OF TIME, ONE IS -- ALL OF THESE HAVE BEEN MENTION TODAY CERTAIN EXTENT IN THE PRESENTATIONS. CHALLENGES. DO WE HAVE A LARGE ENOUGH SAMPLE SIZE, WE WOULD LIKE TO HAVE A META TISSUE PROCUREMENT, WE WOULD LIKE TO HAVE A LITTLE MORE AUTISM BUT AT THE MOMENT WE HAVE THE LARGEST SET OF T WE NEED TO INCREASE THE CELL RESOLUTION, I THINK THAT WAS ONE OF THE THEMES OF MAKE OF THE PSYCHENCODE, WE DON'T KNOW HOW MUCH OF THAT IS RELATED TO DISORDERS WE ARE STUDYING. I DID NOT TALK -- WE HAVE RECENTLY CREATED A PROCUREMENT MAP WHICH WAS A POSTNATAL HUMAN BRAIN DEVELOPMENT AND REALLY TRY TO UNDERSTAND GENE EXPRESSION, PROTEIN CORRELATE, AND REALLY NEED FOR A LITTLE BIT MORE SYSTEMATIC ANALYSIS CHARACTERIZATION OF BOTH ELEMENTS USING -- HUMAN -- CELLS. I'LL THANK YOU ALL FOR YOUR ATTENTION AND I LOOK FORWARD TO QUESTIONS. THANK NIMH, PSYCHENCODE MEMBERS, FOR UPDATES FOR ALL ACKNOWLEDGMENTS, PLEASE VISIT PSYCH ENCODE.ORG. THANK YOU FOR YOUR ATTENTION. [APPLAUSE] >> THANK YOU FOR SPEAKING QUICKLY. IT WAS GREAT. I'M SORRY WE DON'T HAVE TIME FOR DISCUSSION OF THESE. WE HAVE TO MOVE ON TO OUR NEXT TALK FROM MICHAEL GANDAL. YOU'VE BEEN WELL ADVERTISEED BY DAN ALREADY. >> THANK YOU. IT'S A GREAT HONOR TO BE UP HERE, AND I WANT TO THANK NIMH AND THE PROGRAM FOR DEVELOPING SUCH A WONDERFUL SYMPOSIUM. I'M GOING TO PRESENT SOME UNPUBLISHED DATA FROM THE SIEK ENCODE TO TRY AND UNDERSTAND BETTER THE MOLECULAR NEUROPATHOLOGY ACROSS A NUMBER OF DIFFERENT PSYCHIATRIC DISORDERS. THIS IS WORK THAT I DID AS A POSTDOC IN THE -- LAB. SO I DON'T THINK I NEED TO INTRODUCE THE CHALLENGE OF BUT I THINK ONE OF THE CRITICAL ISSUES THAT WE'VE BEEN TRYING TO TACKLE WITH THIS QUESTION OF POLYJE NECESSITY, AS THE EARLIER TALKS THIS MORNING REALLY DID A NICE JOB, HIGHLIGHTING ALL THE ACCESSES IN GENE DISCOVERY OVER THE LAST SEVERAL YEARS, THIS IS SORT OF PARADOXICALLY ACTUALLY LED TO A NEW CHALLENGE, WHICH IS HOW DO WE MAKE SENSE OF ALL THESE DIFFERENT GENETIC VARIANTS AND GENES THAT HAVE BEEN LINKED TO DURCH DISEASES. STEFFAN SANDERS HAS HIGHLIGHTED 64, 65 HIGH RISK AUTISM GENES, AND IF WE START TO LOOK AT THE NUMBER OF GENES LINKED TO AUTISM AS A CHARACTERISTIC EXAMPLE, THERE ARE GOING TO BE MANY HUNDREDS IF NOT THOUSANDS OF GENES ASSOCIATED WITH THE DISEASE, SO HOW DO WE START TO UNDERSTAND WHAT THESE ARE DOG BOTH AT A POPULATION AND INDIVIDUAL LEVEL. THEY TAKE A NUMBER OF DIFFERENT FORMS WHICH VARIOUS PEOPLE HAVE TALKED ABOUT, INCLUDING CNVs, SINGLE NUCLEOTIDE VARIANTS, FORMS OF ASD, BUT STILL REALLY ONLY ACCOUNT FOR A SMALL HANDFUL OF CASES, LESS THAN 1% OF INDIVIDUAL CASES, TEND TO HAVE HIGHLY ADDITIVE EFFECTS AND SHOW STRONG -- SO A MAJOR QUESTION THAT WE'VE BEEN TRYING TO ANSWER OR ASK IS HOW DO WE UNDERSTAND WHAT THE CONVERGENT BIOLOGICAL IMPACT IS OF MANY OF THESE DIFFERENT RISK VARIANTS. SO THE APPROACH WE TAKE AS DAN NICELY INTRODUCED, TO REALLY KNOW CUSS ON THE SYSTEMS LEVEL, THE MOLECULAR HIERARCHY AND THE SENTRA DOGMA OF BIOLOGY AND REALLY UNDERSTAND THAT THE GENETIC VARIATION TRANSLATES INTO CHANGES IN GENE EXPRESSION OR POTENTIALLY TRANSLATES INTO GENE EXPRESSION OR SPLICING ALTERATIONS IN THE TRANSCRIPTOME THAT THEN GET INTEGRATED INTO PROTEIN ABNORMALITIES AND THEN WORK UP IN THE SYSTEMS LEVEL HIERARCHY TO LEAD TO ALTERATIONS IN BEHAVIOR AND COGNITIVE PROCESSES, SO WE USE THE TRANSCRIPTOME AS A QUANTITATIVE PHENOTYPE THAT REPRESENTS THE FULL GENOME IN A LARGE SCALE UNBIASED WAY THAT WE CAN USE TO HELP POTENTIALLY IDENTIFY THE CONVERGENT COUNCIL STREAM EFFECTS OF MANY GENETIC VARIANTS, MANY OF WHICH WE DON'T EVEN KNOW YET, BUT THAT POTENTIALLY CON CONVERGE ON ALTERATION OF GENE EXPRESSION IN THIS HIERARCHY, AND AS DAN MENTIONED, GENES DO NOT ACT INDEPENDENTLY, THEY FORM THESE COEXPRESSION NETWORKS AND WE CAN USE THAT STRUCTURE TO BEGIN TO IDENTIFY MORE SPECIFIC BIOLOGICAL PROCESSES THAT MAY BE JUST REGULATED IN DISEASE. SO TO GIVE AN EXAMPLE IN AUTISM, THE FIRST STUDY FROM THE GESH WIN LAB THAT LOOKED AT THIS WAS FROM 2011 FROM IRINA VONAGUE USING GENE EXPRESSION MICRO ARRAYS TO CHARACTERIZE GENE ALTERATION AUTISM MATCH CONTROLS, AND THIS IS A STUDY THAT WE PUBLISHED MORE RECENTLY IN 2016, USING RNA SEQUENCING TO EXTEND THOSE ORIGINAL FINDINGS. SO HERE WE ARE A NUMBER OF CASES OF AUTISM, ABOUT 50 CASES OF AUTISM AND 50 MATCH CONTROLS. WE HAVE FRONTAL AND TEMPORAL CORTICAL BRAIN SAMPLES AS WELL AS CEREBELLUM, AND RAN A LARGE SCALE RNA SEQUENCING EXPERIMENT, AND WE IDENTIFY A VERY LARGE SUBSTANTIAL CHANGES IN GENE EXPRESSION SHOWN HERE IN THIS HEAT MAP PLOT AND IMPORTANTLY THE CHANGES IN GENE EXPRESSION WERE NOT PRIMARILY DRIVEN BY KNOWN BIOLOGICAL TECHNICAL COVAIR YATSVARIANTS THAT AFFECT AGE, SEX OR QUALITY OF THE TISSUE BUT REALLY SEEM TO BE DRIVEN BY CASE CONTROL STATUS. INTERESTINGLY, WE ALSO LOOKED AT THE CEREBELLUM, SO ALL OF THIS HEAT MAP HERE IS FROM CORTICAL SAMPLE, AND WE ACTUALLY DON'T SEE ANY CHANGES IN CEREBELLUM OR VERY, VERY ATTENUATED CHANGES IN CEREBELLUM HIGHLIGHTING THE IMPORTANCE OF LOOKING AT INTERREGION SPECIFIC MANNER. AND WHEN YOU TAKE ALL OF THESE GENE EXPRESSION MEASURES ACROSS SAMPLES AND YOU USE THIS NETWORK FRAMEWORK THAT DAN INTRODUCED, WE IDENTIFY SEVERAL MODULES OF COEXPRESSED GENES WHICH ALMOST ALL TAG THE PRIMARY CELL TYPES OF THE CNS, AND SO WE CAN ESSENTIALLY COMPUTATIONALLY DEACONVOLVE AND GET A SENSE FOR HOW THEY'RE AFFECTED IN DISEASED BRAIN. SO HERE WE IDENTIFY A NUMBER OF MODULES THAT ARE UP REGULATED IN AUTISM, M9, STRONGLY ENRICHED FOR KNOWN CANONICAL ASTROCYTE MARKERS AND STRONGLY UPREGULATED IN DISEASE, AND HERE WE FIND A PROVIDE CROW GLIA MICROGLIA MODULE ALSO IDENTIFIED IN AUTISM. SO WE CAN USE THIS APPROACH TO BEGIN TO ZOOM IN ON THE MOLECULAR ALTERATIONS THAT ARE PRESENT IN ASD BRAIN. SO ONE OF THE MAJOR QUESTIONS OF THE PSYCHENCODE CONSORTIUM IS HOW DOES -- WE KNOW THAT PSYCHIATRIC DISEASE DOES NOT OBEY THESE DISCRETE BINARY CLASSIFICATION BUT REALLY LIVES ON THIS CONTINUUM, SO CAN WE USE THIS UNDERSTANDING TO BEGIN TO COMPARE AND CONTRAST DISTINCT -- PREVIOUSLY THOUGHT TO BE DISTINCT PSYCHIATRIC DISORDERS AND REALLY IDENTIFY THESE AXIS OF NEUROPATHOLOGY IN GENE EXPRESSION. SO TO DO THIS, WE'VE COMPILED PRELY MUCH ALL AVAILABLE GENE EXPRESSION MICRO ARRAY -- ACROSS FIVE DIFFERENT PSYCHIATRIC DISEASES INCLUDING AUTISM, SCHIZOPHRENIA, BIPOLAR, DEPRESSION, CHRONIC ALCOHOLISM ALCOHOLISM -- TO BALANCE EACH DATASET INDIVIDUALLY SO THAT CASES AND CONTROLS WERE WELL MATCHED IN TERMS OF AGE AND SEX AND TECHNICAL COVARIANTS THAT CAN AFFECT GENE EXPRESSION, AND PIPELINE TO ANALYZE EACH DATASET INDIVIDUALLY SO THAT WE CAN THEN REGRESS OUT STUDY SIZE IESKTS, BATCH EFFECTS AND NORMALIZE ALL THE DATA TOGETHER TO PERFORM DIFFERENTIAL EXPRESSION AND NETWORK ANALYSIS AS WE HAD DONE IN AUTISM. SO USING THIS APPROACH LOOKING AT THE SORT OF 10,000-FOOT VIEW, WE CAN COMPARE THE TRANSCRIPTOME IN ONE DISEASE, THE TRANSCRIPTOMIC CHANGES IN ONE DISEASE WITH THE TRANSCRIPTOMIC CHAIX IN ANOTHER DISEASE. SO HERE I'M SHOWING YOU ON THE X AXIS, AUTISM, THE 10,000 OR SO GENES EXPRESSED IN THE BRAIN AND PRESENT ACROSS ALL THE DIFFERENT STUDY. ON THE RIGHT SITE ARE GENES UPREAG LITED IN AUTISM, THESE ARE GENES DOWNREGULATED IN AUTISM, WHAT WE SEE IS THAT THERE'S -- HERE I PLOT ON THE Y AXIS THE SAME EFFECT SIZES FOR SCHIZOPHRENIA AND WE CAN SEE THAT THERE'S A STRONG CORRELATION SUGGESTING THAT THERE'S A SHARING OF GENE EXPRESSION ALTERATIONS ACROSS THOSE DISEASES. WE CAN QUANTIFY THIS, WE SEE A RO VALUE OF .44, HIGHLY SIGNIFICANT BOTH BY PARAMETRIC AND NON-PARAMETRIC METHODS. IN CONTRAST, IF WE LOOK AT OTHER DISEASE COMPARISONS IN THIS CASE, THIS IS CHRONIC ALCOHOLISM, WE SEE NO SIGNIFICANT TRANSCRIPTOMIC OVERLAP WITH AUTISM, ANY KIND OF -- THAT'S MEANINGFUL, SO THIS IS IMPORTANT BECAUSE WE KNOW THAT SCHIZOPHRENIA, FOR EXAMPLE, HAS AA DEGREE OF CO-MORBIDITY WITH SUBSTANCE EU WE WANTED TO MAKE SURE THAT THOSE TYPES OF FACTORS WERE NOT DRIVING THE EXPRESSION CHANGES THAT WE SEE. SO IF WE REPEAT THESE TYPES OF APPROACHES AND THIS TYPE OF ANALYSIS IN A PAIR WISE FASHION, WE CAN DEVELOP A NETWORK OF TRANSCRIPTOMIC SHARING OR SIMILARITY AT THIS GLOBAL LEVEL ACROSS DISORDERS. WE SEE NOT SURPRISINGLY THAT SCHIZOPHRENIA AND BIPOLAR DISORDER SHARE THE GREATEST DEGREE OF OVERLAP AT THE BRAIN GENE LEVEL AND STRONGLY ASSOCIATED WITH DEPRESSION, ALBEIT LESS SO THAN WITH EACH OTHER, WHAT WAS PERHAPS MORE REMARKABLE WAS THAT AUTISM SHOWS A STRONG CORRELATION ABOUT SCHIZOPHRENIA AND PIE POLAR DISORDER, CHRONIC ALCOHOLISM SHOWED NO POSITIVE OVERLAP WITH ANY OF THE OTHER DISORDERS. SO THIS COULD BE -- THE TRANSCRIPTOME ITSELF IS -- WE'RE LOOKING AT POSTMORTEM TISSUE EITHER CAUSALLY RELATED TO DISEASE OR A CONSEQUENCE OF DISEASE AND IT'S VERY HARD TO DISENTANGLE THOSE TYPES OF CHANGES. ARE THE GEEPS THAT ARE DIFFERENTIALLY EXPRESSED RELATED TO KNOWN GENETIC CHANGES, SO HERE I'M PLOTTING THE TRANSCRIPTOMIC SIMILARITY BETWEEN DIFFERENT DISORDERS FOR THE LAST SLIGHT, COMPARED WITH THE GENETIC SNP BASED CORRELATION FROM THE PGC2013 PAPER AND WHAT WE SEE IS THE SUBSTANTIAL RELATIONSHIP WHERE THE MORE GENETICALLY SIMILAR TWO DISORDERS ARE, THE MORE SIMILAR THE TRANSCRIPTOME PATTERNS ARE. WE COULD ALSO GAIN SOME INFORMATION FROM THE SLOPE OF THIS OVERLAP AND QUANTIFY WHAT WE CALL SORT OF THE TRANSCRIPTOMIC SEVERITY. HERE I'M USING SCHIZOPHRENIA AS A REFERENCE SO IT'S A UNITARY LINE IN BLACK AND WE PLOT THE SLOPE OF AUTISM TRANSCRIPTOME CHANGES IN RED, BIPOLAR IN GREEN, AND DEPRESSION IN BLUE, AND WE CAN SEE THERE'S THIS GRADIENT WHERE AUTISM OR THE EARLIEST ONSET OF DISEASE SHOWS THE MOST SIGNIFICANT OR SEVERE STRONG CHANGES AT THE -- WHERE SCHIZOPHRENIA AND BIPOLAR ARE ABOUT THE SAME AND DEPRESSION SHOWS THE LEAST SEVERE CHANGES. IMPORTANTLY WE'VE REPLICATED THIS NOW IN MULTIPLE INDEPENDENT RNA SEQ DATASETS AS PART OF THE PSYCHENCODE, THE COMMON MIND DATA THAT PAMELA AND PANOS TALKED ABOUT, AND THIS IS THE BRAIN GVEX DATASET FROM THE CHICAGO GROUP. SO THIS REALLY GIVES US LIKE A 10,000-FOOT VIEW OF GLOBAL CHANGES IN THE TRANSCRIPTOME, IT DOESN'T REALLY HONE IN ON SPECIFIC BIOLOGICAL PROCESSES, SO WHAT WE DID NEXT IS TO USE NETWORK ANALYSIS IN THIS COEXPRESSION TO BEGIN TO IDENTIFY SPECIFIC MODULES OF GENES THAT ARE COEXPRESSED ACROSS DISORDERS IN DIFFERENT DIRECTIONS OR WAYS, SO AS DAN HAD SHOWN BEFORE, THIS UNSEUF ADVISED CLUSTERING TECHNIQUE -- TO PULL OUT THE MAJOR CELL TYPES OF THE BRAIN, WE SEE THAT HERE AS WELL, SO THESE ARE THE MAJOR MODULES THAT I'VE IDENTIFIED ACROSS ALL OF OUR SAMPLES. EACH ONE IS VERY STRONGLY ENRICHED FOR CELL TYPE SPECIFIC MARKERS, SO MICRO GLIAL MODULE, ENDOTHELIAL MODULE, SEVERAL NEURON MODULE, ASTROCYTE MODULE THAT ARE STRONGLY ENRICHED IN ONE CELL CLASS BUT NOT ANY OF THE OTHERS. SO NOW HAD WE LOOK AT THE CHANGE IN MODULE IGON GENES, WE IDENTIFY FOUR DIFFERENT MODULES THAT WERE FAIRLY CLUSTERED TOGETHER WITH EACH OTHER AND NOW WE CAN QUANTIFY HOW THESE MODULES ARE CHANGED IN DISEASE BRAIN. WE SEE ARE STRONGLY DOWNREGULATED IN AUTISM, SEVERAL OF THEM ARE DOWN REGULATED IN BIPOLAR DISORDER, NOTHING IN DEPRESSION AND SEVERAL ARE DOWN DOWNREGULATED IN SCHIZOPHRENIA. WE SEE DIFFERING CHANGES IN ALCOHOLISM, AND THIS IS THIS PURPLE MODULE RIGHT HERE THAT DAN MENTIONED IS RELATED TO NON-SYNAPTIC MITOCHONDRIA AND HIGHLY CORRELATED WITH THE FIRING RATE OF CELL. SO THIS BEGINS TO GIVE US A PICTURE THAT SYNAPTIC PROCESSES ARE DOWNREGULATED ACROSS MULTIPLE PSYCHIATRIC DISEASES. IN CONTRAST, WE SEE SEVERAL OF THE GLIAL MODULES SEW STRONG UPREGULATION, IN THIS CASE, THIS IS OUR AS ASTROCYTE MODULE STRONGLY UPREGULATED IN AUTISM, BIPOLAR DISORDER AND SCHIZOPHRENIA, WE IDENTIFY ONE MICRO GLIAL MODULE WHICH IS VERY STRONGLY UPREGULATED IN AUTISM ALONE AND NOT THE OTHER PSYCHIATRIC DISEASES. THIS IS ALSO ENRICHED FOR KNOWN MARKERS WITH HLADRA, SO FINALLY WE WANTED TO GAIN SOME INSIGHT INTO WHAT'S DRIVING THESE CHANGES. WE THINK IT'S RELATED TO GENETICS, AND SO WE TOOK THE SUMMARY STATISTICS, GENE LEVEL SIGNIFICANCE VALUES, WE COLLAPSED THE GENE LEVEL TO GENERATE SIGNIFICANT VALUES AND LOOKED TO SEE HOW VONGLY THAT WAS ENRICHED FOR THE DIFFERENT MODULES THAT WE IDENTIFIED. SO WHAT I'M SHOWING HERE, THE FOUR NEURON MODULES IN RED ARE ALL VERY STRONGLY ENRICHED FOR DISEASE GWAS SIGNAL. SO HERE MOST STRONGLY IN SCHIZOPHRENIA, BUT ALSO IN AUTISM, THIS IS THE LATEST GWAS THAT HASN'T BEEN PUBLISHED YET, BIPOLAR DISORDER, WE SEE THREE OF THE FOUR NEURON MODULES ARE ENRICHED, AND OTHER ANTHROPOMORPHIC TRAITS, ALL OF THESE MODULES ARE DOWNREGULATED IN THESE DISEASES. IN CONTRAST, WE TONIGHT SEE DON'T SEE ANY -- FINALLY WE CAN LOOK AT RARE VARIANTS, SO I TOOK ALL OF THE -- OR SEVERAL OF THE WHOLE EXOME SEQUENCING STUDIES FROM AUTISM AND SCHIZOPHRENIA AND COMPILED A LIST OF DE NOVO NON-SYNONYMOUS RDNVs AND LOOKED FOR ENRICHMENT AMONG OUR MODULES AND FOUND THAT THE NON-SYNONYMOUS DE NOVO VARIANTS IN AUTISM AND SCHIZOPHRENIA ARE ENRICHED FOR THE NEURONAL MODULE WHEREAS THE SILENT DE NOVO VARIANTS WERE NOT, WHEN WE SEE A SIMILAR PATTERN FOR GENES IN -- IN AUTISM. SO TO CONCLUDE, WE CAN GENERATE A MODEL BASED ON THIS DATA THAT BOTH COMMON GENETIC VARIATION, RARE DE NOVO VARIATION SEEM TO CONVERGE ON SYNAPTIC OR NEURONAL DYSFUNCTION IN COMBINATION WITH OTHER FACTORS LIKE ENVIRONMENT THAT'S ASSOCIATED WITH UPREGULATION OF ASTROCYTE AND MICRO GLIAL ACTIVATION. WE SEE THIS PROCESS VERY STRONGLY IN AUTISM, WHEREAS SCHIZOPHRENIA SHOWS A MORE SUBTLE CHANGE THAT'S >> THANK YOU VERY MUCH. PLAIS [APPLAUSE] >> OKAY. SO NOW ALREADY FORESHADOWED, APPARENTLY THERE'S A MEETING TOMORROW, RIGHT, ON MOSCIASM. >> I'M VERY GRATEFUL TO NIH FOR SUPPORTING OUR WORK. IT'S A PLEASURE TO SPEAK ON BEHALF OF THE BRAIN SOMATIC KNOW SAY SISM NETWORK, A GROUP OF INVESTIGATORS CHARGED WITH UNDERSTANDING THE MEANING FOR BRAIN DEVELOPMENT. SO WHAT IS SOMATIC MOSAICISM? THESE ARE MUTATIONS KNOWN TO OCCUR NOT IN THE GERMLINE, AS WE'VE BEEN SPEAKING THROUGHOUT THE DAY, BUT IN THE SOMATIC LINEAGES. THESE ARE THE LINEAGES THAT START FROM THE FERTILIZATION, HERE IS THE ZYGOTE, AND THEY DEVELOP PROGRESSIVELY THROUGH VEL M. DEVELOPMENT SO THESE MUTATIONS CAN HAPPEN AT ANY TIME DURING THIS TREE. AS YOU CAN IMAGINE, THEY CAN ALSO INVOLVE THE GERMLINE SO THE GERMLINE SEPARATES FROM THE FERTILIZED -- ABOUT THE FIFTH YOU SEE HERE PRIMARILY GERMLINE -- SO IF A MUTATION OCCURS EARLY ENOUGH, SURE, IT CAN BE TRANSMITTED THROUGH THE GERMLINE, AND IN FACT, DE NOVO MUTATIONS HAVE BEEN SHOWN TO ARISE FROM THE PROCESS OF SOMATIC MOSAICISM FROM THE EARLY EMBRYO. SO THERE ARE TWO CONSEQUENCES OF SOMATIC MOSAICISM. ONE IS AS I SAID ONLY PRESENT IN THE SOMATIC CELLS, AND AS A CONSEQUENCE OF THAT, NOT ALL SETS WILL SHARE THIS MUTATION, SO IT'S PARTICULARLY HARD TO DETECT THEM. THEY'RE ONLY PRESENT IN SOMETIMES VERY SMALL PERCENT OF THE CELLS IN THE BODY. SO YOU HAVE TO LOOK FOR THEM. WHY WE'RE STARTING THIS? WELL, WHAT WE BELIEVE IS THAT IT COULD ACTUALLY EXPLAIN VARIABLE -- GERMLINE MUTATIONS. OBVIOUSLY WE'LL PROBABLY HAVE -- EFFECT AND PEOPLE HAVE SHOWN IT HAS PHENOTYPIC EFFECT EVEN THOSE IT COULD STILL BE VERY SMALL, THEY COULD STILL EXERT QUITE POWERFUL PHENOTYPES. ALSO -- PERHAPS CONTRIBUTE TO THE PHENOTYPE, BY INTERACTING WITH GERMLINE MUTATIONS. HOWEVER, WE DON'T KNOW TO WHAT EXTENT THIS IS AN ADAPTIVE OR NON-ADAPTIVE PHENOMENON, AND WE DON'T KNOW THE ACTUAL FREQUENCY AND THE TIME OF ORIGIN. IN TERMS OF POTENTIAL CAUSE, IT COULD HAPPEN AT ANY TIME BOTH DURING PROGENITOR EXPANSION IN EMBRYONIC AND FETAL DEVELOPMENT OR IN POST -- SUCH AS NEURONS, AND IT COULD BE A VARIETY OF CAUSES THAT I'M NOT REQUESTING TO GET INTO DETAILS BUT BASICALLY YOU COULD HAVE FAULTY DNA DUPLICATION PROCESSES, DEFECTS IN DNA REPAIR AND EVEN ENVIRONMENTAL DAMAGE LIKE OXIDATIVE DAMAGE, TOXIC, CHEMICAL OR RADIATION. SO AS I MENTIONED, IT'S NOT AN EASY IF HE NO, PHENOMENON TO INVESTIGATE AND IT'S EASILY MISSED. IN THE CONSORTIUM, WE HAVE APPROACH DEVELOPED APPROACHES WE COULD USE. THIS WAS PUBLISHED IN "SCIENCE" VERY RECENTLY. YOU CAN JUST TAKE BULK TISSUE FROM THE BRAIN AND SEQUENCE IT, RIGHT? BUT HOWEVER, THIS IS THE SIMPLEST AND MORE STRAIGHTFORWARD APPROACH, AND PRETTY MUCH DEVOID OF LARGE NUMBER OF FALSE POSITIVE. BUT AS YOU CAN IMAGINE, THE COVERAGE FOR SOMATIC MUTATIONS THAT COULD BE VERY RARE, SAY 1% OF THE CELLS, IT'S VERY LOW. SO WITH THIS APPROACH, YOU CAN DETECT A STANDARD GENOMIC -- PROBABLY SOMATIC MUTATION THAT INVOLVE 10 TO 20% OF THE CELLS. ON THE OTHER HAND, ON THE OPPOSITE SIDE, YOU CAN ISOLATE NEURONS, AND YOU CAN DO WHOLE GENOME AMPLIFICATION OR SINGLE GENOMES AND SEQUENCE THOSE AND THE PROBLEM THERE, OF COURSE, WELL, THE ADVANTAGE IS THAT YOU ARE LOOKING WITH THE HIGHEST SENSITIVITY POSSIBLE AT SINGLE NEURONS, FOR EXAMPLE, AND COMPARE AMONG THEM AND DISCOVER SOMATIC MUTATIONS THAT WAY BY LOOKING AT THE DIFFERENCES, BUT YOU RUN INTO THE PROBLEMS OF WHOLE GENOME AMPLIFICATION WHICH IS NOT GOING TO PRODUCE ERRORS SO THEREFORE IT NEEDS TO BE CAREFULLY THOUGHT OUT. THEN THERE ARE INTERMEDIATE APPROACHES LIKE -- NOT ALL NEURONS COULD BE -- NOT ALL CELLS COULD BE CROENLY EXPANDED. FINALLY YOU COULD TAKE FRACTIONS AND AVOID WHOLE GENE SEQUENCING THAT WAY. SO OUR NETWORK IS CHARGED TO UNDERSTAND THE CAUSES AND THE CONSEQUENCES OF SOMATIC MOSAICISM AND IT'S COMPOSED OF SIX BASIC PROJECTS. EACH PROJECT HAS SEVERAL P.I.s AND SEVERAL NODES. BASICALLY THERE ARE GROUPS LOOKING AT SOMATIC MOSAICISM AUTISM SPECTRUM DISORDER OUT OF HARVARD IN COLLABORATION WITH YALE, AND THEN THE GROUPS LOOKING AT SCHIZOPHRENIA, AND WE'RE LOOKING AT TURRETTE'S SYNDROME AND JOE IS LOOKING AT CORTICAL DYSPLASIA, AND ANOTHER GROUP IS LOOKING ALSO AT SCHIZOPHRENIA AND FINALLY JONATHAN IS LOOKING AT A VARIETY OF DISORDERS INCLUDING AUTISM AND BIPOLAR DISORDER. SO MOST OF US USE A HIGHLY COMPLEMENTARY APPROACH BUT HIGHLY DIVERSIFIED APPROACH. ALL OF WHICH START WITH -- TISSUE. I'M SHOWING THE APPROACHES OF THE WEINBERGER MODERN ENGAGE GROUP. SO THE TISSUE HAS TO BE HIGHLY CHARACTERIZED. YOU NEED TO KNOW EXACTLY WHERE YOU'RE DOING YOUR DISSECTION AND WHAT SENSE YOU'RE ISOLATING IT. OFTEN WE COMPARE SINGLE SEQUENCE MODELED IN VARIOUS SYMPTOMS, TO UNDERSTAND THE FUNCTIONAL IMPLICATION OF THIS PHENOMENON. ALSO AS YOU SEE HERE, GROUPS USE DIFFERENT HUMAN SOURCES, HUMAN BRAIN SOURCES LIKE IN THIS CASE, THE SCHOOL SCHOOL OF MED SIB USE SCHOOL OF MEDICINE USED THE COMMON MIND CONSORTIUM BRAINS WHICH -- EVENTUALLY THE RANGE OF PHENOMENA THAT WE CAN INVESTIGATE. SO I'M GOING THROUGH VERY QUICKLY A FEW SLIDES TO SHOW YOU THE IMPORTANCE OF SOMATIC MOSAICISM FOR TISSUE THAT'S ALREADY BEEN INVESTIGATE BID SEVERAL GROUPS INCLUDING MEMBERS OF OUR CONSORTIUM. SO HERE YOU SEE A PAPER FROM THE -- WHO HAS ACTUALLY LOOKED AT L1 WHICH CAN INSERT RANDOMLY OR NON-RANDOMLY IN VARIOUS LOCI, AND IT HAS ESTIMATED THAT -- THROUGH ANALYSIS OF NEURONAL STEM CELLS AND THEIR PROGENY RAN ABOUT PINT .6 TO .9 -- INSERTION PER CELL WITH AT LEAST 44% CONTAINING INSERTION. THIS IS REMARKABLY CONSISTENT WITH AN EARLIER WORK FROM THE WALSH GROUP AT HARVARD SHOWING THERE ARE ABOUT .6 -- THAT SUGGEST EVEN DIFFERENT GROUPS IN A HIGHLY CONTROVERSIAL FIELD CAN ACTUALLY COME WITH -- CONSISTENT ESTIMATES. I HAVE A FEW SLIDES FROM THE GROUP, WE LOOKED AT PRENATAL FETAL STAGE HUMAN BRAIN, WE ISOLATED SINGLE CELLS, WE EXPAND THEM INTO CLONES, AND THEN WE SEQUENCE THESE CLONES TO LOOK AT VARIANTS INHERITED IN THE CLONE FROM THE ORIGINAL CELL THAT WAS THE CLONE FOUNDER CELL. IN THE CLONES, THE SOMATIC MUTATIONS THAT ARE PRESENT IN THAT ORIGINAL -- HAVE A 50% ALLELE FREQUENCY AS YOU CAN IMAGINE, SO THEY'RE QUITE EASILY DISCOVERED. AFTER VARIOUS PROCEDURES WHERE WE OF COURSE EXCLUDE GERMLINE VARIANTS AND WE COMPARE THESE CLONES WITH EACH OTHER AND WITH ORIGINAL BRAIN TISSUE, WE CAN DISCOVER SOMATIC VARIANTS. HOUR, THESE VARIANTS NEED TOHOWEVER, THESE VARIANTS NEED TO BE VALIDATED TO GET RID OF CERTAIN ARTIFACTS. THIS IS THE WORK OF TWO PEOPLE IN MY LAB THAT HAS PERFORMED THIS HEROIC EFFORT OF CLONAL ANALYSIS. SO BASICALLY THE SENSITIVITY WITH THIS APPROACH IS ACTUALLY PRETTY HIGH AND THE FALSE RATE IS ACTUALLY NOT -- ONLY 5 TO 10%. SO THESE ARE THE DATA. WE DISCOVERED THAT ACROSS HUMAN PRENATAL DEVELOPMENT, EACH PRO GENITAL -- OF VARIOUS REGIONS, ACCUMULATES ABOUT BETWEEN 200 AND 500 SINGLE NUCLEOTIDE VARIATION. AND WE CALCULATED THE RATE OF ABOUT FIVE SMBs PER CELL PER DAY. THERE IS LITTLE DIFFERENCE BETWEEN VARIOUS -- SUBSTITUTION TYPE THAT WE NOTICED. NOW THE MOST REMARKABLE THING, AND I'M SKIPPING VARIOUS THINGS JUST TO GIVE YOU AP IDEA OF WHAT YOU CAN GET FROM THIS TYPE OF ANALYSIS, IS THAT WE OF COURSE HAD TO REGENOTYPE THESE VARIANTS TO VALIDATE THEM IN THEIR ORIGINAL TISSUES, INCLUDING THE SPLEEN, AND BY RESEQUENCING THEM, AND WE COULD RESEQUENCE OBVIOUSLY ONLY THE HIGHEST FREQUENCY, LIKE BETWEEN .3 AND 30% ALLELE FREQUENCY, WE FOUND TREMENDOUS OVERLAP NOT ONLY BETWEEN VARIOUS REGIONS OF THE BRAIN THAT YOU CAN SEE HERE, FRONTAL CORE TECH, BASAL GANGLIA, MANY ACTUALLY SHAI BUT ALSO BETWEEN THE BRAIN TISSUE AND THE SPLEEN. YOU CAN SEE HERE. SO THERE IS A HIGH NUMBER OF -- THAT ARE ACTUALLY SHARED, SUGGESTING THAT BASICALLY THESE MUTATIONS ARISE VERY EARLY BEFORE THE SPLITTING OF THE SPLEEN -- THESE ARE THE STAGES I'M TALKING ABOUT. SO THIS COLUMN, SO TO SPEAK, SHARED SOMATIC MUTATION HAVE AN ALLELE FREQUENCY OF -- WHICH IS EXACTLY WHAT YOU WOULD EXPECT IF YOU CALCULATE ABOUT 12 CELL DIVISIONS PRIOR TO -- WITH AN EXPECTED ALLELE FREQUENCY OF .03 TO 25% IN SOMATIC TISSUES. SO WE'RE LOOKING AT A VERY SURPRISING PHENOMENON THAT IS PERHAPS NOT SO SURPRISING BUT ALL THE SOMATIC MUTATIONS IN THE BRAIN ABOVE 2% ALLELE FREQUENCY ARE ACTUALLY SHARED WITH THE SPLEEN AND, THEREFORE, OCCURRING VERY EARLY. NOW WHEN WE CLUSTERED THE MUTATIONS WE DISCOVERED BY ALLELE FREQUENCY AND CONSIDERED THE SHARED DISTRIBUTION AMONG THE CLONES WHICH ARE HERE, WE FOUND THAT THE MOST FREQUENT MUTATIONS DOWN HERE ON THE LEFT ARE THE ONES THAT OF COURSE ARE MORE SHARED BETWEEN THE CLONES IN THE BRAIN, AND BY COMBINING THESE TWO PHENOMENON, WE COULD ACTUALLY RECOUP STRUCT THE EARLY THREE OF THE FIRST FIVE MITOTIC DIVISION OF THE HUMAN EMBRYO. SO THESE ARE A TOTAL OF 84 MUTATIONS, THERE'S TYPICAL ALLELE FREQUENCIES OF 1% IN NEURONAL TISSUE, AND YOU CAN SEE THEY'RE DISTRIBUTED ACROSS THIS VERY FIRST MITOTIC DIVISION, ALLOWING US TO CALCULATE A MUTATION RATE OF ABOUT 1.3 -- PER DIVISION PER CELL. I'M GOING TO SKIP THIS FOR LACK OF TIME, BUT JUST TO SAY THAT THE MUTATIONS -- PROBABLY DIFFERENT MECHANISMS ACROSS DEVELOPMENT. I'D LIKE TO SUMMARIZE BY SAYING THAT IN NORMAL DEVELOPMENT, OUR DATA SUGGESTS THAT THERE ARE ABOUT 100 TO 400 MOSAIC -- PER CELL, THAT THE FREQUENCY AND CLOAALITY OF THESE MUTATIONS ALLOW TO RECONSTRUCT -- THE FACT THAT THEY ARE MOSTLY COMING AT LEAST THOSE ABOVE 1% AT -- 2 SNV PER DIVISION PER SELL CELL, AND THERE IS A SHIFT IN SPECTRUM DURING DEVELOPMENT AND FINALLY, ONE SURPRISING FACT IS THAT EMBRYONIC NEURON PROGENITORS ACCUMULATE 5 SNV PER SELL PER DAY, WHICH IS ORDERS OF MAGNITUDE HIGHER IN ADULT SOMATIC CELL -- ADULT GERMLINE, SUGGESTING THAT THE NEUROGENIC PERIOD IS A HIGHLY MUTAGENIC PERIOD OF HUMAN DEVELOPMENT. AND WITH THAT, BASICALLY OUR CONCLUSION SO FAR IS THAT POSSIBLY THIS PHENOMENON IS ACTUALLY QUITE A NORMAL FI CAL -- WE DON'T KNOW YET THE IMPLICATION FOR THE DISORDER BUT SO FAR THIS MUTATION SEEMS TO BE DISTRIBUTED THROUGHOUT THE GENOME WITH NO PARTICULAR PREFERENCE. IT CAN AFFECT, HOWEVER, THE MANIFESTATION OF OTHER GENOMIC VARIANTS LIKELY, AND CAN BE SELECTED FOR, CAN BE POSSIBLY DEPENDENT ON AGE, AND WE'RE HOPING TO DISCOVER THIS IMPLICATION FOR THESE ORDERS. AND WITH THAT, I'D LIKE TO THANK PEOPLE OF MY LAB AND I GIVE THE PODIUM TO CHRIS. [APPLAUSE] >> I'D ALSO LIKE TO THANK THE ORGANIZERS AND EVERYONE FOR A REALLY FUN MEETING, AND SO I WAS ASKED TO JUST TALK A LITTLE BIT ABOUT SOME MEDICAL MUTATIONS IN THE DISEASED BRAIN BUT ALSO TO INTRODUCE THE BRAIN SOMATIC MOSAIC NETWORK. SO THE MOST FAMILIAR DISEASE ASSOCIATED -- ARE FOCAL CORTICAL DYSPLASIAS. ONE EXAMPLE OF WHICH IS CALLED HEMIMEGLOENSELFLY, INTENSELY EPILEPTIC. THIS IS AN MRI OF A CHILD WHO HAD INTRACTABLE EPILEPSY THAT STARTED AT BIRTH. THERE WAS LEFT SIDED WEAKNESS BECAUSE OF THE RIGHT HEMISPHERIC DYSPLASIA. HIS SEIZURES WERE SO BAD AND INTRACTABLE THAT HIS ENTIRE RIGHT HEMISPHERE WAS REMOVED TO CONTROL HIS SEIZURES, AND THAT'S WHAT ALLOWED US TO FIND -- WHAT ALLOWED ANNE PODURI IN MY LAB TO FIND MUTATION IN THIS CHILD, AND THAT WAS ACTUALLY LIFE SAVING. HE HAD NO SEIZURES FOR SIX YEARS, HE LEARNED TO WALK, HE DEVELOPED FLUENT SPEECH, HE READS AT GRADE LEVEL, AND HE DOES HAVE PERSISTENT LEFT SIDED WEAKNESS BECAUSE HE DOESN'T HAVE THE RIGHT SIDE OF HIS BRAIN. HIS NAME IS DANTE. I USE HIS NAME AND PICTURE WITH HIS PARENTS' PERMISSION AND THERE'S A VIDEO ONLINE IF YOU'RE INTERESTED IN HEARING HIM SPEAK. SO HEMIMEG LA ENSELFLY -- FOCAL CORTICAL DYSPLASIA, BASICALLY THESE ARE NATURE'S WAY OF ANNOUNCING THAT SOMATIC MUTATIONS OCCUR IN THE BRAIN FAIRLY FREQUENTLY. YOU CAN SEE THESE LITTLE PIZZA SLICE SORT OF THINGS, TYPICALLY ONE OR TWO COUNCILMEMBERS CENTIMETERS IN S IZE. THEY ARE THE MOST COMMON CAUSE OF CHILDHOOD INTRACTABLE EPILEPSY THAT COMES TO SURGERY. ANY GOOD PEDIATRIC HOSPITAL WILL HAVE A NEUROSURGERY PROGRAM WHERE THEY TAKE THESE THINGS OUT BECAUSE WHEN THEY ARE IDENTIFIED RADIO DPRAFICALLY AND REMOVED, THEY CURE THE SEIZURES ABOUT HALF THE TIME AND ABOUT ANOTHER THIRD OF THE TIME THEY MAKE THE SEIZURES MUCH, MUCH BETTER. SO AS BEST WE CAN TELL, THE VAST MAJORITY OF THE CELLS CARRYING MUTE TAITION ARE TAITION THAT MUTATION AND YOU CAN CURE IT BY REMOVING THEM SO NEUROSURGEONS LOVE THEM. SO HEMIMEG LA ENSELFLY AS SHOWN BY SEVERAL LABS BOTH CONTAIN SPONTANEOUS, SOMATIC, GAIN OF FUNCTION MUTATIONS THAT ACTIVATE THE MTOR PATHWAY. AND THIS WAS SIMULTANEOUS WORK FROM OUR LAB AND OTHERS, MANY DIFFERENT LABS SHOWED THAT THESE SOMATIC GAIN OF FUNCTION MUTATIONS IN SEVERAL STEPS TO THE MTOR PATHWAY, IN THEY OCCUR IN YOUR BRAIN, THEY GIVE YOU FOCAL DYSPLASIA. THEY OCCUR IN YOUR SHOULDER, YOU GET OVERGROWTH OF YOUR SHOULDER, THESE ARE OVERGROWTH GENES WHICH IS WHY HALF OF THAT KID'S BRAIN WAS OVERGROWN, YOU CAN GET GAID OF FUNCTION MUTATIONS IN AKT3 OR AKT1, GAIN OF FUNCTION MUTATIONS IN MTOR ITSELF OR IN NEGATIVE REGULATORS. SO HERE'S AN EXAMPLE. IT'S REALLY NOT JUST A MUTATION BUT IT'S THE TIME AND PLACE OF THE MUTATION WHICH IS KEY, AND THIS IS SHOWN BY THESE THREE PATIENTS. THIS PATIENT HAS A FOCAL CORTICAL DYSPLASIA OF THE PARIETO-OCCIPITAL REGION. WHEN THAT WAS REMOVED, THE STUDY OF THE TISSUES SHOW THERE WAS A MISSENSE MUTATION IN MTOR ITSELF ITSELF, IT WAS ONLY PRENLT PRESENT IN 3% OF THE CELLS, NOT IN THE LESION, NOT DETECTABLE ANYWHERE ELSE IN THE BODY WHERE THEY LOOKED. NOT IN THE SALIVA. HERE'S A FRONTAL FCD, SAME MUTATION IN ABOUT 3% OF THE CELLS WITHIN THAT LESION. THE EXACT SAME MUTATION NOW PRESENT AT ABOUT 20% OF THE CELLS, BUT STILL NOT PRESENT IN THE BLOOD. SO THESE EPILEPTIC DISORDERS ARE SOMATIC MOSAIC MUTATIONS THAT ARE NOT ASSAYABLE IN THE BLOOD. THEY ARE SO TOXIC AND TOXIC TO CELLS, THAT THEY NEVER PRESENT AS GERMLINE MUTATIONS BECAUSE THEY WOULD BE LETHAL TO THE EMBRYO. SO THEY CAN ONLY BE STUDIED IN THIS CASE BY STUDYING THE BRAIN. NOW OTHER SORTS OF SOMATIC MOSAIC MUTATIONS THAT CAN CAUSE EPILEPSY OR THAT ARE SOMEWHAT MILDER OR LESS TOXIC ACTUALLY CAN BE TOLERATED IN THE BLOOD SO THEY MIGHT BE GERMLINE OR THEY MIGHT BE SOMATIC MOSAIC MUTATIONS DETECTABLE IN A FRACTION OF THE CELLS IN THE BLOOD. SO A COUPLE OF MEMBERS OF THE CONSORTIUM DECIDED TO TEST WHETHER SOME FRACTION OF AUTISM SPECTRUM DISORDERS MIGHT REFLECT SOMATIC MOSAIC MUTATIONS. WE WERE PARTICULARLY INTERESTED IN THE IDEA THAT. SO MILDER FORMS OF AUTISM MIGHT HAVE AP AUTISM-TYPE MUTATION BUT MAYBE ONLY IN PART OF THE BRAIN. OBVIOUSLY AUTISM REFLECT A LARGE CONTRIBUTION OF DE NOVO VARIANTS TO A FRACTION OF THE CASES, ALTHOUGH DE NOVOS ONLY ACCOUNT FOR PERHAPS A QUARTER OR SO OF ALL OF THE CASES. SO NORTH TO TEST THIS HYPOTHESIS HYPOTHESIS, KINDLY CAME TO ME EXTREMELY WELL TRAINED, WORKED WITH ANOTHER TEAM SCIENCE CONSORTIUM, THE AUTISM SEQUENCING CONSORTIUM, THAT WHICH YOU'VE ALREADY HEARD. THANKS TO JOE AND MARK AND COLLABORATORS IN THAT CONSORTIUM, SHE WAS ABLE TO ACCESS ALMOST 5,000 TRIOS, SORRY, ALMOST 5,000 CASES, MANY OF THEM TRIOS, WITH AUTISM SPECTRUM DISORDERS, AND THEN WAS ABLE TO RECALL THE DE NOVO VARIANTS TO DETERMINE WHETHER SOME FRACTION OF THEM MIGHT BE SOMATIC KNOW SAY, MAY NOT BE GERMLINE MUTATIONS THAT OCCURRED IN THE SPERM OR THE EGG BUT MIGHT INSTEAD BE MUTATIONS THAT OCCUR AFTER FERTILIZATION. SO FIRST YOU JUST PLOTTED THE FRACTION OF ALL THE DE NOVO VARIANTS, AND WHAT YOU CAN SEE IS THERE'S A FRACTION OF THEM HERE, WHICH HAVE A LOCAL -- HETEROZYGOUS MUTATION BUT IN NOT ALL OF THE CELLS, BUT ONLY IN PERHAPS HALF OF THE CELLS. SO THE ADVANTAGES OF THIS STUDY IS THAT SHE COULD ANALYZE A LARGE NUMBER OF FAMILIES BECAUSE WE DID NOT THINK THAT SOMATIC MOSAIC MUTATIONS WERE GOING TO BE A LARGE FRACTION OF THE DISEASE BUT THEY MIGHT BE A SMALL FRACTION OF IT. WE ALSO HAD A TRIO DESIGNED IN MANY CASES. WE ARE PRIMARY BLOOD RATHER THAN CELL LINES BECAUSE CELL LINES WOULD NOT BE RELIABLE TO LOOK FOR SOMATIC MOSAIC MUTATIONS. UNFORTUNATELY, THOUGH, WE HAD RELATIVELY LOW COVERAGE, ANYWHERE FROM 50 TO 90X, SO THIS WOULD BE VERY INSENSITIVE, SO WE COULD ONLY DETECT THEM IN A QUARTER OR SO OF THE CELLS. SO AGAIN, SHE CERTAINLY FOUND THAT CERTAINLY A FRACTION OF THE DE NOVOS LOOKED LIKE THEY HAD A LOWER ALLELE FRACTION. SHE BROKE THESE OUT INTO THREE CATEGORIES, GROUP A WAS ALL DE NOVOS, GROUP B WAS THOSE DE NOVO ALLELE FREQUENCY AND C WAS THE SUBSET OF THE GROUP B THAT HAD A SEE VEER IMBALANCE THAT WAS SIGNIFICANTLY SIGNIFICANT. SHE ACTUALLY RECALLED ALL THE EXOMES, LOWERED THE THRESHOLD TO CAPTURE THESE POST ZYGOTIC MUTATIONS BUT INCREASE THE SPECIFICITY OF THE ALGORITHM BY SCREENING DEET KNOW VOS THAT WERE PRESENT IN NORMAL INDIVIDUALS. AND THEN SHE DID EXTENSIVE VALIDATION USING THREE DIFFERENT VALIDATION METHODS BY SUBCLONING THEM, USING TARGETED SEQUENCING PANEL AND FOUND THAT ABOUT 90% OF THESE MOSAIC VARIANTS VALIDATE AS MOSAIC VARIANTS BY MULTIPLE TECHNOLOGIES. INTO YOU SO IN FACT, REMARKABLY, THE VAST MAJORITY OF THESE MOSAIC VARIANTS HAD NOT BEEN CALLED BY PREVIOUS CALLING ALGORITHMS, THESE WOULD BE NOVEL ONES AS OPPOSED TO THOSE THAT WERE PUBLISHED. AND OVERALL, ABOUT 7% OF THE DE NOVO MUTATION APPEARED TO BE -- THESE EARLY POST ZYGOTIC MUTATIONS. MOST OF THEM NOT DETECTED BY PREVIOUS ALGORITHMS. SOME ARE KNOWN AUTISM GENES, OTHERS ARE KNOWN TO HAVE SEVERE NEUROLOGICAL PHENOTYPES LIKE HNRNPU WHICH CAUSES SEVERE ENCEPHALOPATHY, OR ONE WHICH CAUSES INTELLECTUAL DISABILITY. AGAIN, WE COORDINATED OUR WORK, SUBMITTED AN ANALYSIS OF A SMALLER NUMBER OF EXOMES BUT BEFORE OUR WORK CAME OUT, AND OUR SENSE IS THAT OUR DATA ARE VERY MUCH IN LINE WITH EACH OTHER. AND BECAUSE OF THE LARGE NUMBER OF EXOMES, WE HAD RECURRENT MOSAIC MUTATIONS IN THESE LIST OF GENES, MOACH MOST OF WHICH AS YOU CAN SEE ARE EXPRESSED IN THE BRAIN, EACH WITH THIS LARGE NUMBER OF FAMILIES, THE FRACTION OF THE MUTATIONS THAT ARE MOSAIC IS RELATIVELY LOW SO WE STILL HAVE RELATIVELY SMALL NUMBERS& BUT WE ARE ABLE TO ACTUALLY HAVE FAIRLY STRONG EVIDENCE TO ACTUALLY IMPLICATE SOME OF THESE AS CONTRIBUTORY. SO JUST TO SUMMARIZE, A -- ARE SOMATIC OR MOSAIC. WE'RE QUITE CONFIDENT THAT IF THE EXOMES ARE SEQUENCED AT HIGHER COVERAGE, WE WOULD BE ABLE TO IDENTIFY A LARGER NUMBER OF THEM. WE HAVE ONLY A VERY SMALL NUMBER OF PATIENTS BUT ELAINE ALSO ASKED FOR THOSE PATIENTS THAT HAVE MOSAIC MUTATIONS IN THE RECURRENT GENES, IS THERE I.Q. ON AVERAGE HIGHER THAN THAT WHERE THERE ARE GERMLINE MUTATIONS, IN FACT AGAIN THE NUMBERS ARE SMALL BUT THE MOSAIC MUTATIONS DO SEEM TO BE ASSOCIATED WITH HIGHER IQ IN GERMLINE DE NOVO MUTATIONS. SO IT MIGHT BE AN INTERESTING MECHANISM TO EXPLAIN HIGH FUNCTIONING YOU A TIS TICS. TO DETERMINE WHAT RELEVANCE THEY MIGHT HAVE FOR OTHER PSYCHIATRIC ILLNESSES. I'LL JUST TAKE TWO MINUTES, I THINK I'VE STILL GOT TWO MINUTES, THE TO EXPLORE THE RANGE OF MUTATION IN SINGLE NEURONS IN THE BRAIN, AND SO THIS ENTAILS AS YOU'VE HEARD SORTING SINGLE NEURONS, LICING THEM IN COLD CONDITION, AMPLIFYING THE GENOME AND SEQUENCING IT. YOU'VE HEARD ABOUT THE ANALYSIS DONE IN PARALLEL WITH OUR LAB AND RUSTY'S LAB. SO MIKE AND MOLLY SEQUENCED THE GENOMES OF A COUPLE OF DOZEN SINGLE NEURONS FROM THREE DIFFERENT RELATIVELY YOUNG HEALTHY PEOPLE AND FOUND AND ESTIMATED THAT THERE ARE ABOUT 1400 SOMATIC -- PER NEURONAL GENOME. THIS ANALYSIS IS DONE BY ALICE LEE AND PETER PARK AND THERE ARE ABOUT 1700 -- BUT THEN ADJUSTED FOR COVERAGE AND FALSE POSITIVE, THE ESTIMATES RANGE ANYWHERE FROM 800 TO 2,000. WHICH WOULD BE A DOZEN OR TWO PER EXOME. THE VAST MAJORITY OF THESE ARE FOUND IN JUST A SINGLE NEURON. OCCASIONALLY, THEY ARE CLOAM, AS CLONAL, NOT PRESENT IN THOSE THREE NEURONS OR THE HEART OF THE SAME PATIENT. HERE'S A SINGLE NEURON SNV, HERE'S A CLONAL ONE SHARED BY TWO NEURONS, HERE'S ONE SHARED BY THREE NEURONS. SO THE CLONAL REPRESENT THE PERMANENT LINEAGE MAP OF ANY HUMAN BRAIN AND WE CAN ACTUALLY IN PRINCIPLE COMPARE THE LINEAGE MAPS BETWEEN DIFFERENT POSTMORTEM BRAIN, BUT THEN THE COUNTS IN THE SINGLE CELLS ALSO GIVE US A SENSE OF WHAT'S GOING ON IN THAT GENOME OF THAT SINGLE CELL, AND IN WORK THAT WE'RE NOW GETTING READY TO PUBLISH, WE FIND THAT IN NEWBORNS, THE NUMBERS ARE LOWER AND ACTUALLY START TO DOVETAIL WITH FLUORAS, AND ONE DISEASE, WHERE YOU ACTUALLY HAVE PREMATURE AGING SO THESE KIDS HAVE GRAY HAIR BY THE TIME THEY'RE 5 OR 10 YEARS OLD AND THEY TYPICALLY DIE BY THE TIME THEY'RE IN THEIR TEENS, THIS IS KNOWN TO BE GENETIC DISORDERS TO REPAIR THE DNA DAMAGE THAT OCCURS WITH TRANSCRIPTION. THEY GET INTELLECTUAL DISABILITY AND PREMATURE NEUROLOGIC DEGENERATION. WHAT WE FIND IS, IN FACT, THE COUNTS OF SNVs ARE ABOUT THREE TIMES HIGHER THAN IN NORMAL NEURONS, SUGGESTING THAT WE MIGHT BE ABLE TO USE THE SINGLE CELL GENOMICS AS A WAY OF LOOKING AT AGING AND OTHER FORMS OF NEURO DEGENERATION. SO THAT'S JUST AGAIN AN INTRODUCTION TO WHAT THE NETWORK IS GOING TO TRY TO WORK OUT. WE'RE LOOKING AT CLONAL SOMATIC MUTATIONS AS A POSSIBLE HIDDEN FORM OF CERTAIN BRAIN DISEASES, AND EACH GROUP IS LOOKING AT DIFFERENT DISEASES, AND THEN ALSO ANALYZING THE GENOME OF A SINGLE CELL TO SEE IF THAT MIGHT CONTRIBUTE TO PERHAPS THE PENETRANCE OF DISEASES. THIS IS JUST THE PEOPLE IN MY LAB WHO DID THE WORK AND AGAIN, WE'RE VERY GRATEFUL TO THE AUTISM SEQUENCING CONSORTIUM AND ALSO THE BRAIN SOMATIC MOSAICISM NETWORK. THANKS VERY MUCH. [APPLAUSE] >> THANK YOU FOR THOSE REALLY FASCINATING TALKS. THERE'S CLEARLY MANY WAYS TO SKIN A CAT OR TO GET A DISEASE, AND I THINK THE DEEPER WE LOOK, THE MORE COMPLEXITY WE SEE. THE NEXT TALK IS FROM STEVE MCCARROLL, SINGLE CELL EXPRESSION APPROACHES. STEECH.STEVE. >> SO 800 YEARS AGO, WE THIS ALL FIGURED OUT. THERE WAS AN AMBITIOUS SYNTHETIC THEORY OF ALL DISEASE, AND THAT THEORY HELD THE BODY WAS A BAG OF FLUIDS, AND THOSE FLUIDS WERE SUPPOSED TO EXIST IN CERTAIN NATURAL PROPORTIONS. -- AROSE IF THEY DRIFTED OUT OF THEIR NATURAL PROPORTIONS. IF YOU HAD TWO MUCH BLACK BILE, YOU WERE MELANCHOLIC, IF YOU HAD TOO MUCH BLOOD, WERE, YOU WERE SAN GYN. ALONG CAME THE 20ITY CENTURY. AND THAT WAS REPLACED BY A MUCH MORE SCIENTIFIC THEORY, KNOWN AS THE CHEMICAL IMBALANCE THEORY. THAT THEORY HELD THAT BRAIN WAS A BAG OF NEUROTRANSMITTERS THAT WERE SUPPOSED TO EXIST IN CERTAIN NATURAL PROPORTION. YOU HAD TO HAVE ENOUGH DOPAMINE TO BE MOTIVATED BUT NOT TOO MUCH OR YOU MIGHT DEVELOP SCHIZOPHRENIA. ANY NEUROSCIENTIST WILL TELL YOU THAT THE BRAIN IS NOT A SINGLE COMPARTMENT OF CHEMICALS. THE BRAIN CONTAINS A VAST NUMBER OF HIGHLY SPECIALIZED CELLS OF HUNDREDS IF NOT THOUSANDS OF DIFFERENT TYPES, AND THESE CELLS VARY GREATLY IN SIZE AND SHAPE AND BIOLOGICAL MISSION. THEY USE THE SAME GENOME IN PROFOUNDLY DIFFERENT WAYS, AND IT SHOULD BE CLEAR THAT THIS IS THE LEVEL AT WHICH WE NEED TO WORK TO UNDERSTAND HOW GENES AND ALLELES GIVE RISE TO NORMAL FUNCTION AND TO ILLNESS. THE PROBLEM HAS BEEN THAT UNTIL RECENTLY, IF YOU WANTED TO SYSTEMATICALLY MEASURE THE LEVEL OF EXPRESSION OF AN ENTIRE CLASS OF ANALYTES, FOR EXAMPLE, ALL OF THE RNAs, YOU NEEDED TO FIRST DO IT BY TAKING THIS WONDERFUL BIOLOGICAL COMPLEXITY AND ESSENTIALLY MAKING IT INTO A SMOOTHIE. WHILE IT'S A GREAT IMPROVEMENT TO BE ABLE TO MEASURE 20,000 THINGS RATHER THAN TO MEASURE FOUR THINGS, IT'S STILL THE 20,000 -- THE WORK REALLY STARTED -- ANALYZING COMPLEX TISSUES LIKE THE BRAIN IN A WAY THAT MIGHT BE MUCH MORE LIKE A FRUIT FREUT SALAD IN WHICH YOU COULD ANALYZE THE COMPOSITION OF EVERY INDIVIDUAL COMPONENT. THIS WAS WORK BY A TEAM OF PEOPLE THAT WAS LED BY EVAN MACO SKO, A POSTDOC IN THE LOB WHO'S ALSO A PSYCHIATRIST. SO OUR APPROACH TO THIS PROBLEM, AND WE REALIZE THAT TO ADDRESS THE COMPLEXITY OF THE BRAIN, IT WAS GOING TO BE NECESSARY TO ANALYZE VAST NUMBERS OF INDIVIDUAL CELLS, THOUSANDS, TENS OF THOUSANDS, IN WAYS THAT WERE EFFICIENT AND AFFORDABLE, THAT NEITHER BROKE THE BANK NOR BROKE THE SCIENTIST. AND THE WAY THAT WE DEVELOPED TO DO THAT WAS TO USE TINY ACQUIESCE DROBLETS DROPLETS TO COMPARTMENTALIZE VAST NUMBERS OF CELLS EACH INTO THEIR OWN TINY LITTLE COMPARTMENT FOR ANALYSIS. THE WONDERFUL THING ABOUT DROPLETS ARE TINY. LARGER THAN CELLS BUT YET THEY'RE VERY SMALL. YOU CAN FIT MILLIONS OF THEM IN A TUBE AND THEY HAVE CAPACITY TO MOLECULARRIZE REACTIONS. IT ONLY TOOK US A COUPLE OF WEEKS TO FIGURE OUT HOW TO GET CELLS INTO DROPLETS. WHAT TOOK US A COUPLE OF YEARS WAS A MUCH HARDER CHALLENGE, TO DEVELOP A WAY TO REMEMBER WHICH RNA MOLECULE CAME FROM WHICH CELL IN WHICH DROPLET. WE ULTIMATELY CAME UP WITH A WAY OF DOING THAT BY USING BEADS TO DELIVER DNA BAR CODES TO DROPLETS, THE IDEA BEING THAT CELLS -- WE'D COENCAPSULATE CELLS WITH BEADS AND DROPLETS. EVERY BEAD DELIVERS A PAYLOAD OF PRIMERS FOR REVERSE TRANSCRIPTION THAT HAVE A MOLECULAR BAR CODE ON THEM AND THEN WE LYSE THE CELL SO THE RNAs CAN ASSOCIATE WITH THE BEAD. OF COURSE TO BE ABLE TO DO THIS, WE NEEDED TO BE ABLE TO MAKE MILLIONS OF BEADS, EACH COATED WITH A DISTINCT BAR CODE SEQUENCE AND WE CAME UP WITH A WAY TO DO THAT BY DOING SPLIT AND POOL SYNTHESIS, WHERE WE SINT SEISE MILLIONS OF -- SO EACH BEAD IS LIKE A HAIRY BALL, AND AT CERTAIN STAGES IN THIS SYNTHESIS, WE SPLIT -- I'M SORE EE WE SPLIT THE BEADS INTO GROUPS AND ADD A DIFFERENT DNA BASE TO EACH OF THE GROUPS. BY DOING THIS ACROSS MANY CYCLE CYCLES, WE CREATE A VAST NUMBER OF DIFFERENT BEADS, EACH OF WHICH IS CODED WITH A UNIFORM BAR CODE WHICH -- UNIQUE PATH THROUGH THE SEER EAVES SYNTHESIS REACTIONS. THIS IS A MICRO FLUIDIC DEVICE THAT OUR COLLEAGUE PHYSICIST DEVELOPED TO BE ABLE TO ENCAPSULATE CELLS IN BEADS AND DROPLETS. THIS HAS GREATLY SLOWED DOWN. HERE YOU'RE WATCHING THE CELLS ARE COMING THROUGH, THIS CHANNEL, THE BEADS ARE FLOWING, IN THIS WAY WITH THE LYSIS BUFFER, THEY MEET AND PINCH OFF INTO A SERIES OF DISCRETE DROPLETS. THIS MOVIE WAS GREATLY SLOWED DOWN. THIS MOVIE IS A LITTLE BIT LESS SLOWED DOWN. HERE YOU CAN SEE THE BEAD FLOWING IN THIS WAY AND THEN CO-ENCAPSULATING A CELL WHICH IS HARDER TO SEE, BUT IF YOU LOOK CAREFULLY, YOU'LL SEE IT KIND OF COME ACROSS THIS ELBOW AND THEN FLOW IN TOGETHER WITH THE BEAD. AND COENCAPSULATEING WITHIN THE BEAD. THAT MOVIE TOO WAS GREATLY SLOWED DOWN. THE NEXT MOVIE IS NOT SLOWED DOWN BECAUSE MELISSA GOLDMAN MADE THIS WITH HER I-PHONE CAMERA. THIS IS A MAP TO PC TRANSITION PROBLEM. SO HERE SHE'S JUST -- SO THAT'S THE MICRO FLUIDIC DEVICE ON THE SURFACE OF THE SLIDE, SHE'S JUST HOLDING HER PHONE UP TO THE EYEPIECE OF THE CAMERA AND IF YOU LOOK HERE, YOU CAN SEE THE SPEED AT WHICH DROPLETS ARE ACTUALLY CREATED. IT'S POSSIBLE TO CREATE AN ENORMOUS NUMBER OF THEM EVERY SECOND BECAUSE THEIR VOLUME IS SO SMALL INDIVIDUALLY, THAT THEY HAVE THIS ENORMOUS CAPACITY TO PARALLELIZE AND TO USE EXPENSIVE REAGENTS VERY EFFICIENTLY. SO THERE WERE BIG QUESTIONS AROUND THE EXTENT TO WHICH YOU COULD CLASSIFY CELLS INTO TYPES& AND LEARN ABOUT THEIR TYPES AND STATES BY THESE KIND OF LIGHTLY SAMPLED TRANSCRIPTOMES. THE ANSWR IS THAT YOU CAN, WE FOUND, AS LONG AS YOU COLLECT A VERY LARGE NUMBER OF THEM. THIS IS ACTUALLY A MACHINE LEARNING APPROACH THAT WAS DEVELOPED BY OUR COLLEAGUE IN WHICH YOU KIND OF THROW THE CELLS INTO A PARALLEL UNIVERSE IN WHICH CELLS ARE GRAVITATIONALLY ATTRACTED TO OTHER CELLS IN PROPORTION TO THE SIMILARITY OF THEIR TRANSCRIPTIONAL PROFILES. AND YOU CAN SEE WITHOUT ANY PRIOR INSTRUCTION OR INFORMATION ABOUT MARKERS OR ANYTHING, THE CELLS BASED ON THEIR GLOBAL TRANSCRIPTIONAL PROFILES JUST SELF-ORGANIZE INTO DOZENS OF DISCRETE GROUPS, EACH OF WHICH WE FOUND CORRESPONDS TO A DIFFERENT CELL TIME, IN THIS CASE, OF THE MOUSE RETINA. SO THIS IS A SYMPOSIUM DEDICATED TO TEAM SCIENCE, THE VERY MUCH AN INTERDISCIPLINARY TEAM IN WHICH PEOPLE CAME TOGETHER WITH VERY DIFFERENT AREAS OF EXPERTISE FROM MOLECULAR BIOLOGY TO ENGINEERING AND OPTIMIZATION, SO PHYSICS, TO ALGORITHMS AND DATA SCIENCE AND MACHINE LEARNING. PEOPLE WOULD SAY IF YOU ACTUALLY GET THAT TO WORK, ISN'T THAT GOING TO BE GREAT BECAUSE YOUR LAB WILL HAVE THIS SECRET WEAPON AND YOU'LL GET GRANTS FOR A TECHNIQUE OFF OF THIS TECHNIQUE YOU'LL ONLY BE ABLE TO DO. FRANKLY IT NEVER OCCURRED TO US TO DO THAT. WE SPENT YEARS PARTICIPATING IN THIS COMMUNITY WITH ALL THESE -- IT SEEMED LIKE THE MOST OBVIOUS THING IN THE WORLD THAT OF COURSE YOU WANT TO BE IN A HIGHLY ENABLED THRIVING FIELD IN WHICH EVERYONE IS GENERATING LOTS OF EXCITING DATA AND SHARING IT EX-EUB RAPTLY, SO BEFORE WE EVEN PUBLISHED THE PAPER, WE PREPARED DETAILED PROTOCOLS, VIDEOS AND BLUEPRINTS TO ZO THAT SO THAT LABS COULD BUILD THEIR OWN. THE 30 PAGE EXPERIMENTAL PROTOCOL HAS BEEN DOWNLOADED 40,000 TIMES WITH 35,000 OF THOSE DOWNLOADS COMING FROM OUTSIDE OF RUSSIA. [LAUGHTER] THE ANALYSIS SOFTWARE WE WROTE HAS BEEN DOWNLOADED 24,000 TIMES. THERE'S AN OP LINE DISCUSSION FORM IN WHICH 700 PEOPLE PARTICIPATE WITH MANY POSTS PER WEEK. AND FROM THIS, WE'VE ACTUALLY -- WE'VE MADE FRIENDS ALL OVER THE WORLD AND IT'S ACTUALLY CONNECTED US TO MANY INTERESTING AREAS OF SCIENCE. PEOPLE HAVE DONE ALL KINDS OF CRAZY THINGS WITH THIS TECHNOLOGY THAT WE NEVER ENVISIONED AND USED IT TO ANALYZE AND STUDY ALL KINDS OF BIOLOGICAL QUESTIONS. SO ONE OF THE THINGS THAT THIS KIND OF APPROACH HAS MADE IT NEWLY POSSIBLE DO IS TO ACTUALLY NOW WERE ADDRESS WHAT CELL POPULATIONS ARE PRESENT IN EVERY TISSUE AND WHAT GENES EACH CELL POPULATION EXPRESSES. TO PROFILE ACTUALLY 10 REGIONS OF THE ADULT MOUSE BRAIN AND IDENTIFY EACH OF THE PA TERPS OF GENE EXPRESSION THIS IS ACTUALLY FROM A STUDY THAT WE JUST DID TOGETHER WITH OUR COLLEAGUE THAT WAS JUST PUBLISHED A FEW WEEKS AGO IN WHICH DROP SEQ REVEALED SOMETHING VERY SURPRISING ABOUT THE CEREBRAL ORGANOID SYSTEMS THAT YOU'VE ALSO HEARD ABOUT EARLIER TODAY. IN PARTICULAR, WE FOUND THAT BY PROFILING 80,000 CELLS FROM A COUPLE CEREBROORGANOIDS, IN ADDITION TO THE EXPECTED -- WE FOUND THEY WERE ALSO CREATING ALL OF THE MAJOR CELL CLASSES OF THE RETINA, AND IN FACT IT TURNS OUT THAT CEREBRAL ORGANOIDS HAVE THE CELLULAR COMPONENTS OF A SENSORY NERVOUS SYSTEM. GEORGIA FOUND THAT SHE COULD SHINE LIGHT ON CEREBRAL ORGANOIDS AND THEY RESPONDED, SOME CELLS WITH SPIKE TRAINS AND OTHER KINDS OF RESPONSES AND THAT IN FACT LIGHT INDUCES THE EXPRESSION OF -- GENES. SO THIS KIND OF SYSTEMATIC UNBIASED ANALYSIS THAT'S NOT CONSTRAINED TO ANY PARTICULAR PRIOR HYPOTHESIS ABOUT WHAT'S GOING TO BE PRESENT IN TISSUE CAN LEAD YOU TO SORT OF SURPRISING INSIGHTS THAT ENABLE NEW KINDS OF SCIENTIFIC QUESTIONS TO BE ADDRESSED IN EXPERIMENTAL SYSTEMS. ONE MORE SLIDE ON RECENT WORK, THIS HAS ACTUALLY WORKED TOGETHER WITH EVAN AND BRAD AND -- AND IT'S JUST REALLY TO ADDRESS THIS QUESTION OF COULD WE TAKE THE POLYGENIC KINDS OF RESULTS THAT COME OUT OF HUMAN GENETICS TODAY AND REALLY USE THAT AS THE STRENGTH TO TRY TO IMPLICATE SPECIFIC CELL TYPES FOR THE PHENOTYPES THAT STUDY. IN PARTICULAR, ONE OF THE MORE PUZZLING THINGS TO COME OUT OF GENOME WIDE ASSOCIATION STUDIES FOR BODY WEIGHT AND OBESITY WHICH IMPLICATE HUNDREDS OF GENES IS A VERY LARGE FRACTION ARE EXPRESSED PREDOMINANTLY IN THE CENTRAL NERVOUS SYSTEM, SUGGESTING ON SOME LEVEL, SOME ASPECT OF OBESITY MIGHT BE A STEALTH PSYCHIATRIC ILLNESS, BUT OF COURSE RAISING THE QUESTION OF WHAT PART OF THE BRAIN WOULD MEDIATE THIS, AND HOW WOULD IT WORK. SO WE PROFILED THOUSANDS OF CELLS FROM THE HYPOTHALAMIC -- NUCLEUS WHICH IS KNOWN TO HAVE A POTENTABILITY TO REGULATE APPETITE, AND THROUGH THESE KINDS OF ANALYSIS, WE'RE ABLE TO IDENTIFY -- THESE ARE ACTUALLY JUST THE NEURONAL TYPES THAT WE IDENTIFIED IN THE HYPOTHALAMIC -- DOZENS OF NEURONAL TYPES, MANY OF WHICH WERE NOVEL. ONE OF THESE NOVEL TYPES IS PARTICULARLY INTERESTING BECAUSE IT TURNS OUT THAT THESE HUNDREDS OF GENETIC INFLUENCES THAT IT HAD COME UP IN OBESITY IN GWAS CONCENTRATE VERY SUBSTANTIALLY IN THIS SPECIFIC NEURONAL SUBTYPE. IN FACT OUR COLLEAGUE IN BRAD LOCAL'S LAB SHOW THESE CELLS ARE ACTUALLY DRIVING A RAPIDLY ACTING -- IT'S ACTUALLY PHARMOPOTENT THAN THE PALM C CELLS THAT HAD BEEN KIND OF HISTORICALLY THOUGHT OF AS SORT OF THIS -- THE IS A TIGHT SIGNAL INTO THIS CIRCUIT. SO HERE SORT OF SYSTEMATIC ANALYSIS OF TISSUE COMBINED WITH SYSTEMATIC GENOME WIDE UNBIASED ANALYSIS OF A PHENOTYPE WAS ABLE TO SYNERGIZE TO REALLY POINT TOWARD A CIRCUIT THAT MEDIATES AN EFFECT ON FEEDING AND APPETITE. SO THERE'S SO MANY THINGS THAT I HOPE WE AND THE WHOLE RESEARCH COMMUNITY WILL BE ABLE TO DO NOW GOING FORWARD IN TERMS OF FINALLY UNDERSTANDING WHAT IS THE CELLULAR DIVERSITY AND THE COMPLEX TISSUES THAT WE'RE INTERESTED IN, UNDERSTANDING PATHOLOGICAL STATES, WHAT'S DIFFERENT IN DISEASE STATES AND WHICH CELL TYPES IS IT DIFFERENT IN AND WHAT'S DIFFERENT ABOUT THEM. FINALLY TRYING TO MOVE IN A SYSTEMATIC WAY FROM GENETIC RESULTS TO MODELS OF PATHOPHYSIOLOGY IN WHICH WE CAN SAY THAT RISK COMES FROM INCREASED EXPRESSION OF THESE GENES AND THIS CELL TYPE AND REDUCED EXPRESSION OF THESE OTHER GENES -- REALLY START TO BUILD TOWARD MODELS OF PATHOPHYSIOLOGY. I WANT TO LEAVE YOU WITH ONE THOUGHT, WHAT GOT US THIS FAR? WHAT GOT US THIS FAR ARE GENE DISCOVERY APPROACHES THAT ARE SYSTEMATIC, UNBIASED AND INDUCTIVE. THEY'RE NOT LIMBED TO ANY SPECIFIC HYPOTHESIS OR THEORY OF DISEASE AND THEY GENERATE POWERFUL SYSTEMATIC DATASETS THAT ARE POWERFULLY CUMULATIVE AND CAN BE RE-USED FOR YEARS IN NOVEL AND UNEXPECTED WAYS BY A LARGE AND THRIVING RESEARCH COMMUNITY. WE REALLY HAVE TO WORK TO MAKE MORE OF BIOLOGY LIKE THE THINGS WE LOVE ABOUT GENOME SCALE GENETICS. WE NEED SIMILARLY POWERFUL WAYS TO REVEAL HOW GENES AND ALLELES SHIEP BIOLOGICAL SYSTEMS AND SHAPE DISEASE PROCESSES, AND I HOPE THAT DROP SEQ BECOMES JUST ONE OF MANY SORT OF NEW AND POWERFUL APPROACHES FOR BRINGING WHAT WE LOVE ABOUT GENOME SCALE GENETICS INTO A MUCH WIDER SET OF BIOLOGICAL QUESTIONS. THANK YOU VERY MUCH. [APPLAUSE] >> THANKS, STEVE. WE ARE RACING TO CATCH UP LEST EVERYWRUN MISSES THEIR PLAINS. NEXT WE HAVE ELISE ROBINSON, WHO IS GOING TO MOVE TO A DIFFERENT BUT VERY IMPORTANT TOPIC ON GENOTYPE TO PHENOTYPE DECONSTRUCTING NEUROPSYCHIATRIC DISORDERS WHICH IN SOME WAYS IS WHERE WE BEGAN THIS MORNING. >> THANK YOU. I'M GOING TO TALK ABOUT HOW WE'RE USING OFFICIAL PHENOTYPE DATA TO UNDERSTAND THE ASSOCIATIONS WE SEE COMING OUT OF THE LARGE CONSORTIUM STUDIES, IN BOTH RARE AND COMMON VARIANT SPACE. WE NOW HAVE MANY GENES AND TYPES OF GENETIC RISK FARES TO FACTORS WHAT'S BECOME CLEAR IS THAT THEY COME IN SEVERAL DIFFERENT FLAVORS. YOU HAVE SOME THAT HAVE A LOT OF OVERLAP WITH INTELLECTUAL DISABILITY IN TERMS OF THE PROBABILITY OF WHICH THEY'LL ALSO INCREASE RISK FOR GLOBAL DEVELOPMENTAL DELAY. THERE ARE OTHERS THAT ARE STRONGLY ASSOCIATED WITH ADHD OR EVEN HIGHER I.Q. SO IT'S BECOME CLEAR THAT IT'S NOT ADEQUATE TO SAY SOMETHING IS FOR EXAMPLE AN ASD OR SIZ RISK SCHIZOPHRENIA RISK FAK TO RISK FACTOR. A GIVEN RISK FACTOR RELATES TO DIFFERENT PHENOTYPIC OUTCOMES, MIGHT BE DOING OTHER THINGS DIFFERENTLY AS WELL, IT'S ENTIRELY PROBABLE. ANOTHER POTENTIAL PROBLEM IS THAT EVENTUALLY, WHEN WE HOPE TO DEVELOP TREATMENTS OR UNDERSTAND BUY LOMG CAL PATHWAYS BASED ON THESE ASSOCIATIONS, IT WOULD CERTAINLY BE BETTER TO HAVE A SENSE OF THE SPECTRUM OF THINGS THAT WE MIGHT ULTIMATELY END UP CHANGING BY ALTERING THAT PATHWAY. SO FOCUSING ON THE SPECIFIC EMPIRICAL DATA IN AUTISM, AS HAS COME UP A LOT TODAY, WE NOW HAVE TWO PRIMARY TOOLS TO USE IN GENETIC ASSOCIATION STUDY THAT COME FROM THE LARGE DISCOVERY PROJECTS. FIRST ARE GENETIC RISK SCORES, THE POLYGENIC RISK CREATES THE LARGEST AMOUNT OF GENETIC RISK FOR ASDs, DE NOVO VARIATION, WHICH IS A YES-NO VARIABLE, CAN HAVE A VERY STRONG IMPACT ON INDIVIDUAL CASES BUT CREATES LESS LIABILITY OF THE POPULATION LEVEL. DE NOVO VARIANTS IN ASD, SPECIFICALLY THOSE FROM ASD ASSOCIATED CLASSES ARE, ON AVERAGE, ALSO STRONGLY ASSOCIATED WITH OTHER INDICATORS OF GLOBAL NEURODEVELOPMENTAL IMPAIRMENT, INCLUDING INTELLECTUAL DISABILITY, SEIZURES, AND MOTOR DELAYS IN PARTICULAR. THIS FIGURE SHOWS THE RATE OF STRONG ACTING DE NOVO MUTATIONS IN ASD PRO BANS BASED ON THEIR COUNT OF THESE ADDITIONAL PHENOTYPES, THESE NEURODEVELOPMENTAL COMORBIDITIES. SO AMONG ASD CASES AMONG NONE OF -- YOU HAVE ABOUT A THREE FOLD INCREASE IN STRONGLY CONTRIBUTING -- MUTATIONS AGAINST THE RATE EXPECTED IN THE GENERAL POPULATION, ONCE YOU GET UP TO TWO OR THREE OF THOSE THINGS, YOU VASTLY INCREASE THE EXCESS MUTATIONS. WITHIN ASD CASES IS CONSISTENT& WITH WHAT YOU SEE WHEN YOU'RE COMPARING ASD AND ID IN TERMS OF THE PROFILE OF THE DE NOVO VARIANTS SEEN IN THOSE ASCERTAINED GROUPS. SO 99 GENES NOW HAVE THREE OR MOTHER DE NOVO TRUNK KATEED VARIANTS IN EITHER ASD OR ID. THE VAST MAJORITY INCLUDING THOSE THAT WERE NESHLY DISCOVERED IN ASD COLLECTIONS, PROTEIN TRUNCATING -- ARE MORE LIKELY TO BE SEEN IN I.D. HOWEVER THERE'S LIKELY A RANGE IN THE EXTENT TO WHICH CERTAIN GENES THAT PREDISPOSE ONE FOR RISK FOR ASD ARE ASD -- VERSUS PARTICULARLY RHODE I.D. ASSOCIATED. FOR EXAMPLE, ARID1B IS REFERRED TO AS AN ASD GENE, NOT THAT IT DOESN'T CON IF HE EVER RISK AS WELL, AGAINST CONTROLS, BUT ARID1B TRUNCATED VARIANTS ARE FAR MORE LIKELY TO BE SEEN IN CLKTIONS IN WHICH PEOPLE HAVE BEEN AS TAPED FOR INTELLECTUAL DISABILITY RATHER THAN AUTISM. THIS IS NOT TRUE FOR CHD8 AND A COUPLE OTHERS EMERGING AT THE TOP OF THE LIST IN TERMS OF THOSE THAT APPEAR IN ASD COLLECTIONS FREQUENTLY. IT'S GOING TO TAKE MORE DATA TO UNDERSTAND THE SPREAD IN TERMS OF WHICH GENES, WHEN FUNCTION IS LOST, ARE LIKELY TO CONFER ASD RISK WITH OR WITHOUT CONCURRENT INTELLECTUAL DISABILITY. SO QUESTION WE'VE BEEN WORKING ON A LOT IS DO ASD'S COMMON POLYGENIC INFLUENCES RELEVANT TO A GREATER FRACTION OF CASES HAVE SIMILAR PHENOTYPIC ASSOCIATIONS IN THE LAST COUPLE YEARS, A SURPRISING PATTERN HAS EMERGED FROM THE LITERATURE IN WHICH IT APPEARS FOR BOTH ASDs AND ADHD, THE COMMON VARIANT, PHENOTYPIC ASSOCIATIONS, APPEAR QUITE DIFFERENT FROM THE RARE VARIANT ASSOCIATIONS. TO EXPAND ON SOMETHING THAT MATT I BELIEVE MENTIONED DURING THE QUESTION AND ANSWER SESSION EARLIER TODAY, BOTH ASDs AND ADHD DO SHOW EXCESS IN TERMS OF RARE DLEE TEAR YES VARIANTS IN CASES VER VERSUS CONTROLS, IN FACT THE CONTROL DATA LOOKS QUITE SIMILAR IN ADHD TO ASD CASES WITHOUT INTELLECTUAL DISABILITY. WHICH MEANS IN BOLT OF THOSE PHENOTYPES, THE RARE VARIANT PROFILE, IF SIMILAR, WOULD BE ASSOCIATED WITH LOWER IQ, AND A REDUCTION -- LESS LIKELY TO HAVE CHILDEN. IN COMMON VARIANT SPACE, YOU SEE SOMETHING THAT'S QUITE DIFFERENT FOR BOTH OF THOSE DISORDERS. ASD POLYGENIC INFLUENCES ARE ASSOCIATED WITH HIGHER IQ AND REDUCTION -- ADHD'S COMMON VARIANT INFLUENCES ARE ASSOCIATED WITH LOWER IQ AND HAVING MORE CHILDREN. SO THERE'S AN ENORMOUS AMOUNT OF VARIABILITY IN THE PHENOTYPIC DISTRIBUTION OF THE INFLUENCE -- PSYCHIATRIC DISORDERS. TO BETTER UNDERSTAND THESE OBSERVED GENETIC CORRELATIONS, PARTICULARLY THOSE THAT APPEAR POSITIVE BETWEEN AUTISM RISK AND HIGHER I.Q., WE DID AN ANALYSIS IN THE COLLECTION COMPARING THE POLYGENIC RISK SCORE FOR SEVERAL DIFFERENT TRAITS AMONG PEOPLE WITH A DIAGNOSIS OF ASD TO THAT OF THEIR PARENTS. WHAT YOU EXPECT IS ON AVERAGE, PEOPLE HAVE ABOUT THE SAME AMOUNT OF POLYGENIC RISK FOR ANY TRAIT AS THE AVERAGE OF THEIR MOM AND DAD. THERE'S A DISTRIBUTION AROUND THAT, BUT IF YOU WERE TO SELECT A RANDOM SAMPLE, YOU'D EXPECT THAT KIDS, ON AVERAGE, HAVE THEIR PARENTS MEAN POLYGENIC RISK FOR -- THIS IS BROKEN IF YOU ASCERTAIN THE KIDS TO, FOR EXAMPLE, BE EXTREMELY TALL COMPARED TO THEIR PARENTS, OR TO HAVE AUTISM WHEN THEIR PARENTS DO NOT. IN WHICH CASE A POLYGENIC INFLUENCE ON ASD IS LIKELY TO BE OVERTRANSMITTED, YOU SEE MORE OF IT TRANSMITTED THAN PREDICTED BY CHNCE TO THE OFFSPRING. SO FOR EACH OF POLYGENIC RISK FOR ASDs, SCHIZOPHRENIA, AND GREATER EDUCATIONAL ATTAINMENT, YOU DO, IN FACT, SEE THIS PATTERN WHERE ASD PRO BANS HAVE, ON AVERAGE, OVERINHERITED, RECEIVED MORE POLYGENIC RISK FOR EACH OF THOSE THINGS, THAN YOU WOULD EXPECT BY CHANCE. SO THESE POSITIVE GENETIC -- IN SOME FORM OF ANCESTRAL OR -- WITHIN FAMILIES, YOU SEE THE AFFECTED MEMBERS HAVE MORE POLYGENIC RISK THAN THE UNAFFECTED MEMBERS. SURPRISINGLY, THIS IS TRUE REGARDLESS OF WHETHER OR NOT THE ASD PROGRAM HAS INTELLECTUAL DISABILITY, AND WE DO NOT SEE ANY SUBSTANTIAL DIFFERENCE IN TERMS OF POLYGENIC OVERINHERITANCE BETWEEN PRO BANS WHO DO OR DO NOT HAVE A STRONG ACTING DE NOVO MUTATION. SO EVEN IN THE PRESENCE AFTER DE NOVO MUTATION THAT CREATE SOMEWHERE AROUND A 10 TO 20 FOLD RISK FOR AUTISM, WE SEE -- IN OVERTRANSMISSION OF THE SAME POLYGENIC RISK FACTORS TO PROBANDS. SO WITH THAT OVERVIEW, WE'RE INTERESTED IN USING PHENOTYPE DATA TO BETTER UNDERSTAND HOW DIFFERENT CLASSES OF VARIATION MIGHT BE BOTH IDENTIFIED IN CREATING RISK DIFFERENTLY, AND TO USE THAT INFORMATION TO GUIDE MODELS OF UNDERLYING BIOLOGY AND FUNCTIONAL FOLLOW-UP STUDIES. SO PART OF THE REASON THAT THIS WORK HAS PROGRESSED COMPARATIVELY RAPIDLY IN AUTISM IS SIMPLY BECAUSE THE DATA EXIST EXISTS THANKS TO THE SIMPLEX COLLECTION AND A HAPPENEDFUL OF OTHER DATASETS. IT'S ACTUALLY INCREDIBLY RARE, AS HAS BEEN MENTIONED A FEW TIMES TODAY, FOR OUR LARGE CASE CONTROL DATASETS TO CONTAIN ANY INFORMATION BEYOND PRIMARY CASE STATUS, ARE YOU A BOY OR A GIRL, IN WHAT COUNTRY ARE YOU FROM, WHICH MAKES IT VERY DIFFICULT TO DO THESE SORTS OF ACTIVITIES THAT LOOK AT HETEROGENEITY OF THE CASES. AND PART OF THE REASON THE RESOURCES ARE SO SPARE IS BECAUSE IT'S NOW LESS EXPENSIVE, FOR EXAMPLE, TO GENERATE DATA FROM A SNP ARRAY THAN IT IS TO COLLECT A STANDARD IQ TEST. IT'S ABOUT A THIRD AS EXPENSIVE ACTUALLY. SO THROUGH THE STANLEY CENTER, WE'RE WORKING WITH LOTS OF PARTNERS, PARTICULARLY PEOPLE WITH PHENOTYPING EXPERTISE, TO TRY TO DEVELOP THREE EFFICIENT MEASURES OF BEHAVIOR AND COGNITIVE -- THAT CAN BE DEPLOYED BY THE COMMUNITY IN LARGE SCALE STUDIES FOR PSYCHIATRIC DISEASE. AS AN EXAMPLE OF SOME OF THOSE COLLABORATIONS, FOCUSING FOR A MOMENT ON COGNITION, ONE, FOR EXAMPLE, MOST LIKELY, WHEN MEASURING COGNITION, WOULD LIKE SOMETHING THAT IS QUICK, CAN GET YOU A DECENT FACTOR SCORE IN TERMS OF GLOBAL COGNITION SINCE THAT'S BEEN SO ASSOCIATED WITH RISK FOR PSYCHIATRIC DISEASE, AND RETAIN ENOUGH DIMENSIONALITY FOR THE MEASURE TO GET AT SOME OF THE MORE SPECIFIC DOMAINS OF INTEREST. WORKING WITH RAQUEL ON THE NEUROCOGNITIVE BATTERY AND TEST MY BRAIN, WE'RE TRYING TO DEVELOP TWO OPTIONS FOR PEOPLE TO BE ABLE TO QUICKLY AND FOR FREE ASSESS COGNITION IN A WAY THAT HAS BEEN GENETICALLY VALIDATED IN TERMS OF USE FOR LARGE SCALE STUDIES. WE'RE ALSO WORKING WITH JP ONEL LA WHO'S DEVELOPED A GLOBAL PLATFORM TO ASSESS THINGS LIKE SLEEP AND PHYSICAL ACTIVITY AND VARIOUS SURVEYS ON SMARTPHONES. OVER THE NEXT COUPLE OF YEARS, HOPE TO BE ABLE TO MAKE THESE THINGS PUBLIC AND DEPLOY THEM SO THAT EVERYONE CAN USE [INAUDIBLE] SO WITH AN YOU A TEMPT TO BE VERY SPEEDY, MANY THANKS TO EVERYONE AT THE STANLEY CENTER AND MARK AND STEVE IN PARTICULAR. [APPLAUSE] >> SO THANK YOU, ELISE, FOR YOUR SPEEDINESS AND FOR A REALLY INTERESTING AND IMPORTANT TALK. AND NOW WE HAVE THREE TALKS ABOUT RESOURCES THAT WILL FOLLOW IN, I PRAY, RAPID SUCCESSION, SO LINDA BRZUSTOWICZ IS GOING TO START, THEN THE SAGE BIONETWORKS, AND THEN BEN NEALE ON HAIL. YOU PROBABLY DON'T KNOW WHAT HAIL IS, SO I'LL LEAVE YOU IN SUSPENSE. >> OKAY. SO I ACTUALLY LOOKED AT THE AGENDA AND FIGURED THIS WAS GOING TO HAPPEN SO I PREPARED A SHORT TALK. SO I'M ONE OF THE TWO P.I.s THAT RUNS THE NIMH REPOSITORY AND GENOMICS RESOURCE, JAY TISHFIELD WHO'S HERE IS THE OTHER AND MY PREDECESSOR FOR MANY YEARS IN THAT ENDEAVOR. WE'RE SUPPORTED BY A COOPERATIVE AGREEMENT FROM NIMH SO IT'S VERY MUCH A PARTNERSHIP WITH NIMH IN TERMS OF WHAT WE COULD DEPARTMENT OF. SO WHAT IS THE REPOSITIVE? IT'S ACTUALLY HAD SOME DIFFERENT NAMES OVER THE YEARS, BUT IT BASICALLY WAS ESTABLISHED IN 1998 TO LEVERAGE AND INCREASE THE VALUE OF HUMAN GENETIC SAMPLES AND DATA THAT ARE PRODUCED THROUGH THE NIMH FUNDED RESEARCH. THERE REALLY ARE TWO MAIN SORT OF PILLARS OF THE ACTIVITY. SO ONE IS THE SUPPORT FOR PRIMARY RESEARCH STUDY. SO THIS IS WHERE A LOT OF PEOPLE, I THINK, HAD INTERACTED WITH US, YOU GET FUNDING, YOU'RE COLLECTING BLOOD SAMPLES, YOU SEND THEM TO THE REPOSITORY, THE REPOSITORY WILL THEN MAKE CELL LINES, EXTRACT DNA, EXTRACT RNA IN SOME CASES, PLASMA, RETURN SOME OF THAT TO THE INVESTIGATOR, BUT ALSO CREATES A BANK SO THAT OTHER FOLKS CAN USE THAT. SO THAT WE CAN MULTIPLY THE RESOURCE. WE ALSO PROVIDE SUPPORT FOR SOME ASSESSMENT TOOLS, THE DIGS AND THE FIGS, AND ALSO PROVIDE SOME DATA ANALYTIC TOOLS. BUT WE ALSO PROVIDE SUPPORT FOR SECONDARY STUDIES, SO WE HAVE THIS BIG BIOBANK NOW OF ALL THESE MATERIALS THAT OTHER FOLKS THAT ARE INTERESTED CAN NOW COME TO AND MAKE AN APPLICATION FOR AND GET MATERIALS FROM US. WE ALSO CURATE -- IF YOU DON'T& KNOW WHAT THE DIAGNOSIS IS, WHAT GOOD IS THE SAMPLE. AND WE ALSO HOST SOME BUT ALSO PROVIDE LINKS TO A LOT OF GENOMIC DATA ASSOCIATED WITH THESE SAMPLES SO WE'VE BEEN REALLY AGGREGATING THINGS OVER THE YEARS. THE OTHER ASPECT OF THE RESOURCE THAT'S IMPORTANT TO POINT OUT IS THAT IN ADDITION TO THESE THINGS THAT ARE SORT OF DIVIDED INTO PRIMARY U.S. DIS, SECONDARY STUDIES, WE ALSO CAN PROVIDE A FULL RANGE OF ANALYTIC SERVICES SUCH AS GENOTYPING, SEQUENCING, GENE EXPRESSION ANAL CIRCLES AND ALSO WE HAVE A VERY ACTIVE STEM CELL RESOURCE WHERE WE CAN DO IPSCV PROGRAMMING, THESE ARE AVAILABLE TYPICALLY ON A FEE FOR SERVICE BASIS TO ANY INVESTIGATOR WHO'S INTERESTED. SO IT CAN BE PRIMARY FOLKS, OR A SECONDARY USER CAN COME AND SAY I'M INTERESTED IN THESE PEOPLE I FOUND IN THE REPOSITORY DATABASE, CAN YOU TAKE THAWR THEIR CELL CELL LINES, REPROGRAM THEM AND SEND THEM TO ME. SO WHO'S BEEN USING US OVER THE YEARS? I JUST WANTED TO GIVE YOU GUYS A SENSE THAT WE ACTUALLY HAVE A LOT OF PROJECTS THAT HAVE BEEN FUNDED, OBVIOUSLY MANY OF THE FOLKS IN THIS ROOM HAVE INTERACTED WITH US, SO WE'VE HAD ALMOST 200 APPROVED PROJECTS IN TERMS OF THE SUBMITTING SIDE. OVER 600 SITES AND QUITE A FEW INTERNATIONAL SITES HAVE SUBMITTED TO US AND ACTUALLY ABOUT 50% OF THE PROJECTS ARE STILL SUBMITTING SAMPLES FOR DATA. THAT'S REALLY A COMBINATION OF THAT THERE'S THERE ARE A LOT MORE PROJECTS FUNDED NOW THAN THERE WERE A LONG TIME AGO SO WE HAVE A LOT OF THINGS ACTIVE BUT WE ALSO HAVE A LOT OF PROJECTS THAT EVEN THOUGH MAYBE THEY SORT OF FINISH, THEY'RE STILL GETTING STUFF TO US SO WE'RE TRYING TO MAINTAIN A RELATIONSHIP AND GET AS MUCH AS WE CAN OVER THE YEARS FROM FOLKS AS THEIR WORK GOES ON. SO SINCE 2000, WE'VE HAD OVER 200,000 BIOSAMPLES THAT HAVE BEEN SUBMITTED TO THE REPOSITORY. AND IF YOU LOOK ON THE FLIP SIDE OF WHO'S BEEN ACCESSING THE REPOSITORY, YOU CAN SEE WE'VE HAD 1500 PROJECTS FROM OVER 450 DIFFERENT P.I.s APPROVED FOR ACCESS, AND 850,000 BIOSAMPLES DISTRIBUTED. SO IF YOU LOOK AT THE RATIO OF THINGS GOING OUT TO THINGS GOING IN, EVERY SAMPLE THAT WE RECEIVE TYPICALLY IS GOING OUT TO OVER FOUR DURCH PEOPLE. SO THAT I THINK IS THE DEFINITION OF SUCCESS FOR THE REPOSITORY, WE'RE REALLY MULTIPLYING THE RESOURCE AND GETTING THINGS INTO PEOPLE'S HANDS AND THAT'S JUST FOR THE BIOMATERIALS. THERE'S A LOT MORE ACCESSING OF THE DATA BECAUSE WE DON'T -- THAT'S REALLY EASY PEOPLE CAN JUST APPLY, GET A PASSWORD, AND DOWNLOAD STUFFMENT ALL RIGHT. SO I JUST WANT TO SHOW A LITTLE BIT OF SORT OF A GRAPH OF THE GROWTH OF THE COLLECTION SO YOU CAN SEE FROM STARTING IN 2000, THERE WERE A NUMBER OF YEARS WHERE THINGS WENT RELATIVELY FLAT BUT WE'VE REALLY BEEN HAVING TREMENDOUS GROWTH BOTH IN THE SIZE OF LONG-STANDING -- THINGS LIKE SCHIZOPHRENIA AND BIPOLAR, SOME OF THE FOUNDING COLLECTION, BUT ALSO THERE ARE A LOT OF NEW COLLECTIONS THAT HAVE COME ONLINE IN RECENT YEARS, FOR EXAMPLE, WE WERE HEARING ABOUT THE TWEP 2QQ1.2 DELETION SYNDROME, SO WE NOW HAVE A CHECK CHECK SHUP FOR THAT AND THERE'S SAMPLES AVAILABLE FOR ANYBODY THAT WANTS TO USE THEM. SO WE HAVE A LOT OF STUFF TO SHARE. SO PEOPLE OFTEN THINK WITH R. ABOUT ABOUT US IN TERMS OF THE MATERIALS, SO WE HAVE DNA, WE HAVE THE LYMPHOBLASTIC CELL LINES, WE CAN MAKE IPFCs, WE'VE GOT A LOT OF BIOMATERIALS THAT ARE AVAILABLE TO SHARE. WE HAVE A LOT OF GENETIC AND GENOMIC DATA, SOME WE ACTUALLY HOUSE OURSELVES BUT A LOT OF IT WE PROVIDE LIPS TO DB GAP SO FOLKS CAN GET THE INFORMATION THERE, BUT THE OTHER THING I REALLY WANT TO POINT OUT AND SORT OF HAMMER ON IS WE HAVE AN AWFUL LOT OF PHENOTYPIC DATA. SO LOTS OF FOLKS HAVE USED SORT OF JUST THE OVERALL DIAGNOSIS, WHO'S GOT SCHIZOPHRENIA, BIPOLAR, BUT FOR 70% OF THE SAMPLES THAT ARE IN THE REPOSITORY, WE ACTUALLY HAVE SOME MORE DETAILED ASSESSMENT WHERE WE ACTUALLY HAVE ITEM LEVEL INSTRUMENT DATA AVAILABLE FOR DOWNLOAD. SO WE'VE GOT OVER 100 DIFFERENT INSTRUMENTS, SO THERE'S A LOT OF DIFFERENT -- A VARIETY IN TERMS OF THAT, BUT THERE ACTUALLY IS AN AWFUL LOT OF JUST SYMPTOM LEVEL DATA THAT PEOPLE CAN GO AHEAD AND COUNCILLOAD. SO JUST TO GIVE AN EXAMPLE OF SOME OF WHAT WE'RE DOING IN ONE OF THESE THINGS, SO WE'RE REALLY TRYING TO MAKE THIS MUCH MORE ACCESSIBLE TO PEOPLE. SO ABOUT A 2000 QUESTION -- 2000 ITEM QUESTION YEAR THAT GOES OVER MANY, MANY THINGS OF SYMPTOMS, AND WE'VE GOT ABOUT 25,000 SUBJECTS WHO HAVE COMPLETED ASSESSMENTS FOR THIS. SO YOU CAN GET ITEM LEVEL DATA ON THESE FOLKS. NOW THE PROBLEM IS THE OLD WAY OF DOING IT WAS DIFFERENT STUDIES HAD DEPOSITED THE FILES IN DIFFERENT FORMATS, THEY'D USED DIFFERENT VERSIONS OF THE INSTRUMENT, DIFFERENT FILE TYPES, DIFFERENT ENCODINGS FOR VARIABLES, SO TO ACTUALLY MAKE USE OF IT ACROSS STUDIES, YOU WOULD HAVE TO DOWNLOAD ALL THESE DIFFERENT FILES AND THEN DO A LOT OF HARM NIEDZING OP HARMONIZING ON YOU R OWN. WE'VE ACTUALLY UNDERTAKEN THAT, WE DOS DO THAT AS THE JOB OF CURATION. SO WE'VE GONE AHEAD, TAKEN THESE, ARM NIEZED THEM, PUT THEM INTO UNIFORM FORMATS AND MADE THAT INTO A SEARCHABLE COLLECTION THAT PEOPLE WOULD BE ABLE TO GO ON THE WEBSITE AND SEARCH ON THE INDIVIDUAL ITEMS BUT ACROSS ALL OF THE STUDIES AT ONCE AND NOT HAVE TO DO IT TO SORT OF ALMOST MANUALLY ACROSS INDIVIDUAL FILES. SO THE OTHER THING THAT I WANTED TO POINT OUT IS THAT WE'RE REALLY MAKING A BIG PUSH TO DO STUFF WITH THE PHENOTYPIC DATA TO MAKE IT MORE ACCESSIBLE. ONE OF THE REALLY IMPORTANT WAYS THAT PEOPLE INTERACT WITH US IS THROUGH OUR WEBSITE. SO WE'RE CURRENTLY IN THE PROCESS OF HIRING ACTUALLY AN EXTERNAL COMPANY THAT'S GOING TO HELP DOUSE A MAJOR WEBSITE REDESIGN TO MAKE IT MORE ACCESSIBLE TO PEOPLE TO HELP PEOPLE FIND WHAT THEY NEED. AND ONE OF THE THINGS THAT I'M ACTUALLY HERE TO ASK FOR YOUR HELP WITH IS WE WOULD LIKE TO GET INPUT FROM THE COMMUNITY AS TO WHAT FOLKS WOULD REALLY LIKE FROM US. SO WE'RE LOOKING FOR PEOPLE WHO& HAVE INTERACTED WITH THE WEBSITE, EITHER AS DEPOSITORS OF PRIMARY DATA, SECONDARY USERS, USERS OF SOME OF THE TOOLS, SO PEOPLE WITH DIFFERENT ROLES, WE'RE PARTICULARLY INTERESTED IN SORT OF THE FOLKS IN THE TRENCHES, SO THE POSTDOC, THE GRADUATE STUDENTS. , WE'VE GOT THIS FEEDBACK BUTTON. SO IF PEOPLE ARE INTERESTED IN HELPING US, IF YOU COULD JUST GO THERE, YOU CLICK ON THAT, YOU IT OPENS UP A PAGE TO ASK FOR YOUR ADDRESS AND WHAT YOUR ISSUE IS. IF YOU'RE INTERESTED IN VOLUNTEERING TO BE CONTACTED MAYBE FOR A FOCUS GROUP OR QUESTIONNAIRE, IF YOU COULD PUT YOUR NAME IN THERE, YOU WOULD DO US A TREMENDOUS SERVICE TO MAKE US BETTER AND MORE ACCESSIBLE FOR FOLKS. SO THAT'S REALLY IT. THERE'S A CAST OF MANY PEOPLE WHO HAVE HELPED WITH THIS. AGAIN, IT'S A TEAM SCIENCE EFFORT. PLEASE COME VISIT US, LEAVE YOUR NAME AND CONTACT FOR A SURVEY. THANK YOU. [APPLAUSE] >> THANK YOU. AS YOU'VE HEARD TODAY, SAGE NETWORKS FUNCTIONS AS THE DATA COORDINATING CENTER FOR SEVERAL OF THE CONSORTIA AND COLLABORATIONS YOU'VE HEARD ABOUT TODAY. BEFORE I GO INTO HOW WE FUNCTION IN THAT ROLE, I'D LIKE TO GIVE YOU AN IDEA OF WHAT IT IS THAT SAGE DOES BEYOND DOING THIS TYPE OF DATA COORDINATION. SAGE'S MISSION IS TO ACCELERATE OPEN BIOMEDICAL RESEARCH BY IMPROVING HUMAN HEALTH AND' WE DO THIS IN MANY DIFFERENT WAYS. THESE METHODS CAN BASICALLY BE SEEN AS BEING BUILT ON THREE DIFFERENT PILLARS. ONE IS A TEAM SCIENCE PILLAR, THIS IS WHERE WE PLAY A ROLE IN CONSORTIA, WE GET TEAMS OF SCIENTISTS TOGETHER SO THAT THEY CAN LEARN FROM AND THROUGH EACH OTHER. WE WORK WITHIN OPEN SCIENCE WHERE WE HAVE ANALYSTS THAT ARE DEVELOPING TOOLS AND METHODS THAT CAN BENEFIT OTHER TEAMS OF SCIENCE, AN OVER THE PAST COUPLE OF YEARS, WE HAVE PARTICIPATED IN PART PAPT CENTER SCIENCE IN ORDER TO BE INVOLVED IN THE INDIVIDUALS THAT WE DO CONDUCT SCIENCE ON. AND WE DO THIS BY DEVISING MULTIPLE TYPE OF PILOTS AND SYSTEMS AND APPROACHES AND FOR THE ONES THAT WORK, WE BUILD INFRASTRUCTURE THAT IS TECHNICAL INFRASTRUCTURE, SCIENTIFIC INFRASTRUCTURE, CULTURAL OR ETHICAL, THAT WE CAN USE ACROSS THE PILLARS THAT WE FOCUS ON. WE TURN THESE INFRASTRUCTURE OVER TO THE COMMUNITY SO THAT THE COMMUNITY CAN THEN FURTHER BUILD ON IT AND UTILIZE THEM UNDER OUR PRINCIPALS OF OPEN SCIENCE. A FEW EXAMPLES, WE AT SAGE DEVELOP AND HELP OTHER GROUPS DEVELOP MOBILE HEALTH APPS AND ALSO DEVELOPED AN ELECTRONIC CONSENT PROCESS THAT IS A VISUAL METHOD OF ACQUIRING INFORMED CONSENT TO DO -- MOBILE PHONES. ONE OF THESE MOBILE HEALTH APPS IS EMPOWER, THIS WAS LAUNCHED TWO YEARS AGO THROUGH THE APPLE RESEARCH KIT. PEOPLE THAT PARTICIPATE IN THE EMPOWER STUDY THROUGH THIS APP CAN CONSENT TO SHARE THE SURVEY DATA AND THE SENSOR DATA COLLECTED THROUGH THIS APP WITH THE BROADER QUALIFIED INVESTIGATORS IN THE BROADER RESEARCH COMMUNITY. WE WORK IN OPEN SCIENCE BY CONDUCTING COMPUTATIONAL CHALLENGES IN COLLABORATION WITH DREENL DREEM AND THESE ARE CROWD SOURCING EFFORTS, BIOLOGICAL, TECHNICAL QUESTIONS, AND WE HAVE LAUNCHED A WHOLE VARIETY OF THESE CHAL EPTION, BUT I CHALLENGES, BUT A RECENT CHALLENGE THAT WAS JUST LAUNCHED WHICH IS IN COLLABORATION WITH THE MICHAEL J. FOX FOUNDATION AND IS BASED ON THE EMPOWER DATA, AND THE GOAL IS TO BENCHMARK METHODS IN PROCESSING OF THE SENSOR DATA COLLECTED TO -- THROUGH THE EMPOWER ACT. WHAT THESE PILLARS IN OUR APPROACHES HAVE IN COMMON S THAT THE ENGAGED DIVERSE COMMUNITIES OF RESEARCHERS AROUND BIOLOGICAL AND ANALYTICAL PROBLEMS AND OFTEN ARE TOO COMPLEX FOR SINGLE INSTITUTIONS TO TAKE ON. THIS IS PARTICULARLY TRUE FOR OUR SUPPORT OF COLLABORATIVE COMMUNITIES. WE SUPPORT AND WE ALSO PLAY A COLLABORATIVE ROLE IN A VARIETY OF DIFFERENT DISEASE EA ALZHEIMER'S, AS WELL AS THESE& NEUROPSYCHIATRIC CON SORE IS THAT CON CON CORE SHA THAT YOU'VE CONSORTIA THAT YOU'VE HEARD ABOUT TODAY. TODAY THESE CONSORTIA, WE WORK WITH THE NIMH IN ORDER TO INCENTIVIZE DATA SHARING AND INCENTIVIZE SHARING OF NONL BY DEVELOPING BOTH CULTURE AND STRUCTURAL INFRASTRUCTURE THAT CAN ENABLE THIS. AS WELL AS GETTING THE DATA THAT IS BEING COLLECTED TO THESE CONSORTIA DISTRIBUTED INTO THE BROADER RESEARCH COMMUNITY SO THAT IT CAN BENEFIT ALL OF YOU AS YOU CAN SEE TODAY THAT A LOT OF PEOPLE HAVE TAKEN ADVANTAGE OF. USING THE COMMON MIND CONSORTIUM AS AN EXAMPLE, WE PLAY A ROLE IN THIS CONSORTIUM, BOTH IN THE ANALYSIS OF THE DATA, GENERAL PROGRAM MANAGER OF THE CONSORTIUM AND ALSO BY PROVIDING A CENTRALIZED DATA MANAGEMENT SYSTEM THAT TRACKS ALL THE DATA, THE RAW DATA, THE PROCESS DATA, THE ANALYTICAL OUTPUT, AS WELL AS ANYTHING ELSE CONSORTIUM-RELATED. I'M NOT GOING TO GO INTO ANY DETAIL OF WHAT THE CON CONSORTIA HAS GENERATED AS THEY DID AN EXCELLENT JOB OF THAT EARLIER. BUT WHAT ENABLES HAVING ALL OF THIS DATA TOGETHER, ALL OF THIS CONTENT TOGETHER IN A CENTRALIZED SPACE IS THAT ALL COLLABORATORS HAVE ACCESS TO THE SAME INFORMATION, TO THE SAME CONTENT, WHICH REDUCES THE TIME IT TAKES IN ORDER TO GET DOWN TO THE BUSINESS OF USING THIS DATA, WHICH IS THE BIOLOGIC QUESTIONS AND THE ANALYSIS OF IT. AS PAMELA TALKED ABOUT, THERE IS CONSORTIUM ALL ANALYSIS THAT IS TAKING PLACE WITH THIS DATA, AND THERE ARE INDEPENDENT WORKING GROUPS THAT ARE USING THE DATA TO ANSWER DIFFERENT TYPES OF QUESTIONS BASED ON THEIR OWN INTEREST. THE INFRASTRUCTURE THAT WE USE IN SUPPORT OF NOT JUST THESE COMMUNITIES BUT ALSO THE DATA IS BEING REL EASED THROUGH MOBILE APPS SUCH AS EMPOWER AND ALSO THAT SUPPORT OUR CHALLENGES IS OUR SNAPS SYNAPSE SYSTEM, FREE FOR ANYONE TO USE, OPEN SOURCE, THE CODE IS AVAILABLE ON GIT HUB. IT FUNCTIONS AS A FILE STORAGE SYSTEM BUT REALLY MORE THAN THAT, IT TRACKS ALL ASSETS RELATED TO A SCIENTIFIC COLLABORATION AND ANALYSIS. COLLABORATORS USING SYNAPSE CAN INTERACT WITH IT, WHETHER IT IS THROUGH THE WEB PORTAL OR PROGRAMMATICALLY. USING IT IN A COLLABORATIVE WAY, IN PART AS A COMMUNICATION PORTAL, IS WORK DONE AT SAGE WHERE HE RAN A SERIES OF ANALYSIS USING THE COMMON MINE RNA SEQ DATA AND POSTED NOT JUST THE ANALYTICAL OUTPUT BACK TO THE PORTAL, BUT ALSO PUBLISHED ALL THE ANALYTICAL STEP OF HIS WHOLE PIPELINE INTO THE PORTAL ITSELF, GIVING EVERYONE ACCESS TO EXACTLY WHAT WAS DONE AND SO THAT PEOPLE COULD TRACE ALL THE STEPS DEVELOPS SWELS THE AS WELL AS ALL THE OUTPUT. IN ADDITION TO THAT, HE USED THE PROVIDENCE SYSTEM TO TRACK THE ORIGIN OF EACH FILE. SO YOU CAN SEE HERE, ALL THE UPSTREAM FILES, WHAT CODE WAS USED IN THE DIFFERENT ANALYSIS STEPS AND ALSO WHO HAD CONDUCTED THE WORK THAT WENT INTO THE ANALYSIS THAT HE DID, AND YOU CAN SEE HERE THAT TENIR BUILT ON WORK THAT HAD PREVIOUSLY BEEN DONE BY GABRIEL HOFFMAN AS WELL AS OTHERS. AGAIN IT WAS COVERED ALREADY BUT BEING ABLE TO HAVE RESOURCES IN THE CENTRALIZED SPACE -- COMMON MIND SIEK PSYCHENCODE TO WORK ON THE SAME DATA AND TO BUILD ON EACH OTHER'S RESULTS BECAUSE EVERYONE HAS ACCESS TO WHATEVERYBODY ELSE DID. SO THERE IS A LOT OF OVERLAP, THERE ARE INDEPENDENT PUBLICATIONS BUT THERE'S A LOT OF OVERLAP IN TERMS OF THE DATA THAT WAS USED THAT WENT INTO THESE PUBLICATIONS. THE IMPORTANT TENET BOTH OF SAGE'S MISSION AND THE NIMH AS WELL IS OF COURSE TO PROMOTE THE DATA RE-USE OF THE RESOURCES THAT ARE BEING GENERAL RAISED. GENERATED. WE SUPPORT THIS IN PART BY HAVING PUBLIC CREATING PUBLIC PORTALS WITHIN SYNAPSE, WHERE THAT PUBLIC DATA IS RELEASED, THEY'RE CALLED THE KNOWLEDGE PORTALS, A PSYCHENCODE AND A BSMN KNOWLEDGE PORTAL. THE DATA ACCESS TO THIS DATA IS DONE IN COLLABORATION WITH THE NIH REPOSITORY GENOMICS RESOURCES IN TERMS OF THE COMMON MINE AND PSYCHENCODE DATA WHERE THE DATA IS ALSO DESCRIBEED ON THEIR PORTAL SO PEOPLE CAN HAVE ACCESS TO IT GOING INTO MULTIPLE DIRECTIONS. AND THE KNOWLEDGE PORTALS ALSO HAVE A DISCUSSION FORUM SO THAT WE CAN THEN RELEASE THE DATA NOT JUST IN A PASSIVE WAY BUT WE CAN GET FEEDBACK FROM USERS WHETHER THEY HAVE QUESTIONS ABOUT ACCESS TO THE DATA, OR QUESTIONS THE DATA ITSELF. AND WE USE THIS DISCUSSION FORUM IN ORDER TO GET FEEDBACK ABOUT WHAT THE DATA HAS BEEN USED FOR AND WE GET BACK HERE A LITTLE BIT, WE ALSO PROVIDED RESULTS EXPLORE SO THAT PEOPLE CAN ALSO INTERACT THROUGH THE PORTAL WITH THE DATA ITSELF. BUT WE USE THIS DISCUSSION FORUM IN ORDER TO GET FEEDBACK OF WHAT THE DATA HAS BEEN USED FOR, WHAT THE SECONDARY USE OF THE DATA HAS BEEN, AND WE NOW HAVE A COLLABORATION WITH NRGR WHERE WE WILL COLLABORATELY TRACK SECONDARY USE OF THE DATA SO THAT WE CAN GET A FULL OVERSIGHT OF HOW THIS DATA SK USED. THESE CONSORTIA THAT WE SUPPORT DON'T FUNCTION IN ISOLATION AS YOU SEE TODAY, THERE HAS BEEN A LOT OF DEGREE OF COLLABORATION BETWEEN THESE GROUPS SO THAT THESE THREE CONSORTIA SHARE DATA AS WELL AS KNOWLEDGE ACROSS GROUPS. A GOOD EXAMPLE IS THE CAP STONE PROJECTS WHERE THE GROUP IS PROCESSING DATA COMING FROM A WHOLE VARIETY OF DIFFERENT SOURCES, NOT JUST THE DATA THAT HAS BEEN -- TO THE CONSORTIA ITSELF, BUT FROM THE COMMON MIND CONSORTIUM, BRAIN SPAN, ENCODE, ROAD MAP AND GTEX WHERE KAY TA IS BEING PROCESSED USING COMMON PIPELINES AND USED ACROSS A VARIETY OF DIFFERENT ANALYTICAL QUESTIONS. PART OF THAT DATA THAT IS AVAILABLE THROUGH SYNAPSE IS MADE AVAILABLE IN A WAY WHERE PEOPLE THEN CAN QUERY FOR SUBSETS OF THIS DATA THAT THEY WANT TO USE, WHETHER THAT IS A SPECIFIC ASSAY, A STUDY, A DIAGNOSIS, OR SPECIFIC BRAIN REGION OR CELL TYPE THAT THEY WANT TO GET DATA FROM. SO TO END THIS UP QUICKLY, I WANT TO ACKNOWLEDGE ALL THE CONSORTIA AS WELL AS GITA, LAURA AND THOMAS, APPRECIATE GIVING US AN OPPORTUNITY TO IMPLEMENT OUR GOALS WITHIN THESE CONSORTIA, AND THE TEAM AT SAGE AND PARTICULARLY SALLY AND KENNETH BOTH IN THE AUDIENCE HERE TODAY. THANK YOU. [APPLAUSE] >> THANK YOU. SORRY TO BE IMPOLITE TO PUSH EVERYONE ALONG. BEN IS NOW GOING TO QUICKLY TELL US ABOUT A FORM OF WEATHER, HAIL. >> THIS COULD BE THE SECOND TALK THIS YEAR I'VE TRIED TO COMPRESS 10 MINUTES INTO ONE MINUTE. LET'S SEE HOW I GO. WHY NOT? MORE FUN THAT WAY, RIGHT? OKAY. HAIL. SO SEQUENCING, WE GENERATE LESS DATA, IT'S GROWING, WE WON'T SAY ASTRONOMICAL, WE'LL SAY GENOMIC BECAUSE THAT WILL BE THE SCALE, WE MADE HAIL, LOGO, VERY IMPORTANT, AS ESTABLISHED EARLIER TODAY. TEAM LEAD IS COTTON, HE'S AMAIDS AMAZING, FANTASTIC GUY, AMAZING NAME. WE DO A LOT OF QC, IT'S SLOW, IT SUCKED TWO YEARS AGO, TOOK FOREVER, TWO WEEKS, MANY TOOLS, VERY, VERY PAINFUL. WE INTERVIEWED A BUNCH OF PEOPLE, MOVED AWAY FROM THE OLD MODEL WHERE WE SHOVEL OUR DATA TO LOTS OF COMPUTE IRES, INSTEAD WE BREAK UP DATA AND SEND IT TO THOUSANDS OF COMPUTERS, VERY EFFICIENT, ALL THE NEW COMPETE UR SCIENCE STUFF, NEW ALGORITHMS BECAUSE THEY'RE MATHEMATICIANS AND SOFTWARE ENGINEERS, SO IN SIX MONTHS WE WENT FROM TWO WEEKS TO 42 MINUTES TO NOW 9 MINUTES. THAT'S FABULOUS. WE CAN DO THIS FOR NOMED, SO TAKE 30,000 YEE NOMES AND 30,000 GENOMES, HAVE IT PROCESSED AND REALLY THE DATA SCIENCES PLATFORM, THIS IS A GREAT DIVERSE DATASET SO WE'RE MOVING AWAY FROM THIS IDEA OF ONLY DOING EUROPEANS TO DO LOTS AND LOTS OF DIFFERENT POPULATION AND WE CAN DO THIS QC IN ABOUT A DAY, FOR 30,000 WHOLE GENOMES ON WHOLE SEQUENCING DATA, NOT SO BAD. IT'S BEEN A FEW P.I.S THAT HAVE CONTRIBUTED, CAN YOU LOOK AT THEM THERE. WE CAN ALSO DO MIXED MODELS TO DO HARD STAT GEN THINGS THAT I'M EXCITED ABOUT. THAT WITH CAN DO IN ABOUT THREE MINUTES, ONE YOU HAD THOUSAND VARIANT FOR THE GRM AND 15 MILLION VARIANTS. NEXT CHALLENGE IS UK BIOBANK, 600,000 INDIVIDUALS, 6,000 PHENOTYPES, TWEP 6 MILLION MARKERS, FOUR HOURS IN HUMAN TIME, A THOUSAND BUCKS ON THE CLOUD. SO NOT SO BAD. YOU CAN LEARN MORE ABOUT IT AT HAIL.IS. THERE'S AN ACTIVE GITR, ET CETERA, ET CETERA, COMMUNITIES ENTHUSIASTIC TO ENGAGE, AND I'D LIKE TO SAY THANKS TO NIMH AND STANLEY CENTER. THERE YOU GO. [APPLAUSE] >> SO WHAT TOOK YOU SO LONG? NOW MARK DALY WITH THE BENEDICTION. >> HAIL IS GREAT, GENOMES ARE EXCELLENT, THOMAS, YOU ARE WONDERFUL, THANK YOU FOR EVERYTHING. I THINK STEVE MCCARROLL GAVE A GREAT WRAPUP OF WHAT WE NEED TO DO MOVING FORWARD SO I'M NOT GOING TO BELABOR THAT. I THINK IT'S CLEAR THAT WE'VE LEARNED A LOT OF LESSONS NOT JUST ABOUT THE POTENTIAL BIOLOGY OF DISEASE BUT HOW WE MAKE PROGRESS AGAINST REALLY COMPLEX CHALLENGES, AND WE HAVE TO REPEAT THE SUCCESSES AND AVOID THE PITFALLS THAT WE LEARNED SO HARD OVER THE LAST ONE TO TWO DECADES WITH RESPECT TO GENETICS FOR THE NEXT STEPS BECAUSE WHETHER IT'S A PATHWAY ANALYSIS, WE ALL HAVE DIFFERENT PATHWAYS IN MIND, WHETHER IT'S TRANSCRIPTOMICS, WE ALL MIGHT BE FOCUSED ON DIFFERENT CELLS, DIFFERENT TYPES OF EXPERIMENTS. THE LESSON WE REALLY TOOK AWAY FROM THE FAILURE OF CANDIDATE GENE STUDIES WAS REALLY ONE OF HERE'S -- YOU KNOW, WE EACH HAVE VALID IDEAS AND EVERYONE MIGHT EXECUTE THEIR OWN STUDY ON A SINGLE GENE WITH FULL RIGOR, BUT IF WE DON'T EMBRACE THE FACT THAT THERE ARE THOUSANDS OF POSSIBILITIES AND REALLY THOUSANDS OF US ALL TRYING TO DO THE SAME THING, THAT WE'RE WORKING TOGETHER ON A TREMENDOUSLY COMPLEX PROBLEM AS A FIELD, WE'RE GOING TO REPEAT THOSE PROBLEMS. NOW THAT WE KNOW THAT THEY'RE OUT THERE, I THINK WE SAW TODAY A TREMENDOUS NUMBER OF PROMISING AVENUES BY WHICH WE'RE GOING TO TAKE THE SUCCESS OF GENETICS AND LEVERAGE THAT INTO SUCCESS IN BIOLOGY. SO THANK YOU, THOMAS, FOR PROMOTING TEAM SCIENCE AND LET IT CONTINUE TO MAKE GREAT PROGRESS IN THE NEXT 10 YEARS. [APPLAUSE] CLOSING WORDS, THOMAS? >> NO, I'VE TALKED ENOUGH. THANK YOU TO THE STALWORTHS WHO HAVE BEEN THE WHOLE MEETING, THANK YOU FOR ORGANIZING THIS OVERLY AMBITIOUS AND WONDERFUL DAY, AND WE LOOK FORWARD TO MORE PROGRESS.