 So, we're at that happy time now at five. I have eight slides to go through and it won't surprise you that there's one for each panel. What Dan and I wanted to do was to try to capture some of the things and there's been a lot discussed here, obviously, but at least some that we could discuss a little bit and be sure people agree with at least the way they're formulated here. This is not going to be the entire output of the workshop, of course, as I understand the recordings that are has been capturing the chats, so those of you who are saying snarky things will all know that. And in addition, I guess we should have told you when we started, but we did have this planned as a recorded session with webcasts and all that when we were in person and since we've been sort of partly in person and partly in vitro, our web team has managed to capture a fair amount of the discussion along with the slides as they were presented. So, particularly those who are taking notes want to go back and read or listen to some of the discussions. They'll be available. I think we started that at about 11 in the morning, so we missed a couple of them, but we did it as quickly as we could, given that we had to adapt. So with that, maybe I'll just start with the first panel and Dan, my partner in crime, can kind of comment when I get something wrong, but we did our best to try to capture these things. So we did hear a fair amount at the beginning about the tension between research and implementation and do we want to do perfect research, which we might call a randomized clinical trial versus really perfect implementation is something that's truly ready and has the evidence-based behind it. And I think later on the point was made, forgive me I forgot, maybe it was married, that we could have research or clinical implementation or research on clinical implementation and it's the lab, the research on clinical inflammation that's probably the sweet spot for a merge. Who was it who said that, by the way? Maybe you're not still on. This is Julie. I think I brought that up. Julie, yes. Okay. No, I wasn't hearing. But anyway, so that's probably a sweet spot for a merge and something we should consider, pragmatic clinical trials is another term for that, but there are other terms for it as well. We recognize that learning health care systems are already doing this kind of work. They do the research as well as the implementation and they make their care systems better. So I think everyone recognized that there's perhaps a dearth of expertise in quality improvement, potentially clinical workflow, possibly dissemination and implementation science, although we do have Mark Williams and if he doesn't count for five of just about anything, I'm not sure who does. So at any rate, that seems to be a concern or a suggestion as well. We did recognize the unique opportunities that merge has for recalling participants who have unusual genotypes and so having the genotype we might be able to recall them given that we have the large numbers. We're also uniquely positioned to address penetrance, heritability and pathogenicity, very interesting to the group. And it was recognized that we didn't need huge numbers to assess these. You could do it really in three or five people was suggested, maybe you need a few more than that, but you do need to find a few that have rare variants. So is there anything here that anybody would violently object to in terms of panel one's output? This is Mary. I guess I violently object to the last one. We don't need huge numbers to assess things like penetrance, heritability and pathogenicity. That's what you would object to. That was actually I think almost a verbatim quote of Gale. Yeah, I think you need a huge sample size, but for each variant you don't need huge numbers. Is that, can you agree with that? I think it depends. Are you trying to assess the clinical significance of the variants based on the phenotypes that you're observing in the EMR? Well, for example, a lot of the way we annotate now is we just look at the frequency in the EVS6500. And if it's too frequent, it's just not pathogenic. And if you had 100,000 people, you could really fine tune that. You see what I'm saying? Yeah, I do see what you're saying. But don't you think that that's how we've gotten a lot of false positives in OMIM? That's how we've gotten the false positives out of OMIM is by finding out what the true allele frequencies are for things that are relatively rare, but not ridiculously rare. And not rare, I guess. The example that was given, the breast cancer example, I did not think was a good one. Because it's very common for three people to be able to have breast cancer before the age of 80. Yeah, I appreciate that. But if a bunch of people have the mutation, first of all, if it's just too common, then it's probably not pathogenic. It's too common for the disease. And by individually reporting the phenotype, this is going to have to be community source. We're not going to solve all of this in a merge. But by reporting our phenotypes associated with variants, we add to the community of information for variant annotation. Yes, I agree the presence of the phenotype, but the absence of a common phenotype could be misunderstood then. The absence of a common phenotype has to be taken in context of penetrance, which is often poorly understood for some of these conditions at a population level. But it's still a huge amount of information. I mean, if you look at how valuable EVS-5500 has been for clinicians, I mean, I use it every week for clinic. I don't think that's right. OK, so let me stop. And we'll let Mary and Gail continue this discussion later. But I did modify that bullet to suggest that this is somewhat context-dependent. So maybe you don't always need to use that number. But you do need to find a few with rare variants. Perfect political. OK. I've been in the government a long time. But if you find two cases of something associated with a rare genotype, you know nothing. The only thing you know is you can put that in a big data bank and hope that other people find them. So I'm on Mary's. I agree with Mary that we have to be careful about the words that we use. That's an honest thing. OK. All right, well, we will continue to refine this. Yeah, no, I think it's good to have you. That was pretty good. That's right, Jerry. Jerry, this is read. I mean, to the extent that any families within eMERGE can be capitalized on, then you have an opportunity to look at rare variants and do, in essence, segregation, assuming that they are reasonable penetrants. And that hasn't really been talked about. It obviously depends a lot on how good your family history data are in eMERGE. Excellent point. Yeah, and I think we didn't even touch about, touch on families here. We did a little. Well, we barely touched on families. We did. Exactly. We did. We did. We did. We did it. We did it. We did it. We did it. We did it. We did it. We did it. We did it. We did it. We did it. We did it. Right. They were just making it start to work. Right. OK. We are going to need a kind of commit. OK. Can I make a comment here? Of course you may. This is a kind of a comment here quickly. Please, now. Go ahead. Please, now. Well, I think the one thing I was missing in this whole discussion is no one talked about the study design for addressing penetrance. You know, are we talking about a prospective analysis or are we talking about a case control analysis? So I'm not saying we should get into it at this point, but I think to address that question, you're going to have to talk about exactly the studies designed you want to employ to estimate penetrance. Okay, excellent point. We will note that down. Yeah, as I said, this is not the final word, but we're trying to get, just as you've done, Neil and Mary, the key issues we need to address. Okay, moving on to panel two, EMR and clinical phenotyping. So I'm now hearing an echo in here, and it's gone. Thank you very much. All right, so EMR and clinical phenotyping, there was sort of strong consensus it seemed, which is hard to assess over the phone, but it seemed as though there was a feeling that emergent investigators should convene some kind of a forum with the clinical leadership of academic health systems to try to find some common ground for genetic, genetic medicine-related research. One of the things that they could then do is perhaps speak with the unified voice to electronic medical record vendors. So it seemed like that was the only thing, that was just one issue. So that seemed to be a lot of discovery. One of the things that's supporting genetics is often seen as a money loser, and we need to work to demonstrate the value proposition. We would ensure, we also need to ensure when we deal with the institutional leaders that there's a bi-directional information flow, so that we understand what's important to them, as well as what's important to us, so that we can end things together. And we should explain how to use the EMR for implementation, identify what they're ready to implement. So that's just sort of the same thing. Forgive me for using EMR, it's just when I type EHR, it always corrects it, and I don't figure out how to fix that. So I really mean EHR. Any objections to that? So we had other research areas that Josh talked about when he went through many. Yeah, more than I could probably use. And these don't touch anything on the phenotype. I mean they're all definitely directed at EMR, but there aren't any. Yeah, the modular thing. There are actually several, and I could even email you. That would be wonderful. Sorry. So with Peggy's amendment that there's lots on phenotyping of research that needs to be done, are there other things, anything in here that people would object to? I don't object to anything on this list for sure. I think there was strong agreement on, this is Josh by the way. One thing I would add, I felt like there's fairly strong consensus around the idea that we wanted to try to use the EMR in ways that was unique in terms of the phenotypes we addressed, longitude, no pharmacogenetics, something like that, that may leverage the 350,000 people we had. Okay, so how about, I have an action item to maybe work with Peggy and Josh and share the things that they're being able to do to help with some of the better work. And we can actually, when Josh gets here, we'll meet and help them. Okay, thank you. Did you hear that Josh? I did. Okay, that's great. So for panel three, EMR was able to discover there was a strong feeling that discovery research should remain a high priority for an user on at least unusual and heavily represented and merged. This is the last two years that's been talked about at the center. We suggest the importance of rare but collectively common variants, which obviously are for sequencing. Alternative designs for discovery should be considered. Extreme discordance, discordance, I guess, and pale distributions, I didn't see as really all that alternative. I mean that's been done in genomes for some time, but that was the comment that was made. And we shall also extend a collection of processes to additional sources of DNA and possibly RNA, as well as data on environment. Projections, any of these? The thing came out with Hogan's idea of the chip. I got him already with space because that pops across both the epigens, you know, so you may be better put here. So I don't know the very interesting suggestions. So the suggestion of a chip for to emerge. We've seen a lot of fellowship. I think I've made lots of functions very, very important. The better experiment would be the sequence thing. The better experiment would be the sequence thing because those chips are not going to capture diverse populations. And Debbie, Debbie would be quite a sequence thing. Yeah. The question there is sort of the timing right now versus in about a year or two, you will have sequencing price come down hugely with these sort of new technologies. So there's a window now for about a year and a half to do something where I think a chip could be very, very valuable. And you may have the unwanted chip in any case. There's nothing wrong with genotyping somebody's sequence from bite first. I mean anything we're talking about now is here and a half away and anywhere. I mean we really should be focusing on where the puck is going to be and not where it is right now. The sequence thing is... Anything, any other objection to what's on here? And then omissions, major things that will up off. Other than that chip. So the only other thing I had was nearly Richard observation of the Phoenix toolkit. Yeah. And putting in GIS information. Carrie, this is Marilyn. Can you hear me? Yes. Okay, great. The only other thing I would say is if once NHGRI starts looking at budgeting and what can and can't be done in an RFA, we did talk a little bit about we think even without generating new data, there is more use of the data we have. Of course, we want to generate more data. But if in the end, we can't do additional sequencing or more data collection, I think there is more to be mined out of what we have. So I think that shouldn't be lost. And I think John Harley made that point as well. Yes, I mean the phenotyping right now is the time obstacle. We have the goal. The two last data is continuing to be updated in a very clean way. So if we are going to develop a phenotype to look at the sequence with, see what I am going to want to run through GUI because the data is there. Yeah. Okay. So panel 4 on genetic testing, we need to understand that the trade-offs potentially between the two-stage CLIA where one qualifies or confirms in a CLIA setting in a subset that doesn't do a CLIA compliant process in everybody versus sort of CLIA from the start of a more universal process. This was in discussion of the PTX implementation at the various sites. Some sites, one way, some sites went the other. And I think we were urged by the group to try to understand what are the benefits and advantages and disadvantages of those two approaches. And we are probably in a good position to study that because we have the same platform, many of the same outcomes and so something worth advising. We may also need to ensure that consent and a standard refer to the consent variance. So various ways that consent might be implemented that it allows deposition into the EMR. The process of reinterpretation of variance and understanding our re-annocation of variance is a researchable question. How does one go about that? Who has responsibility for it? How is it best done? What's the frequency that's sort of high-point? And the approach to re-contact participants is also a researchable question. Many researchable questions came up. And Heidi told us about the ACMG standards for clinical live reporting that were something that we certainly should be considering or similar standards at least as they develop should be followed obviously when we do research. Anything on there that people have trouble with? It's in the last point separating the lab methods which is what Heidi was talking about. First is the genomic testing recommendations of forced return and pediatric return. You're talking about the methods that I just want to separate out. Right. The return piece does not mean something that is not well-ordered. Yeah. I think that she actually referred to reporting like what they report out about how they report it in the format. That's right. But not the actionable theory. Yeah, that's a separate report. I just want them to be kind of alluded to. Good. Other rejection? Are you all still there? Mark, you're not saying anything. I'll chime in when I've got something to say. There you go. Please don't ask Mark to say anything. He's been on me for a ride Friday. Keep that in mind. Ask Mark to say as much as he wants. There you go. Panel five on consentification and governance. There were a lot of recommendations from this one including we really need to accept the impact point of clarification. And again, you know, our recurring theme was how it's done when is it useful. He's alive there. Mornings and you know strongly urges us to agree on family history into the EMR. We did an analysis within the merge of the, even the presence of family history information in the electronic medical record. And it was a visible it was like, you know, 20% or so of maybe three of that. In M-U-2. Pardon me? In M-U-2. Yeah. And actually, thanks to work that great hero when he was here. But we should facilitate interaction of LTE research components across multiple networks. That is something that we struggle with. And I think we did quite well in the first phase of the maybe not as well this time because we were focusing more on implementation. But also because there was this whole return of results consortium that kind of sucked the oxygen out of the group when it came to LTE research. And then they went out laughing. Hopefully we can find a way to deal with that in phase three. Yeah, they all signed off actually. As we were saying that there was little work in the legal arena. There are clear and hit-let issues among many, many others that are researchable and we could address. There are also economic issues that we probably aren't paying as much attention to as we could. We don't have the expertise. Engaging payers and stakeholders was another thing that was suggested. It probably hasn't been done as well in this group. Policy development certainly is needed for effective implementation and something that we should try to pay some attention to now that we haven't. And persistence as you know with data has unique policy applications. So the fact that I think this was beginning to work on it, that these data do persist throughout the lifetime. And that has really, you know, there are very few other things that do that. Any objections to what's up here? Any additions to what's up here? I think the only thing I would note related to payers is that if you ask payers in general, they say well we're not in the research business. So I think as we think about engagement, we're going to have to be very thoughtful about what we're really engaging them to do. Terry, it's Susan. I have one addition on legal. It's really easy to lapse into, you know, thinking of legal as the statutes. But legal here is also going to include liability which we talked about a little bit. And also the sort of human subjects regulatory context. Those are really important ads. So when you say humans are just regulatory, that sounds a lot like statutes to me. No, I mean like the common rule and the advanced notice for changing the common rule. And those aren't really, I realize they're not technically statutes, but aren't they? They're regs. I mean the human subjects, right? How about regulations? Liability and regulatory about that? Sure. Okay. Unfortunately I go off the end of the slide, but you all are one. The other additions? The addition. On the other hand, IRB collaboration.code site. Right. IRB collaboration.code site. IRB collaboration.code site. Something that we really haven't done. Yeah. Wouldn't IRB education? Education. Yeah, education and collaboration. Not in traditional IRB experiments. Yeah. There's two decades of central IRBs. And the gamers have sufficiently different issues. It's going to, it needs to draw that entire path into how you get IRB agreements and central IRBs. Although the point was also made that there are a number of us that are also involved in the HMO research network that is currently actively engaging in this discussion. So we could potentially leverage onto that. Well, and the CDRNs that just got, are in the process of standing up have got a very rapid mandate to solve the IRB. Yeah, the clinical data research network, PCORI. Okay. But maybe not. Yeah, since there are many working in that state, maybe that might not be a unique place for emergency, but there may be other places where there are, such as the clinical implementation, how do you get IRBs to understand that? Yeah, the unique aspect may be related to the sensitivity around genetic and genomic data, but I don't think we'll necessarily be covered by the other groups if somebody doesn't put it on their agenda. We don't think IRBs are necessary for research. They're not necessary for clinical medicine. So we're looking at genomic medicine and educating IRBs maybe, you know, not enough to put our paths. And they're getting the hospital. No, I totally understand that. I mean, and I think it may be a space, and I think the CTSA is not a genomic research. And it's also the case that if everybody can imagine a set of institutions that don't, their IRBs are not afflicted by genetic exceptionalism, it would be eMERGE institutions. So maybe they're not the model for generalizable solutions to the kind of arcane knowledge and the fact that IRBs go into reactive overprotectionism if they don't understand something. They don't understand because it's too complicated. Okay, we'll try to capture that. Maybe I can get it from there. All right, moving on. On return of results, the point was made, and I think it's a good one, that eMERGE really has an integrated infrastructure for empirical LT research, probably better than any other network. We really have integrated it in a sort of parcel of eMERGE. And so, return of results is a key topic, but there are many others, and we should probably try and take advantage of it. It's almost a loss in the sense of that panel, but at any rate, I'm going to consider it. Highly penetrative areas seem very highly suited to return of results investigations, and these institutions are interested in them because they recognize their liability and other issues. And there's little work ongoing in those. So how can we address the highly penetrative variants that are going to be rare but still will have a substantial number of people with them in eMERGE? The point was made and well taken that we should ensure return of results studies with diverse populations. There's almost no work in that area today. And eMERGE is well positioned to study the impact of patients having access to their own electronic health record, and we should use that to facilitate research and answer research questions. Any objections to anything here? I'm just going to discuss the issue of sequencing. I mean, give it on about the whole of the sequencing of the eMERGE panel, but we also got a product broader and the implications that might have in its link to the eMERGE. For return of results? The issue is going to become much more complicated. The issue is going to become much more complicated. But to get a highly penetrative variant... No, and I think the separation of return of data from return of interpretations is worse on note because it's really too different. We also talked about finding action in those T by developing some type of scoring risk. I mean, that was a kind of discussion. More broadly than that, we talked about comparing different approaches to actionability, some of which would query patients themselves and participants about what they regarded as actionable and desirable information. But right, also Heidi talked about scoring systems that were more expert clinician driven. So I think comparing different approaches to actionability is going to really be crucial in this area. And particularly in the context of discussing potential partnerships with PCORI, then we absolutely have to have the patient voice represented in this. I would also just modify the last bullet, Terry, to say patients having access to genomic information through their own eMR. I think there is... Can I add another thought, too? We did talk about looking at the issues raised by returning a proband's results to kin, to family. So I think that that's really uncharted work. Our R01 is on that, but that's coming up in a lot of studies. And we're critical. Trust me and add it. I won't check this here. Yeah, but it's 528. We'll send these around. Anything else on eMR integration? There was a lot of talk about the open info button approach and the standards that nicely permit implementation in just about any eMR that uses those same standards. So that's sort of a readily transportable thing that people thought looks good and we should continue. I think that's the first thing that helped your system move in a constant. I think that's a direct quote mark. And so when we address the issue of should we or shouldn't we provide results to them, they probably are the ones that have the most stake in it and are likely the only ones that will keep it throughout and move it from systems that much. It was recognized... It was recognized that genomic clinical decision report requires both a variant and a clinical action and that distinguishes it from non-genomic clinical decision support. I think that was Etta said that. The clinical actions Etta probably do get updated in non-genomic CVS, but the test result was not. And I think that's what's really unusual about very split. The clinical actions are changing much more rapidly in other genes. And there's an important point related to that that Etta brought up, which is because the result doesn't change. That's why the versioning in the EHR is an important thing to study related to the legal liability issue. Because if you give a decision at a given point in time, what was the information you were using to make that decision? So that would tie in with the legal issues that return of results that were discussed earlier. So you didn't know that a variant was pathogenic at the time? You can't be faulted obviously before it was not reporting that. What do you mean, Mark? Yes, but without a versioning to say what was the decision made on at the time, then that can get lost in the shuffle of the tort. Implicit in that, but a lot of work is trying to standardize the components of that interpretation. That this gene clinical scenario means this, and that that's a coded value that you can store for the patient as opposed to running a version of a decision report, you know, a rule every time. But what you really want to do is be able to capture, you know, this person is a high metabolizer, this person is a low metabolizer, capture that interpretation in a codable form. And that's, you know, what Chandling Chris has talked about a lot is that when we're storing these, we have, we bounce back and forth between storing the genotypic detail and storing what the genotypic detail means. Yeah, but so codifying interpretation is this double-edged sword, because if the interpretation changes, then you've got to do the equivalent of a global search and replace for all your additional coded data. Now you added, which did not represent the primary data, but rather a point in time for some of it. So, you know, I think you see it playing into it. It's not to debate the difficulty about it. How to record these sequential interpretations in a way that you can walk them back. I think the other thing sort of just as a general issue that was discussed more in the other ones is this issue of scaling beyond the emerald sites. Codifying our experience in a way that it can be translated more broadly. And just as a general... I'll be down in about five minutes. You know, we've worked so hard to get these systems up and running, but what we have not had a chance to do then is to do the more traditional implementation science of what happens once you turn them on. So, there's the studying the impact of what we've built in this version and what are the lessons of that. Another thing is the value assessing the value of these genomic interventions. So, the cost of actually studying them. We also talked about bringing clinical practice leaders, those who are familiar with that clinical workflow and the point that there are specific clinical workflows in specific practice settings or even for a specific position. So, that does complicate things. And the point was made that when we try to convince people to do this week, we should make this point as well as cost. Because as you mentioned, the 2D system changes will help you take probably many examples and some more data from others. I think we also brought in in that panel and earlier the need for the clinical practice leaders to understand the importance of capturing both the clinical and the genomic data. So, I think clinical practice leaders have this general agreement with the facts for decision curve that when we get into the thousands of genes and tens of thousands of proteins, no practitioner could possibly do this by what they read and remember. But I think it's also the case that they don't see it yet. It hasn't happened. It's a very tiny number of actionable things. And so at some point they'll be the tipping point where obviously you do this because genomic medicine is in the house. It's the driver for the need for testing approaches. So, Dan, I don't think we have to wait until everybody agrees with us. Instead I think we should invite influential leaders to workshop where they can tell us why they don't agree with us or what it would take for them to agree with us. I think it would be that kind of tone that would actually be helpful to get them into such a meeting. That's a possible hypothesis. Pulling back something that Dan said very early on in the day, a part of EMR integration that we have not tackled yet is just in time phenotyping. The idea that the decision support drives data collection at the point of care in order to optimize a genotype-genotype batch. I think that's a very useful construct. And I would almost flip around that we bring, we have special meetings with the leaders who are designing the EMRs, that we integrate genomic science and clinical science and genetics into the already, piggyback it on the already transforming EMRs that are being used to improve health care efficiency. I mean, there's a radical revolution going on now. And this is one, and we ought to have a seat at that table to discuss this aspect of clinical decision support as they're building all variations of other types of clinical decision support with the purpose of efficiency. So I think there are some who have seats at that table, Mike, and I guess, you know, Ken Kawamoto has been involved in that. There are issues as well, but I agree that that is an important issue. Is there anything else that we want to add to this list here as sort of critical emissions? Obviously there will be other things to come through. I mean, Kevin, you want to make sure that when you put something into the medical records that it is an annotation that is associated with it that sort of is automatically brought up to the attention of a clinician. This is sort of the way we manage the epic system here. If you put in something that influences warfarin, for example, you can't write a warfarin dose if that information is on the individual. So the annotation there is critical. You know, sort of that's adjusting timing too, only if they're a beginning warfarin. So one thing that eMERGE may need to champion is the fight against the tendency to both do lossy compression that is do a lot of observations, report only a few, and throw the rest away, which clearly already exists. And the medical legal incentives to delete data as soon as a statute allows you to get rid of it to reduce malpractice liability. It's not a major theme, but it clearly in genomics has a much longer tail of implications for health and well-being of individual patients than, say, an X-ray results or standard clinical labs. Great, we have one more. Hogan's comment made me think of something that we didn't represent here, and that is different types of clinical decision support for effectiveness. Again, he brought up the hard stop in the clinical workflow, and we've talked earlier about the fact that that is not always necessarily a desirable thing. So I'd like to see some representation of exploration of different ways to do decision support to implement genomic medicine and compare their effectiveness. I'm not sure where it goes, but figuring out how to leverage the other elephants that are running around, including the CCD from Meaningful Use and the CDRN from PCORI, that there will be massive piles of potential phenotyping data in new, at least semi-standardized formats that will be coming online during the period of E-merge 3. We didn't have quite a little menu here for one little study, guys. Alright, next panel. This is the last one, so this is genomic medicine and pediatrics. There was a lot of discussion around the incongruity in pediatrics persons in both kinetypes. While that would seem as a disadvantage, it would also seem as a strength. The question was asked is emerged, a pediatric component large enough. Is any study ever large enough? But large enough for what? I think it's the next question, and perhaps we need to define what it is that we feel that the pediatric component can and should do and where it might need to be larger or smaller or whatever. Can we comment it on the size? You have three pediatric sites, but just a lot of them are not possible to have a public cure. Pediatrics, and I think advanced development is starting to increase, so maybe that within these sites it's just relying on trust. I mean, there might be three great hospitals, but it's my size of that. I'm just saying that the issue of size, and I think there's also advantages that can play in the other sites bringing more pediatric patients, because then it's just more of a longitudinal component that the three pediatric hospitals in here issue about don't have. There was always the hope that the three bestial response in the world would have. Then the adult sites would say, oh, this is great, we'll get on the bandwagon. Some of them are starting to. I think that's kind of the striking the size component. And John made the point, can we encourage the adult sites to come more toward the pediatrics in some of the penis sites that may not be as much interest to them, but probably do have applications in adults as well as children, developmental disabilities. We're trying to do that. They are. I mean, we're testing a pediatric penis type right now in our adult population, atrophic dermatitis. It should just become a community, especially developmental penis type. You know, those are things that go away with time. Those of us that do it often, those of us that are exploring adult because I added even more. You didn't speak up when John said that. I know. The point was made that children with adult diseases early in life are likely to have a high genetic load and are very interesting to study. I know that what is not missed on pediatricians is making this on some of the adults. This is Steve Lerner. I'd like to propose that maybe children have pediatric diseases and adults have pediatric diseases that simply occur later in life because they have a lower genetic load. It takes more environmental influence for the disease to manifest. In fact, obesity is a classic example of that. Early childhood obesity versus later on. We'll make it more articulacy, but obviously there's the converse of that. Maybe it's more environmental load. We're recommended to be an untapped resource throughout a merge. We did do a lot of soul searching early in a merge about can we measure CNVs? We determined that we didn't have algorithms that were really reproducible enough. Cameron Elizabeth, are you still there? I think David Cousins is looking at it and we can certainly look at it again. It's still going to be difficult to get it out of these data, but it may be worth looking at again. I know David's trying. Sequencing gets it much better, doesn't it? Yes, sequencing absolutely. We look at your finer regions. We're looking at your copy number variance and you would look at it with sequence. Nonetheless, in the merge one, we were able to connect copy number variation to bone marrow dysregias and even show a longitudinal that they acquired copy number variance before they had the bone marrow dysregia using the longitudinal data. Those were larger things than you find by sequence, but I think it's still useful. David is very aggressive. It's fair to say that it has not been rolled out network-ified yet. There are many other questions one could ask about CNVs. Pediatrics has been underrepresented in genomics and we should consider more Gene by Environment data collection in children. I think the point that we have a real strength in having pediatrics and we should continue to capitalize on that. We should find ways to relay the merge and the new board sequencing project. I was delighted that we had Dr. Stowell and Upholm on the call who were two of the PIs in those programs. And we have one. So that's three of the four now are intimately familiar with merge. We should try to target conditions for genomic analysis that have early clinical utility in children. So children in particular seemed to be felt that the loading through the penetration of clinical utility hasn't quite happened yet. I'm not convinced that it could be something else. And the point was made that it emerges uniquely positioned to lead a translation into children, which I think we may be among certainly genomic translation. We're, I think, the only group that might be doing that. And then the question was shouldn't we consider birth defects? I think that's not something that has been compressed by the 50 after birth groups. It's in the peer. There you go. Yes. Good point. Good point. Although some of them may be sure to be correct. In fact, you can find that patients with heart disease or the parents once they've been corrected, they may not have to know how to do it. There's something in that one that's correct if you don't have it. No, I don't have it. Any major omissions on this? All right. So I think we're done. And at this point I'd like to just sort of give you an idea of what we will try to do soon, which is to try to come up with a summary that's not a transcript of this meeting, but at least the major points, the major things that we identified as opportunities plus challenges for eMERGE, including some of the things that we didn't, you know, that we can't do in eMERGE and shouldn't try to. One of the things that we didn't do was set priorities to leave. And it would be helpful to try to do that maybe in eMERGE to make an address that time in the next couple of days. And then we will, you know, also send out these slides. Anything else we should be doing? I mean, thank everybody. I apologize for Washington's inocidable weather. It really handled it quite well. And I was delighted that this went as well as it did. All right, great. No, I appreciate everybody's flexibility. Thank you. Thank you so much. Bye-bye. Bye-bye.