 So I'm going to talk about ClinVar and ClinGen, trying to focus on obviously the topic for the meeting, which is representing pharmacotonomics information, which I haven't ever done, despite giving hundreds of talks on this topic of ClinGen and ClinVar. So special thanks to Terry and Mary, who had a little pow-wow over the last 24 hours with me as I went through the data that is in ClinVar feedback here. So just to tell, there we go. For those of you not as familiar with ClinGen, so ClinGen is really, its goal is to create an authoritative central resource that defines the clinical relevance of both genes and variants for use in medicine and research. And we take, achieve that through both data-sharing activities, as well as a lot of curation effort around three main topics, looking at genes and their association with disease, clinical validity of that, looking at variant pathogenicity, and looking at actionability. With the goal of putting all of this information to curated genomic knowledge bases, one of which is ClinVar that I'll talk a bit about. A lot of the information about our project is on the clinicalgenome.org website, and I can tell you it's an enormous effort with a lot of people involved, over 500 people from over 100 institutions already documented in many of our working groups. So a lot of effort going on. I'm going to first start with some of the gene-level resources that we have. We've been developing a framework to evaluate the strength of evidence between genes and diseases. And that framework is shown here in terms of defining that evidence as definitive, strong, moderate, limited, no evidence or conflicting evidence. The approach that we take and the materials that we use, as well as the actual results that we are now beginning to generate, and our paper was just accepted at American Journal of Human Genetics and should be out shortly, although it is already on bioarchives. And you can go to the website. You can search on a given gene or variant, or just click on the curated genes button, and you'll come up with a list of genes, or if you search for an individual gene like was done here for SMAD3, you'll come up with what diseases that gene is associated with, as well as links, if there is a clinical validity score that has been done, and if there is any actionability that's been done by our actionability work group. If you click on the validity link, you will get the specific point scoring that's been done looking at genetic and experimental evidence with the overall conclusion of the genes role in each disease that's been evaluated. Now we haven't, we have desired not to duplicate effort with the outstanding effort of PharmGKB and CPIC, so for all of the pharmacogenomic genes, they come under the category of external resources, and if so if you search, for example, for CIP-2C19 on the ClinGen website, although there's nothing that ClinGen itself has done, you will get links to external resources, including PharmGKB, I should mention with discussions last night, we realized that CPIC isn't specifically being represented here, so we are fixing that. But you can also, in addition to PharmGKB and soon CPIC links, you can just click on the button and get all of the variants that are related to that gene through the ClinGen website, and that takes you over to ClinVar. If we look at the activities specifically around variants, there's several ways that we have been trying to support the community in terms of knowledge base, knowledge bases. One of those is to improve very interpretation simply through public interpretation sharing, really creates transparency and allows us to crowd source the work, and this is just a map of the entire world showing the density of submissions that are coming from different countries, and really to point out that ClinVar is not a U.S. database, it is a database of information from submitters all around the world, in fact from 703 submitters from 56 countries have submitted nearly a half million submissions on over 300,000 unique variants, so it's really been developing into a very large resource, and that data comes from different sources, researchers, whether published or unpublished, clinical labs, in fact 80% of data through major efforts that we've had targeting clinical labs is coming from those sources, expert groups like PharmGKB and CPIC clinics directly, patients directly, as well as integrating existing databases, one of which is PharmGKB. We also have linkages to OMIM, locust specific databases, etc., so I took a screenshot of a pharmacogenomic variant to get a sense of how the PharmGKB data is being represented in ClinVar right now, so this is a CIP-2C-19 variant, and what you can see is that it's been submitted by PharmGKB, if you click on the supporting evidence button, you will get the summary of the detail that has come out of the PharmGKB database, you can also click on, in addition to the PubMed literature references that supported that curation, PharmGKB's link is shown here, so you can actually then drop over to the PharmGKB website to get the rich wealth of resources that are there, so this is a balance between getting some basic information in this resource, so that people who download the entire ClinVar data set to use in their genomic pipelines will be sure to get pharmacogenomic variants, but it is not our intent to represent the wealth of PharmGKB and CIPIC curated data in this website, and that is to then send you over to those resources. A critical thing that Mary touched on is the importance of common standards in both terminology as well as rules for how we interpret variation, this is a list of all of the different clinical significance terms that have been submitted to ClinVar to date, you can see there are a lot of ways people call things pathogenic, uncertain, and penine, and that was a real challenge in being able to compare across different submissions, these are some other terms that aren't the sort of Mendelian terms, most of the pharmacogenomics comes under these three terms, this just seems to be a hyphen difference coming in from different sources, so a lot of effort has gone into cleaning up this process, so it's easier to map interpretations for whether they are the same or different, as many of you know we published the guideline through ACMG and AMP on Mendelian variants, standardizing those terminologies to the five shown here, that has been hugely important in our field of Mendelian disease interpretation to get a standard in place that has been now widely adopted, recently AMP published standards for somatic variant interpretation in terms of the terms that should be used for those variants, and as Mary mentioned, CPIC has published terminologies for interpreting variants at the allele function level as well as the phenotype level, and then as she mentioned, they are now working closely with the ACMG lab QA committee to develop specific guidance around the clinical significance terms that accompany the both the allele function as well as their relationship to phenotype, and Mary showed you this proposal that is now being worked on with ACMG to come out with a guideline that we hope everyone will adopt. There's also a lot of effort that we put into interlaboratory conflict resolution when multiple submitters submit the same variant to ClinVar but then have different interpretations, we know that at least one of them is wrong, so that is an area for ripe activity to try to improve resources, and we just had a paper accepted in genetics and medicine where we are able to show from four major submitters to ClinVar the ability to resolve 87% of the differences in variant interpretation, either by applying the ACMG AMP guideline rules that are now standardized or sharing internal evidence that one lab did not have access, some of the labs did not have access to, which helped resolve these differences. Often the 13% that we couldn't resolve, they often are around things like subjective interpretation of functional data, which is very hard to decide if it's real or not real, and this is where engaging experts in systematic consensus-driven interpretation is really critical. And in fact, this is how we move variants of the STAAR system within ClinVar to make them more trustworthy for our community, and as Mary pointed out, FarmGKB has expert panel status and CPIC practice guideline status, and when they submit things, it can resolve the differences in variant interpretation that are in there, so it's really critical that that expert panel review can resolve inconsistencies in ClinVar. Now, not every variant that's described in ClinVar will be resolved by expert panel, and that is why we also engage in just lab-to-lab inter-lab conflict resolution, which is very important. We have formed within ClinGEN, a pharmacogenomics working group. The intent of that was not to replicate data, but simply to liaise with FarmGKB and CPIC to ensure that their efforts were best represented within the larger genomic medicine resources that we are building, and that work group is chaired by Terry Klein and Marilyn Richie, who we'll be talking as well in a moment. And the goal is really to provide standardized terminology to guide pharmacogenomic ClinVar submissions from not just FarmGKB and CPIC, but also other laboratories that we work with across ClinGEN, submit the dosing recommendations from CPIC and the pharmacogenomic annotations from FarmGKB to ClinVar, as well as submit allele function information using CPIC standardized nomenclature. Some of the challenges that have been encountered in submitting to ClinVar is no surprise, and they have been experienced by FarmGKB and CPIC in trying to do this and trying to represent the star alleles in ClinVar is not easy, as well as representing dosing recommendations, which are often based on a haplotype or a diplotype level information where you're trying to represent multiple variants across a haplotype as well as by allelic information. So there was a need for a template and a vocabulary design for the disease associations for drugs, as well as figuring out how to link back to the resources that are really has the wealth of data and make sure that that's not missed in users of ClinVar. I took downloaded all of the submissions from FarmGKB and displayed them in four columns here. You can see there's an enormous amount of information that's already in there, 342 pharmacogenomic submissions around drug response. You can see by looking at the terms here, they range from efficacy, toxicity, dosage recommendations that are in there already. I did search within ClinVar to see who else besides CPIC, sorry, besides FarmGKB has so far submitted pharmacogenomic information to ClinVar and that list is shown here. I added the countries those data are coming from as well as looked up exactly what it is they've been submitting on. So this is my laboratory here. Most of what we've submitted is around somatic variation and responsiveness in this case to tyrosine kinase inhibitors, as well as some of the basic warfare invariants, a group submitting on GLA treatment for Fab Ray disease, prednilazone response, another somatic submission, CIP2B6, RARA, I can't say all these drugs, TPMT, PARP inhibitors, and morphine dosing. So you're seeing a variety of different things that are coming in, but largely most of the information in ClinVar not surprisingly is coming from FarmGKB, which is where we'd want it to come from. I already showed this representation of a classic variant from, in this case, CIP2C19 being represented from FarmGKB, but it's also important to remember that there is other variants that are being represented. And we have to figure out a way to display data around pathogenicity for Mendelian disease, for example, versus the interpretations that are around therapies, which there are, in this case, therapies for CFTR. And so ClinVar has separated into germline interpretations where you're seeing pathogenic for this particular variant with respect to causation for cystic fibrosis, as opposed to FarmGKB's submission that is around response to therapy, where we can display both of these types of information, but they're targeting different content. Same thing with somatic variation, where we want to be able to talk about variants that are contributing to risk for cancer, which may be different in their interpretation compared to responsiveness to therapy, which is now being shown down here in the pharmacogenomic section. So this is how we can aggregate information around the same variant, but represent different types of information that are relevant to that variant related to cancer risk or disease risk, as well as responsiveness to therapy. Not yet represented in ClinVar are the CPIC submissions. And that, as we pointed out, is because this is really a more challenging effort to represent the diplotypes and the phenotypes in there. So there is extensive work going on with Terry and Donna McGlott at ClinVar to really get this data represented appropriately within ClinVar, and they are starting with CYP2C19 as a test case to really represent this appropriately, so it's useful to the community. And with that, I'm going to stop. Thank everyone who's involved in ClinGen, as well in particular, Terry Klein, Marilyn Richie, and of course, Mary Rellen, and a shameless plug for our curating the clinical test case. So thank you for your time, and thank you for coming up. Thank you, Heidi, for being on time. We have time for one or two clarifying questions. Yeah. So in one of your early slides, you showed if somebody searched CYP2C19, then it would list these external databases. And so in that example, it shows PharmGKB first. So are you actually sorting? Actually, that's a important point. This is not completely representative. It's a long list. And I actually cut and paste from the bottom up to the top, PharmGKB, which it occurred to me that this is something we really actually need to address is the ordering of information based on what you're searching for. And that is not yet addressed. And Mark has been heavily involved in some of the development of these resources, and maybe can speak to this better than I can as being involved in representing this through info button efforts. But it is something that when I went searching for this at 5 a.m. this morning, I was like, gee, really nice if you search for a pharmacogenomic gene that those resources like PharmGKB and CPIC would go to the top. So we're going to have to think about some of the logic that is, you know, how to reorder those resources as they are relevant. So it's a good point. Yeah, so we've had an update of the website. And so in the prior version of the website, we actually did have CPIC and that being represented. It was less than ideal, but it was coming up. And then when we redid it, we are now in the process of fixing things. And so one of the things that's critically important for this community is to give us feedback about how this resource is working and what we can do to improve it. Because if you go to the site, there are a huge number of resources and it's impossible to kind of keep track of everything. And so if all of the all of us in different content areas can go in there and make sure that our content is being represented in the most effective way, that's the way we can make the ClinGen resource the best. And then the other thing that we've been doing and for those of you who are have any touch of your electronic health record system, we do have the ability to link into your EHRs through something called InfoButton, which is a meaningful use standard. So if you have a certified EHR, you can turn on InfoButton functionality. You can then create a link to ClinGen around certain things. So if you have a drug where you wanted to represent pharmacogenomic information, you could actually add an InfoButton link to either the CPIC guideline or the ClinGen resource to lower the barrier to that access. And so if any of you are interested in standing up that type of an InfoButton link from your electronic health record, just send me a message and we'll make that happen. So just following up on what Mark just said, so to implement something like that, is it still going to be a major hurdle to overcome inconsistencies in the allele nomenclature in order to do that? Is there enough standardization in the community? And how these things are reported and represented in EHRs that it would make that integration straightforward? Or is there still a lot of manual curation and standardization that needs to be done? So it's a good question and I don't know that I can completely comment on that because there is so little pharmacogenomic information that has been submitted to ClinVar except for from PharmGKB that it's hard to get a good sense. And as you saw, that really spanned a lot of different topics from treatment of diseases like febrile and CF and somatic cancer, where it's slightly different in terms of how we think about representing that data. So I do, having watched the uptake of the ACMG AMP guideline for Mendelian disease variation be just enormous and widespread. I do feel that with CPIC coming out with guidance that that will, you know, one of the reasons that Terry and Mary reached out to ACMG is to really ensure that that guidance is coming from a series of authoritative groups so that there's no question that this is the standard recommended by CPIC, PharmGKB, ACMG, AMP, etc. And then in my mind that really creates the motivation for groups to change their systems and adopt that standard. And the, you know, having taken all of the Delphi surveys that were done for the CPIC approach, it really was an incredibly thoughtful approach to ensuring the community voice in the development of those standards. And a lot of us running labs and being the ones who actually write reports on this information, we're a part of that. And we did the same thing slightly differently, but with our ACMG standard where we did surveys to the laboratory community, we held open workshop sessions at AMP and ACMG, and we made sure we got the voice of everybody. Because if you don't get that, then when you put your thing out and everybody feels it was developed by six people, then people are like, I didn't have a voice, I don't think they did a good job, I'm not going to adopt it. But I think with the efforts that are being done here, both through the Mendelian as well as from a co-genomic, I do think they're going to be adopted in my opinion. But time will tell. And I don't know if Mary or Terry or anyone else has more understanding. It could comment on that. So I think what we'll do is keep this discussion alive in the, in the discussion section. But that was a great segue to the next topic, which is digitized. And Sandy Aronson is going to speak on that topic.