 Okay, welcome back everybody. We're going to continue with our open session agenda. I'll turn this over to Rudy Thank You Eric So we're going to immediately get into three concepts that are going to be presented to the council and I'll remind you that before NHGRI or any Institute NIH can publish a funding opportunity announcement We need approval of an advisory group that's done in an open meeting We always use the council so that you are always aware of whatever opportunities are being published by NHGRI So we're going to hear a presentation from the two members of the staff For the first concept we encourage a discussion lots of questions from council and at the end of the discussion I will ask to take a vote to approve the concept So the first RFA is so the first concept rather is title Consortium for understanding the impact of genomic variation on genomic function and the presentation is going to be a tag team with Mike Paisen starting and then Dan Gilchrist, correct Thanks everybody. We're going to tell you about a draft concept first I'd like to check can people on the phone here and can you see that we have a title slide up? Actually that reminds me Joe Ecker. Are you on the phone? Yes, can you hear me? Yes, we can so Joe Ecker is joining us for this particular concept He's a former council member at the Salk Institute. Welcome Joe Thanks, Rudy You can see our title slide up there. Yes Okay, thanks. I gotta go ahead and proceed. I like to start off by thanking a number of people obviously this Is a big effort required a lot of contributors. I'd like to thank NHGRI leadership I'd like to thank the NHGRI function team program directors and analysts that work on this Thank the we're in the division of genome sciences, and I'd like to thank the division But also we've gotten a lot of an input from a key workshop that we had genome to phenotype workshop And I'd like to thank the participants there as part of the parts of the discussion and letting us hear about Different kinds of ideas that we could incorporate or not So I'll give you a little bit of background Tell you about where this comes from why we think this is an important area of science to put a have a major undertaking Then you'll hear about the proposed scope and objectives. What is the concept that we're interested in putting forth? That's what we would like to hear discussion from council members on and you'll vote on later We'll also tell you a little bit about how we think that this would relate to other ongoing activities If it moves forward at the end we'll have to counsel discussion, and then we'll also have a formal vote So I'd like to begin by quickly showing you what is the structure of the program that we're going to talk about So that's not a surprise later We expect to have these five components And we think it's very important for them to work together as a group and help each other out with their different science And they're going to cover different aspects of how genome structure determines genome function in different ways they'll get at in different they'll get at different aspects about how variation affects genome function for instance We'll have centers that will look at how Elements function will have centers that map activity of different elements will try to put elements and variants into networks and pathways We're interested in doing integrative analysis and also predictive modeling on the data as it arises And then have a data coordinating data processing center So so the background where did we get how did we get to where we are now? You NHGRI strategic planning process you heard this morning started out in February of 2018 that's when we first previewed it with the council and Community input is very important to NHGRI. It's very important to NIH as well hearing What are the needs of the scientific community in order forward? So we've been looking through and here and hearing input from different venues a Very important venue for us has been this genome phenotype workshop this meeting took place in January of 2019 and Report on it was presented in council in May of 2019 I should say that the proceedings of that workshop and also the council report are archived They're available to anyone that has access to internet. They're freely available So the focus of this workshop was the very general problem of how does Genome changes in genome structure genome sequence relate to phenotype It was not focused on functional genomics yet We got a lot of important info about functional genomics at this workshop and The concept is built in large part on feedback that we got from this workshop though also other strategic planning venues So the workshop was planned around for different scientific topics again Not a functional genomics workshop and one thing that struck us is across the different topics We heard a strong need to do more functional genomics a lot of ideas emerged in different areas of the Different topic areas of the workshop calling for more functional genomics continued investment in functional genomics So the pro the concept that we're Presenting to you today the proposed the concept proposal is intended to address a number of different scientific challenges The key is this overarching idea of how does genomic variation change genome function? All right, so no we're not going to solve that in just five years But that's the rubric that we wish to work under we'd like to go from Today we have a lot of association studies Can we get closer to having a causal understanding of what variants that are associated with disease or phenotypic changes do? We'd like to improve our ability to interpret variants so in some cases we have variants with no interpretation in some cases we have variants Associated with the trade or phenotype often. We don't have much understanding of how they're mechanistically connected and Because that space is so huge Looking at the number of possible variants the number of possible phenotypes number of possible cell types We don't think that in the near future Let's stop there because it's harder to predict in the far future They don't see a way in the near future that this will all be experimentally tested So I think it's very important to develop approaches that one can do predictive modeling and make predictions about unseen variants Made to provide some idea of the scope if you were to look at for instance the NHGRI GWAS catalog You'd find something like 70,000 associations from GWAS Few of those have been characterized or if you look in nomad 3 you'll find some 600 million snips Some 100 million structural variants few with any of those it's known whether they're associated with any phenotypic change or disease So we think that this is an area that's at the forefront of genomics. It's very important We think NHGRI is positioned to make an important contribution here building on our past efforts So we want to continue perhaps increase our effort in this area It's important for all to understand that in our view this concept Is only a portion of our area of our work in function area for example following this You'll be hearing about a concept looking at developmental gene expression another function on concept We think that this is going to be a foundational part of the strategic plan. We heard loud and clear it From input that functional genomics is something we should be working on now And that's why we're bringing you this concept now rather than waiting for the strategic plan to be complete Last council round you also had some concepts brought to you on the same premise So I'll stop there and transfer it down Thanks, Mike So with this proposed program NHGRI is taking on both a challenge and an opportunity as Mike just spoke about Identifying disease associated variants is now more or less routine, but interpreting variant effects understanding how they impact phenotype Is a major bottleneck. So the overall objective of the program we're talking about today would be to improve understanding of how Genomic variation impacts function and phenotype to enable significant advances in our ability to Interpret genomic variants and really help relieve that bottleneck. So I'm going to talk now More about the implementation that we envision this will involve components of both research and resource building and include data collection analysis and predictive modeling So we're proposing an initiative Composed of interrelated ideas that would build upon decades of work by the genetics and genomics communities and Also bring together threads from a number of existing or former programs Systematic genome perturbation and characterization has been a central part of the common fond links program And it's also been an important part of the current phase of the end code project Systematic identification of functional elements and gene activity has been central to the end code project the roadmap epidemics project as well as GTX and Predicting predictive modeling both of regulatory networks and Of variant causality has been explored in the genomics of gene regulation program funded by NHGRI along with the Non-coding variants program so work in each of these areas. We think would be Important and integral to the program that we're going to be proposing So as I mentioned the overall objective of the program is to transform understanding of how variation impacts function and phenotype This would begin to be addressed through a number of interrated interrelated activities Testing the impact of genomic variation on functioning Determining where and when regulatory elements and genes are active with single cell resolution Exploring the roles of specific sequences and regulating function in networks developing computational approaches to model and predict relationships among variation phenotype and function and Through establishing a resource to enable future studies in this area by the community as Mike touched upon earlier In the proposed program each of those five Interrelated activities would be focused within one of five interacting components Functional characterization centers would experimentally test how variants impact function and lead to phenotypes Mapping centers would identify where and when functional regions of the genome are active with single cell resolution regulatory network projects would advance network level understanding of the impacts of sequence and function on phenotype predictive modeling projects would develop and test computational approaches predicting impacts of Genomic variation on function and phenotype and a data coordination center would generate a resource of data tools Models to advance community investigation in this area in the future So in a moment, I'm going to talk about how we envision each of these components working in more detail Individually and also how we think they could synergize to reach Or work towards the program's larger goals so The program would include both research and resource building activities and we envision each Component of this project to exist somewhere on a spectrum from those where the science is most mature and where there's the least risk involved to those where the science is at an earlier stage involving more risk, but also potential for Excuse me innovation and reward so mapping centers and data coordination centers are envisioned to exist on the Resource building end of the spectrum dependent upon more mature science and Functional characterization centers are envisioned to contribute significantly to resource building However, we recognize that the science in this area is less mature and so these Components this component will include significant research activities as well At the other end of the spectrum regulatory network projects and predictive modeling projects are envisioned to be Research-driven sort of pushing the boundaries of our understanding in some ways similar to NIH R01 research project grants But nonetheless making important contributions to developing a community resource So now I'm going to go into a little more detail about each of these components and the goals that they help address To improve understanding of relationships between variation function and phenotype characterization centers would systematically apply high throughput genomic perturbation methods testing the impact of variation on either protein coding elements or non-coding elements or both As saying impact of genomic variation on molecular cellular and organismal phenotypes Importantly these centers would catalog the results the tested variants and their impacts and share them with the research community Because we're still learning about the best ways to test the effects of genomic variation Refining generalizable approaches to perform these tests and catalog the phenotypes would be an important part of these centers To determine where and when regulatory elements and genes are active with single cell resolution Mapping centers are proposed to obtain biological samples of high value to the community and the consortium to establish multi-omics pipelines for high throughput mapping of functional regions at single cell resolution while preserving information about biological and spatial contexts of those activities and in doing so identify cell type specificity of regulatory elements and link them to specific genes to explore the role of sequences in regulating networks and establishing phenotypes regulatory network projects are proposed That would collect multi-omics data systematically with temporal and spatial resolution Measure differences in gene and regulatory element activity in systems undergoing state changes for example profiling systems undergoing a change in cell state or cell fate Develop analytical approaches to identify regulatory networks using the collected multi-omics data and identify network level relationships between Genomic variants functional elements and phenotypes So it's important to acknowledge that It's not feasible to experimentally test the impact of all genomic variants on all phenotypes in all contexts so a Key part of the proposed program is developing computational approaches To more accurately predict the impacts of genomic variants on function and phenotypes To this end predictive modeling projects would develop and Test computational approaches including novel methods predicting the impact of genomic variants on function and phenotype the location and function of specific elements within particular biological context Also interactions of genomic variants Projects may also create tools to enable inferences about genome function As we envision these projects, they would also play an important role in the larger program and be called upon to provide expertise Helping to prioritize samples approaches So that the consortium could perform the most informative experiments most efficiently Finally an important goal of the program would be establishment of a community resource To enable future studies in this area and this would be anchored by a data coordination center Which would lead efforts to provide community access to all the data software models and resources developed by the program Make the data available in a form that's ready for advanced machine learning approaches work with standards development groups and other consortia Towards facilitating data integration Performing joint analyses and developing standards and sharing those standards and best practices and Finally organizing consortium activities including community working groups and leading outreach activities Some activities would be shared across most or all of the components Making data software and other products of the consortium available through the data coordination center would actually be a shared responsibility of all the groups Planning and implementing projects that span multiple of these components would obviously be shared activity as Would be developing standards data quality metrics and best practices and we do think developing standards and metrics Is one important thing that can come out of a larger program like this that's working together All groups would also contribute to outreach efforts This is being proposed as an interactive consortium because we think that there are important synergies that can be leveraged Both within each of these components and across different components Within a component coordination Will be important for we think generating data that maximally covers the biological space that there is to be explored While avoiding redundancy As well as ensuring that data are interoperable and have uniform metadata Or easily to compute on in the future Synergies between different groups would also be leveraged For example characterization centers could prioritize samples that would be mapped by mapping centers sites of regulatory activity identified by the mapping centers could be Tested by the functional characterization centers Data from both characterization and mapping centers could inform predictive modeling efforts And similarly predictive models could help prioritize efforts by the characterization and mapping centers And we envision similar synergies would exist across all components of the consortium additionally The data coordination center would work with data collection groups to produce interoperable data That are uniformly processed in reproducible manner and to develop reproducible pipelines for data processing and analysis that can be shared with the community So in these ways and others we think an important part of the proposed program Will be each component learning from the others to move towards their individual and their shared goals a number of significant outcomes are envisioned for the program These are focused on advancing understanding of how function is impacted by variation and leads to phenotypes and Enabling communities working in this area to move forward at a more rapid pace So movement in this direction will be aided by we think identifying and characterizing functional elements both when and where they're active and How genomic variation impacts their function and Modeling variants and elements how they contribute to different networks and pathways Towards improving variant interpretation a key part of this that's envisioned for the program will be catalogs of the impact of Genomic variants on function and phenotype that would be shared with the community as Openly as possible and taking community input into account to make them as broadly useful as possible a Data resource would also be a key outcome of the program This would be as I mentioned structured to enable machine learning approaches It would include this database of tested genomic variants reporting the functional and phenotypic effects in particular biological context It would also importantly include predicted effects of untested variants Coming out of the predictive modeling centers All raw and process made it metadata would be available made available as part of the resource And shared with the community with as few restrictions as possible tools models method standards that best practices and Technological advances are also seen as important outcomes So we're proposing that the program would contain a number of centers or research projects within each component Along with a single data coordination center It's important to note that these groups would be working together closely But also would exist in a much larger community of researchers and other projects working in the same area So now I'm going to turn it back over to Mike to talk a little bit about the relationship between This program how we envisioned that working with other programs So I hope as you've been hearing about this you've been thinking about what did we choose to put in this? Proposal and what is it that we left out and why there are some choices that we made where we didn't include topics as part of this Propose as part of this concept because we think they're happening. Well already in other places in NHG or I So for example, if you look at technology development, we think that our unsolicited Program and our novel genomic technology program Covers very well development of new experimental methods in this area. So that's not an explicit part of this proposal Similarly with respect to development of new computational methods. We think that's very important All these problems are not solved yet a lot of that is happening already in NHG or I again through our Unsolicited portfolio and through our computational genomics data science part We're also very interested in the idea of how this information can be used in a clinical context We don't have a Place for that in this particular fellow Our variation function and disease funding opportunity though does take applications in that area and I'd also point out that Our sex program is not bound by particular areas of science But we have a number of SEGs now that overlap with experimental and computational methods in this area So the field is benefiting from those as well We're also thinking that we don't want this if we move forward with this We don't want this consortium to exist in a vacuum to function on its own It's very important that it be integrated with other science. That's happening at the time So for instance, we're thinking that variant discovery projects and Sources of curated variants such as ClinGen are going to be very important We're thinking that this interaction is going to be bidirectional because people in the consortium will benefit from having variants that they can test and Characterized variants that can be training sets, but also information about those variants. We want to get back out to the community Similarly, there's a lot of single-cell work that's happening in HubMap and in HCA and We are we envision that we would work with them as partners On shared data shared processing methods shared technology Right now NHGRI as part of IHEC the international human epigenome consortium We would like that relationship to continue right now Shared resources and shared experimental methods with respect to mapping So what we're laying out here is a concept and idea to tackle one of the great challenges in genomics How is it that genomic variation changes phenotype or changes phen or changes function? We think now is the time to try and make a bigger step forward and continue walking forward on this problem So I'll stop there and we're going to move on to discussion in a vote We're very interested in council feedback on any of these points We're going to start off by first I'll call on Joe Ecker and then Steve rich to provide high-level thoughts on what they've heard Then we'll open it up for all council members for discussion and you can discuss amongst yourselves You can ask us for clarification on individual points as you see fit and at some point Rudy will call for a vote So thanks So Joe can we hear from you first please? Yes, can you hear me? Yes, we can loud and clear Great, great. Thanks. Thanks, uh, Mike and Dan and Stephanie for the very nice presentation um I think the concept clearance captures very well a lot of what we talked about what was discussed at the genome to phenotype workshop um And I congratulate you guys on doing that. I mean you got the message about function loud and clear and the map the functional Centers will be really value additional value that hasn't really been realized. I think at this scale I just had a couple of questions about I think the goal here is to have the The the whole be greater than the sum of the parts and therefore the interactions that you described I think are very very useful. There's just a couple of questions. You mentioned something about high value samples in the I don't know if it was the mapping centers or the or the the Functional centers, but do you envision that you know, I can imagine that that tissue or material that is of value to one group Could provide an input for example for for other groups To be able to utilize with their technology. For example, if you have groups that have developed a system For studying function that you know, understand the variants in that system would be would be very useful So how do you envision? What do you think about sharing of high value code samples? So so we would very much like the different groups to work together and for example Um, a mapping group of a mapping type group might learn about the biology of a particular organ or tissue and that might spur Ideas for how a function a networks group might study that and we think that then they could work together That if there were for instance mapping information that might inform doing network studies Or network studies might suggest Samples that one might follow up or functional characterization or predictive modeling I just Would add that A number of the projects that mike and i and others already work with Are you using? Oh, yeah, sorry. I'll use this one Programs were already involved in have used sample sharing across components of the encode project Hub map which is one that we work with in the common fund I think possibly the 40 nucleon program also does this so there is some precedent for sharing samples across components like that I had another question also about What if any Is the role of you know model organisms for a studying function? I know we talked about some examples at the At the workshop of where would you see that that would fit in to this function functional study? right, so I think the current thinking on this is that The goal of the program is to really advance understanding How this is working in human health and disease how variation is impacting function and phenotype and so There would be a strong preference for work to be performed in mammalian systems and human systems where possible But it would be up to investigators to make the case that if a particular area Was more amenable to study in a specific model system And there'd be high value in that that's something that could be considered And my last question So you recently and it's your eye funded reference genome mapping centers, which I think is really terrific That was the other thing that was discussed at the genotype to phenotype workshop was having real Real genomes So can you imagine that some of the systems that are in this program? If it's if it's voted successfully could be That the genomes could be developed from the mapping centers so that you actually when you map It's mapping to you know a complete genome. Do you see any interactions with that program? So you're asking the question of with the particular reference genomes coming out of the HGRP the human genome reference program Be sources for this project Yeah, I am thinking about integrating so that mapping or functional data is actually carried out with the full genome and it could come later that is the reference centers might Consider using some of the high value samples that are coming from this project for the references Yeah, we're certainly open to the idea that if it turned out a number of samples came with this uh From an individual so with the same genome Then if that could become an HGRP sample that would be a benefit And Joe, this is carolin. I one of the things we've talked about internally and I just want to touch base if this is sort of what you're talking about is that You know as we as we improve our reference genome and sort of Address the dark matter if I'm going to use probably the wrong term but getting a better idea of the genome Functionally making sure that our functional characterization is covering That entire space and so is your question sort of thinking about sort of identifying what What we're functionally characterizing and how efforts like the genome reference Can improve our understanding of what we need to be characterizing or am I not or was your question Precisely Precisely you that is yeah the the goal of that program is to really flush out what's missing and You'd like to have that within the functional characterization and the mapping as well. And so yes carolin exactly right And I'll give Jen Troyer a little shout out for being one of the people who who Push this as something for us to be considering as we think about activities as well Okay, I have a couple. Okay, that's all for me Great, uh, this is Steve so I have just a couple of comments and couple questions number one. This seems like an incredibly ambitious program And not just from the science standpoint but also logistically and how you're going to have by my counts seven to ten awards to functional characterization Three to five for mapping centers five to seven regulatory and then six to eight predictive And so in a sense, you're going to have individuals that need Say the the seven to ten functional characterization centers Have to work together in a sort of a working group model to figure out what they want to do And coordinate that with the three to five centers that are mapping Who decide what they're going to do and how to integrate with the functional group and then the regulatory network Same way with predictive modeling and that's a huge Sort of just a logistical question that It's hard to even see how a data coordinating center can can coordinate all that much less keep track of all the data and curation and so forth so I I just want to you know Indicate that there is some concern about just getting this program up and running And my guess is it's probably going to take A significant amount of time just to get these people to know each other and start hammering through the protocols So to think that it's going to get started on day one It's probably you know, you probably don't believe it anyway But it's going to take some time for them to just get up and understand what's going on um And it gets to the point of you're really looking at the modification of the genome For variation to predict function in time and space under a stable environment Because you're not manipulating the environment at all and yet we know that for You know different cell types different, you know, organ systems Whatever happens through the environment may change totally what the function is of how the genes operate So could you sort of comment on how you're going to think about having A series of cells tissues organoids intact, you know units That makes sense across time and space To allow the network people To divine the networks that are used in prediction what the function is I agree this is going to be a big challenge Um, you know, so part of by having this as a consortium the simpler part But no might but by no means the easy part is to try and manage overlap and to have Sample to have centers that are applying related techniques have only a small amount of overlap So we can compare results across centers, but you know, not everybody does the same biological system Um, we are hoping that we get impact information from the computational groups predictive modeling groups So from the beginning here What would be a good way to look across biology to what extent should we be looking across development To what extent should we be looking across body plan to what extent should we be drilling down in a few particular cell or organ types But I personally don't think there's a single unique correct answer to how best to do this So we'll be we'll be looking for input from the different groups and how to do this work together with program staff, but Yeah, that's the approach Right, can I interject can you give some sort of a contrast? Because one of the issues that steve raises is just sheer number of components that you're trying to herd Can you give some sense, um, encode at its largest size? How many? Components did it have and how does that compare it to this overarching program? Yeah, so I think uh, we're on the order of mid 20s to 30 components Right now within encode for encode. There's one data coordination center that Helps a great deal with managing all those different groups. So not Radically different than the sheer number of components here I'd realize there's some other differences and now it's organized, but just by sheer number of cats being herded Other thing I would point out and this is true of encode today as well We would envision that some of these projects would be more closely managed Some of them would be more loosely managed So current thinking is that somebody that's doing a regulatory network project We don't know so much about how best to do this today To a large extent they would be proposing their system doing their own thing Benefiting from being in the environment of the consortium So not all of these things are subject to the same amount of management and coordination is point yeah, just couple other quick things I guess is you're going to be have generation of primary data and the people who are working in Regulatory networks and predictive modeling will be working with that primary data But they have to wait until the primary data are generated qc. You know so forth So how are you going to I think it's really important to get the people who are doing the networks and the predictive modeling in at the beginning So they have some not only knowledge of what's going on, but also some idea of how the data are being produced and And actually give feedback as to well these are the types of data and so forth So I think it's really critical to make certain that these people are involved at the beginning and Really will help the project and the other part is just as I as I've mentioned before I think Having all the data and and as much modeling and prediction work done on Anvil probably would be a useful thing Especially if you can make it free To the investigators so you don't have to pay for the computation and the space and everything else but And then see how that can be integrated with other sources of genomic data And I'll stop there I've got how and then trey and then mark So, um, it seems that the comparison to n code is in some ways relevant, but in other ways not quite fair I mean n code was meant to be a catalog if you will encyclopedia of elements But here to understand how they work together Is a it's a much greater challenge and perhaps a much bigger burden For the coordinating center Not to simply collect Data as it comes out of the other initiatives, but to make sure that the other initiatives are coordinated in their consideration of system and context to be able to speak to Interaction to cross talk to I Having each of these Entities Doing its own thing will not necessarily lead up to the The sum that's greater than the parts So is it envision that the coordinating center is going to have a sort of a proactive role in Actually dictating what types of questions or problems are going to be addressed in what order and I think the current thinking on that is that that's largely a program responsibility and fall outside of the activities of the data coordination center, so So I I mean, I think this is a thing that we are also really Interested from thoughts and feedback about are there we recognize this is a challenge in this whole how do you make Programs bigger than the sum of their parts and we like to think that we have lots of examples where we've done that And then the opportunity to learn from those things, you know I think some of it falls on the coordinating center Some of it falls on a really well-designed steering committee within the group some of it comes from the sort of nih Idea, but if there's things that you think we should be thinking about as we sort of if we move forward with this concept To help with that We also and this is i'm looking at you because you asked but we're open to anybody on terms of if people think there's some Key things that you really think we should be Considering as we to that would really help address that and help mitigate that concern I mean, I think we try to do it through building in good Incentives and good ways and ways to resource the group so that they do have that ability to to Work together both the coordinating center and the individual groups But if there's more to it than that, you know, I just want to make sure we're hearing That from you as well I'd love to hear the perspective of others, but it doesn't seem that an occasional way in from Steering committee is going to be sufficient, you know, what i'm imagining is day to day Operations are going to need to be intimately coordinated protocols are going to need to be standardized Cell types tissue sources It's going to require meticulous detail to date with regard to day to day coordination just from from hearing Description and in this follow-up question And this was one of the thoughts I had when I read the description is that because of what how just described Seems like perhaps a better approach is to have several groups Each with all these or some of these components that work together and Remove the coordination because that might end up introducing inefficiencies and let each group Do its thing and at the end then share everything through an bill or whatever When each group has its own person that or group that does these things that they're they're going to be Much more often and with a with a with a goal of as a group Delivering something so just so I understand so you mean that Groups will form that would have representatives from each of these four functions and they will Act as an entity and then there would be another group doing things in parallel Is that the thought? Yeah, so you'll have I guess it would be more like a segment I guess So a a research institute will Put in a proposal for doing This on a subset of the of the ambitious goal Okay, so so as you guys know I I am a supporter of this idea of this concept and and I think it's shaping up to be A even nicer concept since the last time you guys described it to us I Support everything that's being discussed about the organization of a complex program And the role of the environment, but let me so I think I think my comments. I'm just going to reduce to Names of things but in the name I think is maybe an important clarification at least if not if not discussion and I want to sort of clarify two components of of the naming up here one relates to what you call a regulatory network And the other relates to what you call mapping So so first things first so so when I read the objective up here, it's quite clear and I actually like Exactly how you've thoughtfully worded that to explore the role of functional genomic elements in regulating networks and influencing phenotypes Presumably when you say that you don't just mean transcriptional gene regulatory network But you also mean potentially metabolic networks Signaling networks, maybe even cell cell communication networks and beyond And so my if that's the case then my suggestion is remove the word regulatory from the name And call it gene network or if you want to include cell-cell interactions at a later date or at least let people propose that then call it Molecular and cellular networks, but but generalize it That so so that would be my my recommendation on the other hand if you really want to make this just Just about that one kind of gene regular, you know, cis or trans gene regulatory network Then I think you got to change the language in your objective. But but anyway, that's that's One what's hopefully a helpful comment now. Let me get to the mapping So I think now now steve's comment about three I made me unclear about whether the network projects are mapping projects or not I thought they were when you first proposed them But at least for now the way I've heard this is that the first three components are types of mapping The first component could be called mapping the effects of genetic perturbation The second component is really mapping as joe pointed out genome elements And the third component I I thought might have an experimental component It wasn't clear relating to mapping of gene gene and other kind of molecular interactions And so again, maybe just to clarify if that's actually true I would consider using the word mapping More equitably across those three either stripping it Entirely or or specifying there's three kinds of maps Which all sound really cool actually and necessary, but just clarifying that that's that's actually what's intended Okay Mark So I'm I'm very enthused about the the program also sounds great And one thing I was happy to see was that when you were talking about the Interplay between the components you were saying that the predictive modeling may influence what gets experimentally characterized But I would just encourage you to draw one more arrow there Because I think the the network projects also a lot of that will be predictive modeling And those projects as well might influence what happens in the first two components um Well, so while we're on nomenclature, uh two quick comments and a more general comment So genome function is actually kind of an interesting term and I Really thought about it like genome function meant replication or something, you know what I mean like I didn't think of it of How the variants impact how the genes work? so I just the language should be clear about that but my biggest concern really is the mapping centers and so I actually had emailed I forgot I think the two of you before this I'm also very concerned about the number of different components and the number of different grants um, and I think the first one could frankly Sop up all the money. Um, and not not finish the job. So The functional characterization if you want to go back to the four of them might be easier I think it I think that's critical and the field is really Calling out for that Uh, and I do think trying to coordinate all these different Components, I you know the answer about encode. I don't think encode started with this many Like they may have gotten to this many somewhere along the way But I'm really concerned about the mapping centers because to me That really seems a different question It almost seems to fit more with the developmental expression that we're going to talk about in a minute Maybe I misunderstood it and I think that really trying to understand How genomic variation in impacts gene structure and gene function Um, particularly in multiple tissues is a huge project that I really want to see done well Um, and so my concern by trying to pull all these different components Into this program that there it may be more complicated The network the network of these grants may be more complicated than needed Uh, and you may lose some of the goals. Um, I also was a little surprised There wasn't a little more discussion about the types of models like Organoids or mentioned somewhere in the in the write-up But I think that's going to be critical And really thinking about what models these different groups use so that they can easily compare data Is going to be really important So to me really emphasizing the coordination of those types of Questions to really get the functional characterization centers off to a great start. It's critical And I'm concerned that all these different components may take away from the overall success Steve So I'm I'm also very supportive of the program. I think it's in a great direction. I guess one question or one Perhaps one view of this may be that you know, this may be The great vision of what you're trying to achieve long term and I think The question is are there a few things that you could evolve into this program with and a couple of the Comments that have been made here. For example, I really like the perturbation model of perturbing a genome and asking what's happened What you did emphasize sequence, but of course there are a lot of other perturbations There are perturbations as mentioned. I think you mentioned that there's environmental perturbations. There's environment There's perturbations on a single genome they can give different results and The other point Is on model organisms or maybe model systems, you know, it could be different organisms It could be different organs. It could be different cellular conglomerates of some sort communication between cells, but the the question I would have is It kind of coming back to what Raphael said is it makes sense That you know, do we have a model system that says by studying perturbations, you know genetic or environmental perturbation That we can actually have a predictive model Of how perturbations affect phenotypes or how phenotypes may backward integrate into genotype and You know So it may not be that you want to jump in to all of these things simultaneously. It may be the small groups Have some ideas of a small system that can actually demonstrate this and you can learn from it and then Kind of understand how you should populate this as time goes on And and it wasn't clear from your presentation whether you know, you would just jump into this and populate it immediately But those are just some things to think about But overall I think it's a great project Okay, I've got Jonathan and then Brent gravely. I'm warning you. I'm going to call on you next So I want to go back to the sort of the the organization part of this because it really does worry me that we're talking somewhere between 20 and 30 different You know individuals projects going on and trying to coordinate all of that so I mean one way to do that is just to You know fund all those things put all those things out there Let them come together and then try to figure out how they're going to all interact I think that's dangerous because I think it's going to take a really long time for everyone to try really try to get together The other approach which graph graphy has already talked about Is you know have themes, you know get subgroups together that might include three of these Different activities or four of these at different activities or maybe even two of these different activities But have them get together under some Some you know coordinate some coordinated thing and then maybe they coordinate in some you know broader way The the other approach would be To throw put some sort of framework on this on this before it goes out So have some better idea and then this is I think goes back to the idea that Program might be part of where the coordination comes in Maybe having a framework out there so that people aren't just totally going in without any idea of how it's coordinated Before you put it there So I don't know what that framework would look like But you know that might be another way of doing it, but I think some sort of Coordination activity needs to happen before this before this gets out there Okay, bram. Do you have thoughts? So I go Please go ahead Um, sorry. I I had to unmute. Um, no, I think I'm in favor of it. I think uh the concerns of it kind of being a Larger sprawling project have been addressed. Um, and I think as long as there's sufficient coordination and Um management, I think it should be fine, but I think this is clearly an area that needs to be pursued in the field and I think the direction has Great promise for new insights Jeff go ahead so a question I Since this is really going to focus primarily on human Systems and I assume it's going to be new recruitment into sample acquisition and that sort of thing I think the question is whether you thought about it some sort of modest lc component to help I'm guessing with the coordination center to help set standards and Issues around privacy data sharing consent return and results those sorts of things might well be Issues that somebody of that sort could contribute to And then perhaps one smaller point if indeed the coordinating center is going to be Responsible for more than data and it sounds like that's what folks are talking about is Name change call it a coordinating center and really make them responsible for a lot of the planning and the work groups and Standards that are All part of this that may go beyond the data elements Yeah, no, I think you know, I think that's a really good idea We have examples in this space and Mike can speak to this a little bit more than I can so like I heck which is one of the groups we talked about coordinating that international group has a Ethics and LC group that sort of talks about this and I agree As we move in the sort of functional genomic space more into Working with more directly with humans. That's an important Area to go and we haven't put that much thought into it yet and the concept as we've been developing it But we have a really good, you know, Dave already had some really good thoughts of things to do And I think that's something we'll take under consideration as we move forward on how to best Add that in and make sure that we are Doing this in that in a good way the other part that we didn't spell out, but somebody else brought up at one point is thinking about as we Go into this making sure that we're having diverse representation of who's being sampled and those types of Issues as well. I think there are a lot of sort of Underlying things that it would be important to make sure that we're really focusing on as we make this type of level of investment I don't know. Mike if you or dan if you'd anything more on top of that So it was suggested that some of the guidance might be programmatic Do you imagine early on in the process saying, you know, let's tackle glucose homeostasis immunologic maturation and myogenesis first Go do what you do But let's use those as test projects and things to Learn how to model from Is that a concept? For the for the network type projects, which I think get most into how systems actually work Barring several groups posing to do the same topic They would largely likely do as they propose rather than program officers tell them. This is what you would study or how you would study it Whereas if we look at component to the identify when and where Regulatory elements are active there is very important to make sure that we don't have centers just duplicating each other's effort There it's important to put together some kind of coherent plan Sampling across the body plan sampling across development So there there would be programmatic involvement with the PIs steering committee But I don't think we would get to detail scientific questions as you as you pose but to get back to Rafael's original comment This is the issue right so you could imagine take myogenesis, right? You could then have functional characterization centers looking at myo d1 and dmd and da da da and you could have Mapping centers looking at the regulation of that and the networks, but if you have all these different groups coming in With very different I mean these models are all going to be different and you're going to have to model you have to do the functional characterization differently if you're talking about development of muscle versus hereditary cancer So I do think they're having some idea of how these different groups Are going to attack the same problem It is important I don't know if this addresses the question that you're talking about but one one strategy we could do is For instance, every system that's proposed to study regulatory networks We could say well, that's something where we should also be identifying when and where elements are active That's also a system where we should be doing some perturbation and seeing what the elements that are found in the networks do So so I think it's actually more acute than that I'm not sure how you do component one without having made these decisions So I think it's not just making sure that component one covers some of the the model Pathways people choose to study, but I don't you have to choose your functional readouts You know, so you're perturbing the genome and what are you measuring? And in component one and I think that that was where I understood the the questions really be most acute and Some of this discussion and need to harmonize across groups because if you're not careful, you know, the most basic thing you're going to do is You know a genome-wide CRISPR assay for growth through competitive growth and then you're going to move beyond that to Someone's going to study homeostasis Someone's going to study cell cycle arrest someone's going to study cell migration and blah blah blah And those are all great projects, but they're not going to harmonize. So I so to me I think you've got to somehow guide or coordinate one first off and foremost I agree that that that kind of level of coordination is important in the Characterization centers. Otherwise, they'll have readouts that are not directly comfortable So I one more thing and it's it explains why I'm having trouble with With accepting that this is going to work as it's as it's currently explained and it's It helps to think about the deliverable. So the human genome project is very clear what you were doing You were getting for every location a base for an encode It's a three-dimensional matrix of zeros and ones and you and it gave us a subset of that and now they're trying to fill in the rest with Imputation but here I don't I don't know what the The literal could be it's just so massively the possibilities are as massive Jonathan Pritchard. Did you want to comment? You have to come off mute first Jonathan Sorry, thank you. Um, so I think that the concept is great and uh, you know, I think it's You know just what NHGRI should be doing and uh, and the discussion's been really good Um, so one kind of thing that I wanted to add a little bit on is I'm trying to understand exactly What scales you uh expect this to be? focusing at so You know, so I guess I would draw a distinction between understanding how genetic variation is affecting what's going on within cells versus at the scale of tissues and whole organisms and um You know and I think different parts of document left me a little bit confused about exactly, you know, whether you whether you see understanding the sort of the whole pipeline from, you know variation to Organismal phenotype is being part of this and and I bring that up because in some ways I think that that's the most challenging problem that we have that we face now so, you know, if you think about You know experiments for understanding how genetic variation is going to affect You know aspects of what's going on inside cells You know for most of these you can imagine what the experiments are But you know scaling these up to understand, you know, how variation is, you know feeding through to produce organism level effects How to do that in high throughput is still much more challenging And yeah, so I wanted to get your get your views on how you're thinking about that and You know and whether that should be made clearer going forward Thank you Yeah, thanks for that comment. So I think you're right as you go up that scale of complexity As you move up the scale of complexity from my cell to an organism It does get much much more difficult. Um, I think that we are open to exploring that whole spectrum, but that At least out of The first expectation is that a lot of that work would be done at more of a cellular level than Okay, thank I'd say the one exception would be in the network modeling projects Where we're you know, it's easier to take into account that one could use spatial resolution and say These cells are next to each other. They're quote the same cell type with the same genome Yet, they're doing something very different from each other Why are they different can we learn something possibly about what's happening from the set about what about the cellular context is important But I do see how this relates raffa to your point of sort of the This determining bounding on this if we if we say, okay, this is the theoretical matrix We're trying to fill out that helps provide a way to sort of put boundaries on some of these questions as they come up Versus saying we can go out and Anything's game if you match some of these words And so I do think that's something as we develop as we move forward with this That does need to be part of what I think is developed and the better that's developed the better We can figure out the right structure around this And so certainly if there's people have sort of thoughts on that I we're happy to sort of hear that but you know, obviously going from the concept moving forward That has to be designed at some point. Otherwise. What is this that we're trying to collect together? But the ultimate goal is to sort of get a predictive model right out of all of this then it does seem Like using a simpler system would be a big benefit now There are going to be applications in human obviously that maybe are low hanging fruit and it makes a lot of sense completely applicable but You know, somebody's got to put the fundamental effort Into trying to get a very basic simple system where we can get a predictive model of functions Well, I mean I would Disagree with that. I mean We've spent however many 50 years of human genetics learning about what genes are important for disease and I think that Certainly focusing on At least a subset of those and understanding their impact on disease phenotype or appropriate proxies Is certainly within our scope And by the way, I don't disagree with that right so I I just hope the the message doesn't get across that it should be just Very the the issue that's come up in the medical genetics community is to make sure that the assay you're using is Actually predictive of disease and that's where I think some of the predictive modeling projects Might be particularly useful Similarly clingent for example, you had clingent and kind of a funny place on your diagram We're like waiting for Data from group one. I mean we're setting up all of our systems to try to rapidly ingest Large amounts of functional data so that it can be combined with what genetic And disease specific data is available for those genes. So I mean there's definitely a Huge need in the medical genetics community for large scale Well, really Statistically robust functional characterization assets So Sharon whether it's a model system like budding yeast or whether it's a model of disease Would you nonetheless support this notion that I think is evolving? Which would be something like for for activity one Program does script a list of say disease phenotypes that are of high interest I mean, I personally somehow scope the activity. Yeah, I actually don't think it's that different than the current common sequencing centers where um a group of disorders if I remember it I didn't apply so I can't tell you but Uh a particular disease group were selected and so I do think if you really want to do all five You're hearing all of our concerns about coordination one way to help the coordination Is to pre-select some very large? disease areas and at least let people Apply saying they're going to focus on this one or that one I think it will make it a lot easier to then try to coordinate across these multiple different grants And you're suggesting pre-select prior to issuing the funding opportunity as opposed to Select after we see yeah Well, I mean it's interesting. So I guess they I'm sorry if I'm getting my acronyms wrong Is I think ignite took an interesting approach where it kind of said you could propose them but then For trials in this case and then the consortium selected, right? I just think you need something You need some way of sorting whether it's pre or post So that there's some vertical integration across the slide that the coordinating center doesn't have to do So so so to couch this in terms of rafa's matrix. I feel like the fundamental problem is The genome is is numerical. It can be numbered. There are three billion base pairs. There are 20 000 genes There are this many chromosomes and so that defines the sort of some of the dimensions of that matrix Or you know encode has a large number of elements But the problem with phenotypic space, of course, is that it is infinite and so you have to deal with that By making sure that your proposals have some overlap Jeff and then howl looks troubled. So he's next. Yeah, well Since we're brainstorming here wondering about say category one as an example instead of funding 7 to 10 independent awards why not fund three trios where the applicants have Decided that they have synergistic activities and they describe in their applications how they're going to work together How they're going to coordinate their activities still need a coordinating center, but you minimize the Large number of potentially wide-ranging projects that then you'd have to synthesize So I like the concept that's emerging from rafa and from Sharon and others I'm not sure that I would focus on a disease entity I think focusing on some developmental or physiologic process is going to inform the diseases that derive Yeah, no, absolutely Right, and that's why I thought your example of myogenesis was a good one because there are some diseases But there's all kinds of interest in physiology. It relates to health. It relates to activity But at least people know where to aim Their assays and their models so I agree. It shouldn't be specifically a disease, but a process an ordinance Yeah, I mean and I like an early concept that was introduced where a center from one through four Is they're going to say hey, let's form a little group and attack myogenesis or let's form a little group and attack whatever But then it would be necessary to have that kind of synergy already developing at the time of applications So that's going to be a tough but probably worthy exercise Okay, I just counted to 10 and it was silent. So unless there are any last thoughts I'm going to bring us to closure here. Anybody on the phone want to Speak some more. Joe. Do you have additional thoughts? Yeah, I just just a comment I You know, I I brought up the idea of using sort of common tissues and things like that with the idea that you'd want to have some systems for example, if you were studying for example, brain And you were then had a group developing functional assays in Organoids that were brain-like then they might be able to coordinate. So but I wouldn't prescribe I wouldn't necessarily prescribe a disease nor a model system But I think that the centers could decide after funding that they wanted to work on x number of Model systems to be able to inform not only, you know function, but then the regulatory network project. So I think with the right kind of Guidance that this could be possible within the program that's been outlined Okay, Brent or Jonathan final thoughts Nothing to add Okay, can I get a motion to approve the concept? Go ahead Sorry, are we going to be voting on the agreement to I heard a lot of support for the overall concept But people's concerns about the complexity and some recommendations about how to focus I didn't hear any consensus about what we're focusing on But I did hear that people felt a lot more comfortable knowing you would go back And find some way to either pre-specify or after you get your applications Focus is that what we're voting on or the 40 million dollars for this so originally So I'm choosing to view this as a lot of advice Some of which we will embrace some of which we will not As opposed to we have rewritten you the council have rewritten this this concept to mean the following things We you're rough as right you're voting on that You're voting on that but we may choose to Modify numbers structure. I mean they're based on the feedback we've got Now can I get a motion to approve the concept? Thank you a second All in favor Keep your hands up, please Okay, and opposed and abstaining And the people on the phone if you would just send me an email Please letting me know whether you're voting to approve Or disapprove or abstain Thank you very much. Thanks guys. Thank you Okay, Jyoti do you want to come up to the podium, please?