Introduction to Yale Center for Genomic Health

April 30, 2021

ID6545

To CiteDCA Citation Guide

00:00And I'm just really happy to
00:02be here today to talk to you.
00:04You know, for the first time since
00:07I joined the community at Yale.
00:09And so today's you know Workshop
00:11is focused on human genomics,
00:13data science and precision medicine,
00:15and it's sort of designed to highlight
00:17some of the plans and activities
00:19for Center for genomic health.
00:23And you know, I'm going to start things
00:25off here by giving you an introduction
00:26to sort of what we're trying to do,
00:28and I'm going to keep it
00:30at a really high level.
00:31I have no data and very few slides.
00:33I really want to sort of outline,
00:34you know where the field is at and
00:36where we think that we can make
00:38a difference in the long term.
00:40But before I could start like to reiterate,
00:43you know what Antonio said,
00:44which is, you know,
00:45this center is really the product
00:47of a joint effort from the school
00:49and the hospital and really was born
00:51out of a shared vision and support
00:53of the prior Dean bulb output in
00:54the prior President Rick Tequila.
00:56And you know,
00:57really is the product of a lot
00:59of people's work over the course
01:00of the last couple years.
01:02You know,
01:02long before I got here and I really
01:04appreciate that and I appreciate
01:06the ongoing support of the current
01:07of the current leadership.
01:09You know in this.
01:10This joint leadership is really emblematic
01:13of the two prong mission of our center,
01:15which on one hand is to leave
01:17cutting edge genomic research.
01:19And on the other hand,
01:21is to do our very best to implement
01:23these technologies into the clinic
01:25to make meaningful improvements
01:27in health care and and both of
01:29these arms need to work together,
01:31and they both need to be strong if
01:33we're going to be successful in our vision,
01:36and so you know, I'm a basic scientist,
01:39genome biologist,
01:40Human Genetics.
01:40Bring that sort of expertise to the table,
01:43but there's really lots of different
01:44perspectives that are important
01:46here and I look forward to working
01:48with everybody for a long time
01:49on these important issues.
01:50And so, without further ado, you know.
01:53So what is what is genomic health, right?
01:55So it's an emerging medical discipline
01:58that involves anomic information about an
02:00individual as part of their clinical care.
02:03So that substantial fraction of the
02:05human disease burden has a genetic component,
02:07right?
02:08So 5% of the world's population
02:10suffers from a rare disease.
02:12Many of these are caused by
02:14rare pathogenic mutations.
02:15Most people at some point in your life
02:17will suffer from a common disease,
02:19and we know that these show
02:23substantial heritability.
02:24Now we can collect genomic data.
02:26You know affordably in that scale, right?
02:28So the obvious thing that we want to do is,
02:31you know sequence everybody's
02:32genome and collect lots of other
02:34types of OMICS data as well,
02:36and use these data to inform
02:37health care in the process.
02:39We want to improve care and
02:41obviously reduce costs,
02:42and so this vision, you know,
02:43is not controversial.
02:44This has been the vision for the
02:47past 30 years, and you know,
02:49there's lots of really exciting applications
02:51we won't have time to cover them all.
02:54And there's been some really nice
02:56great success stories along the way,
02:58right?
02:58But I think it's fair to say that you
03:01know, for the vast majority of
03:04heritable conditions, you know we're
03:05barely scratching the surface of
03:07what we could be doing. OK and.
03:12You know we need to do our best
03:14to push the envelope here because
03:16it's an important problem, right?
03:18And so so why is that so?
03:19It's worth sort of taking a step back
03:22and thinking about the big picture about
03:24where we are as a field right now and
03:26where we need to go because really motivates.
03:29Sort of how we're thinking about the center.
03:32And so you know,
03:34we've come a long way right?
03:35So you're 20 years ago we didn't even
03:38notice single human genome look like OK,
03:40and the Human Genome Project, you know,
03:42sort of gave us this solid foundation
03:44of human genome structure and function,
03:46and it allowed us to sort of start to do
03:49Human Genetics in a in a systematic way.
03:51OK, and the next big landmark was the
03:53development of these high throughput
03:55DNA sequencing methods that allowed
03:57us to go beyond one genome start.
03:59Look at many genomes,
04:00start to implement these technologies
04:01into the clinic.
04:02I mean,
04:03you know a couple years ago we finally
04:05reached that that long awaited landmark
04:06of being able to sequence a genome for $1000.
04:09And this was sort of what
04:10everybody was waiting for,
04:11right?
04:13So these first two steps notice there's many
04:15more important things that we need to do,
04:17but you know,
04:18we're on pretty solid ground right now.
04:20And right now we're kind of in the
04:22middle of these second, third,
04:24and fourth steps where we're
04:26trying to take these technologies,
04:28apply them at scale across the
04:30human population,
04:31learn about how genetic variation looks
04:33across all different ancestry groups,
04:34learn about how genetic variants
04:36operate in cells,
04:37and obviously to take.
04:40You know to look at genetic variation
04:42in the context of lock the whole
04:44companion of human diseases and start to
04:46catalog all the different variants that
04:48actually have an effect on disease risk,
04:51right?
04:51And we're kind of in the very beginning
04:53of this last stage or sort of taking
04:56this information and trying to
04:58implement it in the health care system,
05:00right?
05:00Some of the really exciting technologies
05:02are using polygenic risk scores to
05:04partition people by common disease risk.
05:06Using this technology is to
05:07increase the diagnostics diagnostic
05:09yield for rare disease.
05:10And using these knowledge to
05:12drive drug discovery or CRISPR
05:15based therapies right and so the.
05:17There's a lot to do here.
05:19The possibilities are really immense,
05:21right?
05:21And I think you know we've been
05:24trying to do this for 30 years and I
05:27think 30 years from now we'll look at
05:29the moment that we're in right now.
05:32As you know, maybe the Golden age, right?
05:34Maybe the inflection point between
05:36what came before and what came after.
05:38But like right now,
05:39it's it's kind of moving slow,
05:41actually,
05:42and we're sort of in this hard slog
05:44of trying to lay the foundation
05:46of knowledge and technologies
05:48that allow us to do this.
05:50Not in an anecdotal way,
05:51but in a systematic way in a real way.
05:54OK so. We're going to slide for awhile,
05:57so get comfortable.
05:57I mean, I want to discuss or some of
05:59the some of the challenges here because
06:01these challenges really would motivate
06:02like what we're trying to do, OK?
06:04So the first first challenge
06:06here is genome analysis. OK,
06:08so you know we can produce genomic data now.
06:11Incredible scale,
06:11but we're still not there.
06:13Still a lot of challenges in how
06:15we analyze and interpret it,
06:17so there's types of genetic variants
06:19that are very difficult to detect.
06:21There's parts of the genome that
06:23are really hard for us to look at.
06:26It's very difficult for us to
06:29predict the function or the impact
06:32of genetic variants computationally.
06:34You know this is the famous variants
06:36of unknown significance problem,
06:38and it's a huge problem not playing
06:40field and there's no easy solution,
06:42and it's something that that
06:44we need to solve.
06:45You know, one approaches, you know better,
06:47fancier machine learning algorithms,
06:49and this is important.
06:51It helps to just have a
06:53lot more genomes around,
06:54so that's important too.
06:56But we also need a couple this
06:59effort with efforts to produce.
07:01Catalogs of what variants do
07:02in cells using high throughput
07:04functional genomics methods so that
07:05we have good data to train the next
07:07generation of AI based methods for
07:09for interpreting genetic variation.
07:10And this is what we need to do if
07:13we're going to have these technologies
07:15being the clinic in a robust way.
07:17I'm at least questions.
07:18These are questions that our
07:20center is very interested in.
07:21Is something in my own lab,
07:22has worked on for a long time and we think
07:26it's going to really push the needle.
07:28So the second big challenge here
07:30is that this effort to catalog
07:32variants that cause disease is.
07:33It's just really hard.
07:35OK,
07:35that's fair to say that it's a lot
07:37harder than people appreciated
07:3910 or 20 years ago,
07:40and there's lots of reasons for that,
07:42but I'll but I'll go into a few of them,
07:46right? So on one hand.
07:48We now know that common diseases,
07:50and in fact most human traits
07:51are highly polygenic, right?
07:53Which means we have we need to
07:54study very large sample sizes in
07:56the range of 10s to hundreds of
07:58thousands of people if not millions
07:59of people you know to really get
08:02a handle on the genetics.
08:04And even for rare Mendelian
08:05diseases where we sort of think
08:07about them as being sort of simple,
08:09they can also be quite complicated due
08:11to the effects of incomplete penetrance.
08:13I'm very well expressivity.
08:16And this can also require
08:17larger sample sizes,
08:18and we specially need that if
08:20we want to map the modifyers.
08:22The protective alleles that
08:23suggest drug targets, right?
08:24So for both of these reasons,
08:26no one institution,
08:27no one lab,
08:28maybe not even any one nation
08:30can really do this on their own.
08:32We need to be participating in large
08:34scale consortia and team science that
08:36really that really get it that we
08:38also need to be more clever about how
08:41we assemble human cohorts and how we
08:43incorporate deep phenotype information.
08:44So every health system needs to be a biobank.
08:47And and every bio bank you know needs
08:49to be connected to every other bio
08:50bank in a network that allows us to
08:53communicate and identify patients
08:54that have similar genomic profiles
08:55and similar phenotypic profiles.
08:56Just something that we need to do.
09:00And the third thing we need to do
09:02is make every effort to make sure
09:03that we do a better job at including
09:06diverse ancestry groups.
09:07In the studies that we do,
09:09for historical reasons,
09:09you know most of our knowledge
09:11is built upon large studies of
09:13European descent individuals.
09:14This is a real problem because
09:15it can actually as this trickles
09:17down into the health care arena,
09:19the algorithms that we use for risk
09:21prediction and clinical decision
09:22making are going to be biased, right?
09:24So we all need to do our part to
09:26alleviate this potential serious issue.
09:28And so this general question of how do we do?
09:30Gene discovery,
09:31the next generation of gene
09:33discovery projects that are bigger,
09:34use better technologies,
09:35and there are more diverse is
09:37a real key goal of our center,
09:39and in fact you know many of our
09:42members are participating in if not
09:44leading some of the most high profile
09:47high impact studies in the world right now.
09:50And the last thing I'll mention here,
09:52I'll do this a little bit faster.
09:54Maybe is that you know the
09:56last challenge here is,
09:58is disease mechanism?
09:58OK,
09:59so all of the things I've talked
10:01about this far oftentimes at the end
10:03of that you still have a correlation.
10:06You still just have an Association
10:07you don't necessarily know how that
10:09impacts the biology of the disease,
10:11and so it's really important that
10:13we take the results of these
10:15large scale studies and these
10:17clinical sequencing efforts,
10:18and we try to translate them into concrete
10:20knowledge about disease mechanism.
10:22And this is really hard because the
10:24approach will vary a lot depending on
10:26which disease you're talking about,
10:28and so we need to engage with
10:30disease experts.
10:31People who know exactly how
10:32how that disease works.
10:34We need to, you know,
10:35engage with animal models.
10:37We need to use stem cell models,
10:39organoid models.
10:41And we need high throughput
10:42functional methods that allow us to
10:44interrogate what these genes do in
10:45cells in a high throughput ways.
10:47A lot of results to parse through.
10:50And then another solution that
10:52we're really interested in from the
10:54standpoint of getting up disease
10:56biology is using health systems as a
10:58platform for learning about this right?
11:00And so if you have,
11:02if you have a lot of people where you have,
11:05you know genomic information
11:06and you also have well organized
11:08electronic health records,
11:09you can start to design studies
11:11where you select groups of people
11:13based on their genotype and do a
11:15better job at looking for phenotype
11:18and doing focused investigations.
11:19And this is really crucial.
11:21I think for taking this types
11:23of studies to the next level,
11:24and of course this is something that
11:26we're trying to build here at yo
11:28with the generations project in the
11:30computational health platform and you'll
11:31hear more about that later today.
11:33So I think look, we covered it.
11:36There's a lot of ground covered here.
11:39I think you know there's a few
11:41take home messages you know one.
11:43This is a really hard problem.
11:46It requires collaboration.
11:48It requires input from lots of
11:51different types of expertise.
11:54Right, and actually,
11:55you know the most exciting projects.
11:57The most impactful projects and
11:59initiatives are going to come at the
12:01intersection of areas that I just mentioned.
12:04OK, and that's really the motivation
12:06for forming the center is to have a
12:09team to have a venue for combining
12:11people with lots of different expertise
12:14that can tackle these questions
12:16in a in a really impactful way.
12:21And so we sort of formed this center.
12:23Now you know there's been going
12:25on for a couple years on the
12:28clinical side's been very active.
12:30We've now sort of assembled the
12:32first initial group of members who
12:34sort of span the whole range of
12:37expertise that I just talked about.
12:39And this is just an initial group.
12:41You know.
12:42I'm new here so I don't know everybody yet.
12:44And so if you know if you're doing
12:46relevant work and you want to get involved,
12:49you know, please contact me.
12:50So what are we going to wear?
12:52Center of Excellence in genomics?
12:54Data science in precision medicine?
12:55You know we're trying to harness
12:57these technologies to improve
12:58healthcare and our principles are to
13:00collaborate on high impact projects,
13:01to share data and tools and to
13:03really be a global partner.
13:05To participate in these very large
13:07population scale efforts that
13:08we really need to do.
13:09To push the envelope here,
13:11but to bring all that technology,
13:13all that knowledge, all that data,
13:14all those tools to bear in
13:16our local population.
13:17OK, that's the mission.
13:18And we've got a lot of
13:20important partners in this.
13:22You know,
13:22we're not doing this alone,
13:24most notably the Yale
13:25Center for Genome Analysis,
13:26which is super important partner.
13:27And what we're trying to do Center for
13:29outcomes evaluation of her research and
13:31evaluation core is also another one,
13:32and there's probably others I left off here.
13:34They don't know about more
13:36when I've been here longer.
13:38So very,
13:38very excited to do all this just to
13:41sort of be a little bit more explicit
13:43about what we're trying to do.
13:45You know the the the one of the main
13:48goals is to sort of build shares,
13:50core technology platforms for the
13:51integrative analysis of genomic data,
13:53and the HR data.
13:54This is super important,
13:55and it's something that's going
13:57to benefit everybody.
13:58This is supposed to be a a shared
14:00resource that anybody at the school
14:02medicine can access for research projects.
14:04This includes generations,
14:05project, highly cloud project,
14:06led by Mike Murray that we hear about.
14:09Computational health platform led
14:10by Wade Schultz as part of core,
14:12and we're also my group in collaboration
14:14with Jim Knight and Y CG or building
14:17in Genomic Data Science platform,
14:18which is essentially a set of pipeline
14:21of really cutting edge genome analysis
14:23tools that are designed to really
14:25get the most out of the genomic data
14:27that we're producing here at Yale.
14:29And to make sure that all of these
14:31three things get integrated really well
14:34together to really push the science.
14:36I think that you know big goal here
14:38is to catalyze collaborative projects,
14:41focus on all the areas that I just talked
14:44about to work on with the hospital to
14:47implement these technologies into the clinic,
14:49and then just more generally
14:51to build a bigger,
14:52stronger genomics community here at Yale.
14:54From the standpoint of recruitment,
14:56training, workshops,
14:57seminars and so really excited to do this,
15:00really excited to be here,
15:02I can't wait to get started.
15:04Work with all of you here an.
15:07With that, I'll say thank you.