 What we're going to do now is we're going to summarize this planning workshop that happened last summer. This was on opportunities for GM sequencing and beyond. And I'm going to just basically just give you the summary of what was in the workshop. And then in the second part of the talk, Rick is going to kind of come up and kind of give you a little bit more of a distillation of this and kind of our additional suggestions related to genome function. And just the high level of the message that we want to get across is obviously we think that functional annotation is critical, and especially on the workshop, that tools need to be developed for this, and that we believe, of course, that we need a systematic catalog of these things. So I'm just going to start with the summary of the workshop. So the workshop was very carefully documented by NHGRI. We essentially can get all the materials, including a video cast and whatnot, from this link. And there was two main questions that this workshop focused on. One was discussing questions and opportunities addressed using genomic studies, starting with sequencing, but looking at other technologies, and then thinking about future NHGRI programs addressing these areas. Now I think one thing that's useful to talk about before we go into the details of the workshop is I think the backdrop for a lot of the discussions was this idea of thinking about a future where NHGRI is only going to do a minority of human genome sequencing, and this is a really kind of interesting future to think about. And I think the way to think about this backdrop is it's really a tremendous success story. I mean, it's a success story of developing something that really takes off, and now it's kind of, you know, kind of genius out of the ball in a sense. But the question is, the question at the workshop was what to do now, okay? And in the workshop minutes, this is what was written. NHGRI needs to position itself to positively influence the large amount of sequencing that will occur. And so that's the question of what that, what is that? And there was a lot of things discussed in the workshop, for instance involving NHGRI, should involve itself in partnerships. There's also a lot of discussion about creating exportable technologies, platforms, and standards that could be used by others looking at the genomes. And there was a tremendous amount of discussion at the workshop, this idea of scaling. And so the idea, of course, of scaling an assay to the genome, but now really scaling something to more and more experiments, really scaling beyond the scale of NHGRI into large amounts of human genome sequencing. And so that was the backdrop of a lot of the discussions. And so there was four sessions in the workshop. One focused on the genomic architecture of disease. This is really large-scale sequencing, the Mendelian sequencing. Then there was, of course, looking at the genomic function, genome variant discovery function, which I'm going to focus on. There was a bid on clinical genomic sequencing at scale, and then comparative and evolutionary genomics. And these are just the executive summaries from the minutes for the workshop. Basically, all of these areas, I think, the consensus was that they were very important areas, so that, of course, this should continue to be an important activity. It was felt that integrating genomic variant discovery with function was critical to NHGRI. For the clinical genomic sequencing, what I got was that it would require a quick evaluation of the utility of sequencing approaches to clinical implantation. And then for comparative and evolutionary genomics, the report says this is still needed to inform the prioritization and interpretation of genomic variants. And this is just directly from the report on the website. So now let me tell you a little bit more about the session we're going to focus on here, integrating genomic variant discovery function. Why NHGRI set this up is they had one overview talk by Joe Ecker, sort of similar to some degree to what you just heard, but not exactly the same. Then they had this discussion group of about 25 people, and they had listed their names here. A lot of the same people here now is from that group. And then they had two people, Rick and myself, to sort of summarize the consensus points of these 25 people. And so what I'm going to do is first give you the, essentially, the presentation we gave in Washington earlier where we summarized these 25 people's point of view. So overall, what did they think? They felt this was an opportune time to study function on a large scale, and basically the reason is we have a huge number of genomic variants being sequenced. I mean, just tremendous amounts of sequencing. And of course, the way to connect these variants to kind of function in biology is, or to biology is through function. And there's also a tremendous amount of development of new technologies for getting it functioned, such as CRISPR, single cell sequencing, and so forth. And there was a feeling through the group of the 25 that we needed a foundational resource to integrate functional information on many discovered variants. And there was a lot of discussion of what would be in such a resource. There was a bit of discussion that function occurs at many different levels. There's, of course, a molecular biochemical level. There's a kind of cellular level. And there's, of course, the organismal phenotype. And it's not clear, you know, what's the best scale to study the function at and so forth. And there was this feeling that NHRI should focus on the sweet spot. It might as well be clear what that sweet spot is, but they should focus on it. And there was a lot of discussion about where, you know, the best models, cells, mice, model diseases, and so forth. There was a discussion of a dichotomy of directions for looking at function. On one hand, we can sort of do, to some degree, what we've done in ENCODE, where we've developed this catalog of elements and all possible variants, and then intersect this completed catalog with the variants found in disease that used to interpret them. And a great example of this is Jay Shinduri gave a challenge talk at the summer workshop. We kind of talked about this very large-scale overall catalog making approach. The other approach is sort of a bottom-up approach, where we sort of take the variants that we find to be associated with disease and then, after the fact, characterize them functionally. And this is really an approach that's sort of moralized on developing the kind of genome technology to be there to characterize those variants after we find them. And the group of 25 felt these both had merits. Related to that, there was another sort of associated dichotomy between studying function with a really high throughput experimentation in a standardized high throughput way versus, you know, really going deep and really trying to understand what a specific gene is really doing. And people felt that to do the latter, you really needed to have a domain expert. A person was really studying that thing. But the group felt that this second thing was not really the province of ENCODE, at least not on its own. But the group felt that both of these things were very important, and that ideally what one would have is some sort of specialized informatics infrastructure to tie these two things together, the high throughput and the sort of domain expert. So there was other considerations that the group raised. One was that when we think of scaling, we often think of scaling to the genome, but of course there's more and more of the scaling to the population to a whole group of individuals. And there's been tremendous success in scaling a lot of these functional genomics assays to many people, such as the success of a lot of these EQTL projects. And a related type of scaling is kind of looking at a kind of personal functional genomics, looking at functional genomics of a person over time, doing a longitudinal study. And there was a feeling that this is also a very powerful way to scale functional genomics. And then a final thing the group felt was that functional genomics is valuable beyond just variant characterization. Remember, that was the charge of the workshop really to think about functional genomics in the terms of variant characterization. But the group felt that functional genomics could obviously be used to characterize cell types, develop a set of their biomarkers, and there's a very nice challenge talked by Aviva Gev on single cell transcriptomics in the human cell atlas project. Okay, so then Rick and myself presented this summary to about 100 people. There's a four breakout sessions of 25, so the 100 people, and then there was a lot more discussion. And NHGRI, I guess, wrote this down like they're doing now in their Google Doc or whatever they're doing and filming it, whatever they were doing. And then they synthesized it into their recommendations, okay? And now I'm going to present you back their recommendations, and you'll see they slightly differ from the original groups and they expand on some things more because of the discussion. So first, there was three recommendations from this whole discussion. First is that the overall question of defining the function of coding and non-coding sequences is foundational for genomics. And, you know, as we expect, one of the recommendations is we should develop and deploy assays reporting disease-relevant functions at the variant gene. Now, the key thing I make right here is the pathway level. There was a lot more discussion in the bigger group than the little group about thinking about pathways and networks and really thinking about kind of interactions, and so that's why I highlight this. Now, under this big recommendation, there was four bullets, and I show the rest of them here. And what I did is I colored these two in green because they were essentially just restatements of what was said in the smaller group. You know, we should look at different scales. We should do this function first or variant first thing. So I'm not going to spend time on that. But there was an additional bullet added here really on the importance of computational methods that need to be developed to predict the effect of coding and non-coding variants. The second recommendation really is more of a technology-oriented one. We should develop tools to manipulate genomic sequences to scale and experimentally characterize their impact. And here, I think the sort of things that were emphasized is that what we get often from these disease studies is statistically significant variants, but we really need to get it sort of biological causality. And we really need to think about ways of measuring functions so we get a causality, and that was felt as very important. And it was also felt very important at the workshop that NHGRI should really be the leader in developing these genomic technologies and should raise to the technical challenge of how to scale these things up. That is very important. So those were two of the bullets under here. And then there were a number of other bullets that really sort of restated what the smaller group just said. For instance, that personal genomics should also include personal functional genomics. And there's this idea that the assays can occur on many different scales. There's this discussion that we really need to improve our understanding of how proteins interact with the genome. Some people were making some comments in the bigger discussion that we really have a feeling naive understanding of what that interaction really means. Then under the third recommendation, this is really from the tools to actually make the catalog. So there was a feeling that we need to systematically catalog molecular components and their interactions. And of course, the key thing here, there was a lot more discussion at the meeting on this biology level, interactions level, as opposed to elements level. But there was also a feeling that the catalog of elements is not complete and additional profiling should be done. And then these are the ones that, I guess, kind of restate things again. Again, the overall summary restated this idea that sends obvious that functional genomics is much more than characterizing variance. But there was also this kind of qualification that NHGRI should limit its consideration to genetic effects. Okay, so now I'm going to hand it off to Rick and he's going to give you our discussion.