 All right, welcome back, everyone. I'd like to talk today about progress and plans for strategic plan for the NHGRI Genome Sequencing Program. Now, you've seen drafts of this before, and I think it's moving on towards time to begin to implement it. And I just want to start with an overview of where we are now with the program. Some of this has come up, came up in Dr. Green's talk this morning. The Genome Sequencing Program has these four components, which Eric described. These are the current budgets, the large-scale sequencing and analysis centers. It's at this per year, and headed down. You'll see that in a second. Centers for Mendelian Genomics at $12 million a year. Some of this is from NHLBI Co-Funding, Clinical Sequencing Exploratory Centers, CSER, it's 19.2, and some of this is from NCI, Genome Sequencing Informatics Tools. And the program was funded for four years in early FY12 with the second round for CSERs in 13. This just shows the amount of funding over time. This is in year-by-year dollars, this curve is year-by-year dollars is actually coming down in 2013 dollars from over $200 million of funding. Down to its current 80 and a bit, and headed down by 5% per year, and in red bars is production. It's not that there was no production in these years, you just can't see it compared to this production. And everybody knows what happened around here. And this boom, and even with this very significant drop in funding from the last time we renewed, there's only been a bit of a drop in sequencing, and actually it looks like we're on course to go start going up again, but I will reserve judgment until I see it. This is just two different plots, two very similar plots. This one is the last four quarters, this one's just this quarter showing production by project type, and at this point we're mostly medical sequencing, and a lot of cancer sequencing, a little bit of organismal sequencing, some thousand genome still going on in the last quarter for which we have records, we're actually into the next quarter from this, but this is where we have complete data. And it's pretty stable at this point in these proportions. There are a lot of active sequencing projects at large scale. There are in the last five quarters the program has touched 21 complex disease projects, 16 Mendelian, I put them in quotes because they're not all strictly Mendelian, there's kind of a blur between complex and Mendelian, 33 cancer including ATCGA. Many organisms, well over 100 organismal projects touched at least in the last five quarters for microbial and thousand genomes, and I just want to, you can't see all of these, and this is actually a really nice live table, you can open some of these up and look at everything you did, and we'll eventually post something like this in a better place so anyone can take a look at it. But I just want to point out some of these larger ones, 47 and 43 deposited past megaterobases for those, and that's a schizophrenia project, and diabetes multi-ethnic cohort project, those are two fairly recent very large projects that would not have been possible five or six years ago, and of course this is TCGA, which is a composite of eight projects right now, which is over 58 terabases at the time that these data were collected. I should go back just a little bit and say, there's two things I wanted to say. I'm obviously talking about the large scale, right now about the large scale sequencing and analysis program, but throughout this talk I'll be sort of moving back and forth between other elements of the program and large scale, but I'm going to send a lot of time talking about large scale, because the large scale component is the one that is in some ways deliberately least defined by its grant proposals, I mean it's obviously carefully planned, but designed in order to take advantage of opportunities as they change and they change very, very quickly. That said, at the beginning of the talk, you'll hear me talk about the current, this whole thing in terms of the current structure of the program, but as I move on, the talk will get more and more agnostic about the current structure and you'll see why at the end. So there are many different project types at multiple scales. These projects were chosen in a number of ways through community input, collaboration, center initiated, we initiated some of these. They are all vetted by NHGRI staff, by our sequencing advisors, some of whom are here today, and for many of the larger projects by council. And what is important and possible and appropriate for the program all change rapidly. So really because of this last point, we're constantly changing, and even though I'm giving a talk today about a new plan, in fact this is a continuum with the kinds of planning we've just been forced to do in order to keep the program staying, keeping ahead of the opportunities, and we just have to do more than predict these opportunities and stay on top of the changes, we have to actually try to drive them, that's the whole point of the program. And that led us to disease 2020. So I think we had come to over the past couple of years with all the increase in sequencing capacity and the increasing attention on the medical sequencing and complex cancer and then complex disease. We asked the centers to draft a proposal about the best opportunities. And these are opportunities explicitly from the point of view, from an ingenomics of human disease, from the point of view of genome sequencing. Again it's a strategic framework for genome sequencing, initially put forward by the large scale sequencing analysis centers. We had a workshop for input in January, which included representatives from all the center, from all the sequencing components, actually from CSER and CMG, and others. That output of the workshop was refined, including there was one round of council comments already. And also went through the efforts of an editing team, who you'll see at the end. And just to say that we intend to use the resources of the program, of the sequencing program over the next 18 months to demonstrate the feasibility and the value of the disease 2020 plan that was proposed. So what is it and why now? We're discovering genes underlying human diseases as fundamental to what we do and it's a fundamental scientific importance. There's been a lot of progress in terms of technology and knowledge, but we're still at the beginning. I think everyone realizes that. We're far from knowing what we're doing and it really is time to use the tools that we've developed to systematically define the basis for human disease. If there's one thing that I take away from disease 2020 is sort of the realization that doing this is now almost entirely technically feasible. It's arguably or not more or less practical. It's not that every problem has been solved to doing this, but by and large it looks technically feasible to do that. This is not a project that can be done solely by NHGRI, obviously. I can only talk today about the part that we want to do, that we would like to do. You'll see immediately it depends critically on other ICs. That's where the samples are. That's where a lot of the domain expertise is. And we can't move very much farther on this without many strong collaborations. Fortunately, we already have some. However, we're in a good place to lead this just by virtue of our history with large-scale sequencing. So I'm going to apologize here for the use of the word domains because it's been used in another context today. I don't want you to get confused. These are not the same domains as our NHGRI strategic plan domains. It's five domains of D2020. So the first is to identify the key genetic factors contributing to 100 important common diseases. It's important to have a big, ambitious number here. I think you'll see why, and a big round number. Identify the genes underlying essentially all Mendelian diseases as domain two. Domain three is identify the genes that drive cancer initiation, progression, and treatment response in all significant cancer types. Domain four is identify microbes and their communities that cause and correlate with disease. And five is to really push harder into genomics in the clinical and healthcare study, which we're doing, obviously, you've heard of other programs discussed today that are doing that as well. So domain 100 common diseases. Then GWAS has discovered hundreds of loci. There's a lot of misinheritability. Causal variants are largely still unknown. Sequencing approaches directly assay the full range of these. So full range of complexity, heterogeneity, the full spectrum common to bear of variants. There's plenty of places to start. There are projects already ongoing. I showed you that long list, and you obviously didn't have time to look at all of them. It was too small anyway. But we can start with these are ongoing projects. There's an autism project, Alzheimer's disease project that we've talked about in extensively previous councils, early onset MI, schizophrenia is another big project, several cancers, chronic kidney disease. There are also disease endophenotypes that could be subject to study. Importantly for many of these, we know that there are sufficient samples to do these. And I can't emphasize the samples stuff enough. It's the major practical constraint to doing a lot of these projects. In order to get power, I mean, you've seen the Alzheimer's disease plan already, and that's 10,000 cases plus controls. Other complex disease studies are likely to need samples like that. People ask me really 100 diseases, complex diseases that we could even go after, or that could be gone after. This is a slide from Terry. You can't see all of these, but these are traits with published GWAS studies. So, 331 of those. Some of these are not disease traits you'll notice immediately, but there probably are 100 in here that could be picked out. I did a different cut of these. I just went back to the data from the GWAS catalog, and Terry is more of a lumper than I turned out to be. I split these. I got more categories, but I mostly just wanted to see where the GWAS explored territory, where it might be a good place to go. And I looked, the blue bars are actual disease phenotypes, and the red bars are endophenotypes. You could argue one endophenotype might really be a disease phenotype. And I divided up amongst a number of categories, disease categories, and you can see there's an awful lot in pharmacogenomic. There's a lot in cancer, and this is heritable cancer that could be gone after, heritable aces of cancer that could be gone after. A lot of sort of neuropsychiatric, neurological. It's an interesting area. And autoimmune and allergy, I think we've got a number of studies in cardiovascular, but there's still more to be done. There are a lot of interesting non-disease phenotypes. Some of them are more clearly related to health than others, and I think those would have to be looked at quite carefully to figure out what might be profitable there. This is a point again about adequate power, we're gonna need thousands of samples to learn about how to best to do all of these. We don't just wanna do one, one or two, we wanna really learn and establish how to nail these down as best, the best ways to nail these down. So this is kind of a key idea here. We need to do comprehensive projects. We have, and others have a number of initial projects over the last few years as permitted by sequencing capacity in samples that are sequencing a thousand here and a thousand there. And depending on how they design them and what the disease architecture is, they're learning things. But these don't look, a lot of these don't look like comprehensive studies, at least, at least right now they don't have comprehensive studies. We also need to try diseases with multiple architectures to understand what the differences in design might be. Very likely these would be multi-component mixed studies. For example, they would include case control, maybe families, maybe some targeted sequencing, whole genome, and others. We would explore both case control and cohort designs. This is an important area. We need to pilot large scale studies where recontacting and re-phenotyping is possible. It's a real limitation of current sample sets that you often, of many current sample sets, that you often can't do that. And finally, we need to push a transition from whole exome sequencing to whole genome sequencing. That's in the midst of happening right now. I'm sure you've heard many things about this, about the cost of whole genome sequencing and the best possible cost. And the advantage is it's not just cost limited. Another major, probably one of the biggest technical hurdles that at least I can predict right now for whole genome sequencing is the analysis tools are lacking for analyzing whole genome sequence. You'll hear a little bit more about that in Lisa's presentation. In domain two, genomic basis for essentially all Mendelian disease. This is technically feasible now for many Mendelian diseases, especially the ones with clear patterns of inheritance. The centers from Mendelian genomics and other groups are already doing this and will continue. There are opportunities to expand this to invest more complex modes of inheritance, so disorders that are Mendelian, but with high allelic heterogeneity, looking for de novo mutations, screening in large populations. Another opportunity to explore this genotype to phenotype design if you can re-contact and re-phenotype patients, sorry, participants. In domain three, cancer initiation, progression, and treatment response in all significant cancer types. We are well along the way with TCGA, and this would continue in alliance with NCI. We are in regular contact with them about this. They were at this small meeting. They had some direct input into the D2020 document. But there's more to do. So completing the catalog of characterized tumor types for primary tumors. We know very little about progression in tumor heterogeneity or treatment response, and there are a number of animal models that could be characterized. In domain four, microbial basis of disease. We know this is important for disease. Establishing causality remains a major challenge. According to the plan in D2020, we would pilot, at least to begin, with a small number of studies in the context of complex disease. We frequently get this question. You're looking into so many complex diseases, why don't you just take one of those disease studies and have associated with it a microbiome project? And the problem again is that the samples very frequently aren't, they're not sampled. Microbiome is not sampled. You can't go back and sample over time, which is really the kind of thing that you need to do. So because samples are so limiting, we have practical worries about doing a lot of this in a short period of time. But we have no question that it's going to be important. In the meantime, there's still lots to be done developing reference sequences and tools, including analysis tools, for detecting microbiome components. Domain five, sequencing in the clinical and health care setting. And this is clearly where a lot of us want to be. The CSER program and other programs at NHGRI are already addressing this. There are opportunities to push farther in this direction. And here I'll be a little bit agnostic about the right way to do this. But at scale, there are certain kinds of things that you can do to combine clinical and discovery. And this needs a lot more thought. But some examples, newborn screening, combined with allele discovery or tumor sequencing, along with to guide treatment or assess response, along with gene discovery. And there are other opportunities along this line. So implementation is kind of in three phases. So one is what we need to do now. In fact, what we need to already be doing. And this is this notion of demonstration projects. And you especially saw some examples of demonstration projects in that domain one. We can get going right now. In fact, we already are going with large domain one, two, and three projects. We can sort of reframe the ones we're already doing into these demonstration projects. There are still, as I showed you, a lot of smaller projects. There are clusters that are related. There's probably a productive way to consolidate some of those small projects and sort of move them towards being disease 2020. But this needs a lot of detailed discussion about the science. Again, lots to start with already schizophrenia, autism, Alzheimer's disease, diabetes, a few others. The next stage, over the next roughly 18 months, we really need to get additional input, especially from other ICs at NIH, about what large sample sets there are, what else is being supported about possible collaborations. Again, we're already with all the large projects. Any large project I showed you has already required such collaborations with NHLBI, with MCI, with NIA. We would like to start identifying additional projects, especially in domain one. We need to start getting the demonstration project data. And I think this is already coming from schizophrenia and type 2D genes and other large projects. And think about how we would design pilot projects for domains 4 and 5 and begin them if possible, sometime over the next 18 months. 18 months from now, we need to come back to all of this and assess feasibility and progress from the demonstration projects. Are we really able to, I mean, both the practical and the scientific? And ask again, are these goals still well justified? And then ask ourselves, what's the right program structure to achieve this if we believe that this is feasible and well justified? That's a lot of value. So this is an overview of the implementation timeline. This is now start demonstration projects. Community input, gather data, seek new projects, design pilots for the domains 3 and 4, and I should say possibly 5. Sorry, it should be 4 and 5. Can't count. In 2015, we will have a meeting to ask the question. Is D2020 still justified? Is it feasible? And what's the best way to do it? And in 2016, the entire sequencing program is slated for renewal. And this will be the input into how this is aimed and structured. With that, I want to acknowledge all the folks who have helped out with this. So from NHGRI, all these folks especially want to thank Deborah for help putting together some of the data slides and Chris. And these folks were the domain editors for the D2020 document. And Carlos is here today. And with that, I'd like to open it up for comments and questions. I'm going to start with Didi, since you've been here for most of this. Well, I think you gave a good overview and kind of history of how we got to this. Baby just mentioned that one of the things was to, in light of the strategic plan for genomics, and then taking advantage of the sequencing centers, how can we best utilize all these resources and capability. So I think this is a good direction to be heading in. And really, how the implementation is done is critical. I like the idea of the demonstration projects. I think doing somewhat the what's in the pipelines now is good, but also maybe developing a bigger, longer term demonstration project could also be very useful to really demonstrate how we can get a better understanding of a disease and what we can do about it. So thanks. And Carlos? I think you did a fantastic job of encapsulating what's been months and months of calls and meetings. So this went through many, many, many different rounds. And I think the final document really doesn't encapsulate what's a pretty exciting vision for what can be done with sequencing to address public health concerns and disease across the spectrum, given the reality of constraints. I mean, in an ideal world, of course, we just do massive whole genome sequencing, and it would all be integrated in electronic medical records. Given where we are now, what can we do? And given the collections that existed. I think the idea of leveraging existing investments across the NIH into other cohorts is such a no-brainer that we need to figure out how to make that happen. Yes. I'm going to be the devil's advocate, recognizing that even the devil's advocate sometimes has sympathy for the devil. So from being outside this, I can sort of look at it and listen to it, this is really the first time that I've heard this presented. So I worry about a couple of things. One is I worry about that what we have is a large infrastructure looking for a mission. And it smacks a little bit of that. Number two is I'm very concerned about using existing cohorts and samples as convenient samples that really lack the deep phenotype information that we need to really understand this. I'm concerned that this is going to take us the same place that GWAS did, which is some good solid information, no question about it, but with huge questions left at the end, unless we also add in the epidemiological aspect of this. And third is I do have a lot of sympathy with the idea of not having the domains all separate, but try to have cross-domain studies where we do look at, for example, at autism and the microbiome, or Parkinson's disease and the microbiome. Because I think we're missing a huge facet of the pathophysiology of some of these disorders if we don't combine these domains. Yeah, so let me start from your last point first. So I agree with you. I think it's not just complex disease and microbiome. It is complex disease and some of those other areas. There's plenty of opportunities. And the way that I stated this and the way that the 2020 has written, it says that they're not supposed to be clear boundaries and that there's plenty of opportunities, but it doesn't really demonstrate it by giving examples of kinds of projects. Maybe if you get to the higher domains, if you look in the medical, in the clinical, it talks about more explicitly. I think there are potentially those opportunities. What I'm worried about is if there are not the samples, what needs to be started today is getting those samples. And we already have a large-scale sequencing program and a more than just large-scale sequencing program. And it's going until 2016 already. They're already engaged in some of these. So this is not self-preservation. This is what's already going on. This is what's already funded. And it does have to get better definition on the question. And I agree with you. It is hard to separate the question from the current structure. But we have to do that. And we have to concentrate on whether the goals are possible. And we have to explore issues like, are there the right phenotypes out there? I know that from the Alzheimer's Disease Project, there's been a lot of thought. I mean, literally a year of practically weekly meetings talking about getting the right samples. And much of that has to do with the deep phenotyping. I know the same discussions have gone on with, and this is a practical, during the demonstration project phase, this is a very practical and critical issue to get ironed out. Again, to go back, yeah, when we get to the stage, we really have to ask, is this still the right focus? And what's the right structure to do this? I want to point, I'll get to Pamela in a second. I want to point out that, for example, in Domain 5, I don't think that there's any single component of the sequencing program overall that's ideally suited to do some of those things. So I think that needs some real thought. Pamela? Following up on your focus on the program assessment in 2015, have the criteria by which the program will be assessed been developed? No, we have to do that with the help of you guys on the SAP. Yes? You've chosen some very interesting diseases to start with, and I would say those that have been refractory to many other genetic approaches like schizophrenia and autism. So there's always an argument out there what's causing the misinheritability. So I hope that there will be pilot demonstrations with the exome and whole genome sequencing to define how much extra yield you're getting for the investment. Because if it gains another 5% or 10% of cases defined by DNA change, fine. But I'd hate to see the herd effect of everyone jumping off the cliff that we have to sequence thousands and thousands more with whole genome sequencing because the misinheritability is out there somewhere, rather than pausing and trying to decide whether we got our assumptions wrong or there's other biological mechanisms we need to identify, such as whether there's a microbiome interaction. I've seen this at every stage of the genome. We either blame the technology or we blame the phenotype. But the question is that we really got our assumptions right about the heritability of these disorders. Yes. So I understand why you can't go and recruit the right cohort, if you will. So I wonder whether you've reached out to things like the UK, the large UK study or some of the other initiatives where you can get part of what you're missing in that way. So they're doing their own thing. They're doing it in a very different way. And in some ways, it's complementary because you're going to be digging deeper and have better phenotype for each disease. But they're going to have the broad into phenotype and other just broad data. So it seems like that might be a way of getting both without having to spend the extra money. And I know it's complicated, but it seems like a way to consider it. Yeah, I agree. And we're going to have to take advantage of every opportunity that there is like that, including internally there's emerge as well. Terry, do you want to talk about UK Biobank? Internally as well as with the sequencing centers about UK Biobank, which is 500,000 people, characterized fairly extensively at baseline and then followed up in their medical records. There was not a lot of enthusiasm for pursuing it. Rick may want to comment on that as well. And we've continued to be in contact with the Biobank folks mainly to make sure that the samples and data would be available. And they said, absolutely, there's no reason that we couldn't post these either in dbGaP or have a link from dbGaP to the Welcome Trust. So it seems like a real opportunity to be able to do this. Yeah, I mean, you would come to the party with something. Yes. And right now, you're not coming with anything and they don't have anything. Once both parties have something, then you can put it together. Yeah, and they're embarking on an effort to do the exome chip backbone on the Biobank cohort, probably not the entire 500,000, but a sample of them initially, and then use that to impute if they can sequence a smaller proportion. But they are getting some funding there, and it would be awfully nice, I think, if we could be involved in that. I'm sorry, the lack of enthusiasm was from whom? I won't say from whom, but I will interpret about what. So it's sort of related to the same sets of kinds of questions that people have been asking, including your question. If the point is just, is only to sequence without regard to sets of individuals that could be informative because of their phenotype, there's not enthusiasm. I think in general, if there is a way to, for example, consolidate some of the phenotypes in similar phenotypes and there are sufficient numbers, there would be enthusiasm. And I think that you also, the degree of enthusiasm that you get, of course, depends on exactly who you ask. So can I interpret what you're saying, then, is two things. One is that it's a cohort, not a case control. And secondly, that it's EMR as the source of phenotype, as opposed to a scientifically rigorous phenotype. Or is that? I was saying something less than that, actually. I was just saying that the prospect of simply of contributing only capacity to that, rather than trying to ask some of these questions with that seemed to be less attractive. Well, I mean, at the risk of conflict of interest, there is the Northern California Kaiser Project. And I think that should also be considered as a source of material. Well, well, over 100,000 people. I'm not involved in it, but people I work with are. So I wanted to mention that. Yeah, Carlos. Just to make a small addition to the UK Biobank, which I think is a great resource, one of the points that's laid out in the document is the need to have a portfolio of different populations and the importance of having a multi-ethnic design. And so that's one potential population. But of course, you'd need to make sure that we'd had good representation, particularly of ethnic minorities in the United States as part of this. And I think the other point is to think about how this is really one of the first starting efforts to do the large-scale sequencing, to drive down into rare variants with the realization that doing rare variant association is turning out to be far harder than we had expected. And so in order to do that correctly, we need a concerted effort. And if we can get a portfolio of diseases under a systematic sequencing approach, then we do a better job of maximizing our chances of finding the set of associations as opposed to dividing up what could be the sequencing budget or whatever budget you want to invest in this across a set of uncoordinated projects, if that's the alternative. So I think that's part of the rationale. And I think the driving idea here is let's not let the perfect be the enemy of the good. This is a logical thing to try. And we'll see over the demonstration projects how it works and then take a mid-course correction. Sloan. Question for Terry while you're still up here. How far are we along, I propose, this very context, driving eMERGE for G2P recontact? So in eMERGE recontact is generally no problem. There's one cohort that it can't be done. Is it fed in here then? Yes, so that is part of the discussion as well. And it could easily be done. It could. It could be. It could be. I mean, you have the resources, and you need to recontact the people and bring them back in. But for the most part, these are people that are undergoing care at the various eMERGE centers. So they're already engaged, very much like the Kaiser cohort, which is people undergoing care at Kaiser. Pamela? Reading over this version of the report, a phrase that comes up repeatedly is large numbers, large numbers. We have to have large numbers. And that's across the write-up of domains 1, 2, and 3 that's emphasized. And I know that, I guess it was at the January meeting, there was some data that actually had lists of where the samples were and how many they had. And there were certain conditions that seem to have already a pretty robust set of samples. But I guess what I'm wondering is, has there been in the last several months an attempt to go through and to really identify where those samples are? I mean, so talking about, yes, you could do Kaiser. Yes, you could do UK Biobank. But if you're talking about an 18-month timeline, presumably you're talking about samples that we already know are quality samples and that we have access to. So was there an assessment that actually went through and looked? Not for individual projects, yeah. So in particular for Alzheimer's. But no, not for others. And we really need to have that discussion and a number of others. Yes, David. I agree completely with the importance of all the goals that are laid out. Two real questions. One is overlap. So let's just pick Mendelian disease as an example where there's separate investments being made in Mendelian disease centers. And it just wasn't clear to me how the Mendelian diseases picked for part of 2020 would be delineated from those that are ongoing at the other Mendelian center. So I don't think there's a need to delineate. I think all of them get reframed because the problem gets looked at at a higher level. And all of that data are useful. I think that the question that I've asked is actually a level more detailed than that, which isn't whether or not this should be a goal, but actually has to do with the dynamics between the CMGs and the LSACs and what's appropriate for each. And believe me, Lou and I talk almost every week about what the right boundary is and where the real opportunities are for expanding into something that is appropriate for the large-scale centers. But who does it in some ways as the next level of detail down, that it should be done in defining what kinds of projects those are is more the point. I guess that's related to my other question, which just does have to do with timelines because 18 months of very short timeline, you mentioned there was almost a year of discussions about the Alzheimer sample cohort. And that's one thing as an example of how long it may take to, I realize that was a particularly problematic one. But it just doesn't seem like there's very much time to identify the targets, collect the samples, and get the data in a way that is going to make it possible to inform the 2015 decision about how best to proceed. So I agree with you. And that's why it's good that we have some large projects that are already underway. A lot of the smaller projects probably aren't going to be adequate on their own. Maybe if some of them can be combined and analyzed together, we'll get some additional mileage out of those. But I agree with you. Even if we had all the samples today, I'll give you an example, Alzheimer's disease. It's not on that table because at the time I produced that table, there's no data. Data are just beginning to come out. By the time, by 15, 16, 18 months from now, we'll have all the data. I doubt we'll have a lot of the analysis done. Maybe for the family samples, we'll have the analysis done. So I agree with you. It's a very short timeline. And we're going to have to make the most out of the projects that we already have, like Schizophrenia, Autism, Type 2D genes, where the data is already collected, essentially. We're going to have to get a lot of mileage out of those. We'll have to assess it every step of the way whether we're giving this a realistic test. We're not going to be, probably also the only ones, doing projects like this. And so there may be information coming in from elsewhere. Or else? The projects you just mentioned, the Alzheimer's and Schizophrenia and so forth, you and others around the table no better than me have been almost intractable. And in 2015, there's going to be a program assessment based on whether or not the sequencing. I'm wondering, are you setting yourself up for an extremely difficult time at program assessment? Is this realistic? I mean, sort of in general, in favor of what you've been talking about, but the most recent discussion is bringing all kinds of concerns to my head. Yeah, I'd like to get into this more in the closed session when we talk about implementation in more detail. But yeah, it's been noted by others. All right, thank you, Adam. Thank you. So that's it for the reports and presentations. We're now going to move on to a concept clearance. This will be from Lisa Brooks, program director of the genetic variation program. Concept is titled, Interpreting Variants in Non-Coding Regions of the Genome.