 We're on the home stretch So if you look at the program my task was a discussion and setting priorities based on the last section and I really because we've had some very good discussion. I was brave enough to attempt some prioritization and So I in bold So what I've done is basically made a list of criteria and it really follows very nicely from Pat's sort of last maybe four slides and What I've done is list some criteria for sample selection or cohort selection But in addition I have set some priorities and I've put two of them in bold. It seems like Two things that we've heard over and over and over again is sample size and broad phenotyping and From the sample size Point I don't think that there is a single Cohort or beautiful cake a single cake that will satisfy all of our needs So we're gonna have to have some kind of an amalgamation of data from multiple cohort studies and I agree with the comments of the last couple hours that there are some challenges of Harmonizing data both from the sequencing point of view and the phenotyping point of view But there's there are challenges that are not insurmountable There are challenges that we need to know up front We have quite a bit of experience and we need to carry that forward One editorial by the way one thing I keep hearing we've we many people at this table spent a lot of time harmonizing phenotypes make sure we save those data and build off of them So because we can continue to do that, okay The other is The the idea of the phenotyping themselves I and I sense a little bit of disagreement among the group about the importance of I selected the word exquisite phenotyping here. There's no doubt. We need quality phenotyping. I Think it was I forgot who it was actually about you know, don't let me was Maynard I think let's not have perfect be the enemy of the good that we have some very good phenotyping and Probably what we have is actually better than good enough and we don't have to set the bar so high That we don't include certain cohorts. So we just to repeat we have the importance of sample size We can bring together than multiple existing cohorts that are that with has high quality phenotyping Then the second is this idea of very broad phenotyping at least for the initial discovery We we can benefit and learn from the GWAS experience that we can do a lot of harmonization So having broad phenotyping it would be very beneficial across the institutes and probably we can pick You know Eric Lander likes to talk about the 100 most important diseases and my guess is we can make a lot of headway by Identifying a set of core phenotypes of great public health Importance where there is not good Treatment and we can make a lot of progress So that I put those two in bold by the way and you can argue with me But I said I thought those were the most important or highest priority number three We have a we seem to be satisfied and in fact we're embracing really is a mixture of direct measurements And that means you're actually taking a person and putting a tape measure around their belly or putting a needle in their arm We're having them spit in a tube Right in front of you We're leveraging with follow-up questionnaires and also then linking this data with electronic Health records of various kinds. It does not always have to be an EMR. There's a lot of other health information out there Both in this country and in Europe, and I would guess elsewhere Particularly since many of us are looking at phenotypes of in older individuals There's there's Medicare Medicaid and national death indices that we could leverage and I don't know that much about the VA But we probably should consider linkages with the VA. I don't know not Okay She's gonna tell VA stories Then next is We seem to benefit a lot from two things they're not really related But I have them on the same line because it's really time is ongoing Contact and and that ongoing contact allows us for continuously measuring and updating information And then it also allows us later when we want to find them and bring them back So really if you're a cohort person in the room and we need to get this message out is keeping ongoing contact seems to be very important for Allowing cohorts to to remain active and hopefully enter into this program Unlike a lot of meetings, and I think that it shows growing comfort within the field I have number five obviously there needs to be appropriate informed consent But I am impressed that that my opinion is the field has matured for in many different ways and We probably have reasonable informed consent in most of these cohorts right now. We've tried to update it We've learned a lot. I may not be perfect, but my guess is it's reasonable But extremely important. I put in I Sort of highlighted is that ability to recontact so so it's it's important both We have ongoing contact in that ability to recontact Because we're not only satisfied with broad phenotyping, but we on specific individuals We may not want more in-depth phenotyping and recontact is key It can be in-depth phenotyping via questionnaire or using my example of bringing that homozygous Individual in for a loss of function variant putting them into a CTS a like setting. I think there's a lot of advantages. I Also think it's important that we consider diversity across the board not just Ethnic diversity and ethnic history this this country has a lot of other kinds of diversity And if indeed for for these complex traits or chronic traits that are common that it is a function of gene environment Interaction we're going to need to see a lot of diversity and be inclusive and Probably a lot of people view that last statement is as one of maybe its political correctness on one hand And also it makes the study a challenge. I would I would put forward There's actually a lot of advantages by by bringing in more diversity into the sample set We there's more information there that we can leverage there may be discoveries You cannot make in in group one. It is quite possible in group two because the obvious one is the variant is Present in group two, but anyway, I would encourage us to think of this diversity as a positive Moving forward So I'll pause there. I tried to get all of these priorities on one slide and Let me let me first ask are there any high priority Points that you think I'm missing so we don't Keep sorry try to keep the discussion on on point Daniel I'm perhaps this is captured within the population diversity aspect, but I think There are there's some very strong benefits to including isolated and or consanguinous populations Particularly for the specific goal of capturing homozygous rare loss of function variants for instance Don't regret Just to add to that. I think family is important, but also I think we spent some time talking about You know a very unusual or interesting diseases or modifiers and in the like that the question is are many of the Coorts going to have in essence filtered them out because of the criteria for joining a cohort Maybe not having a number of pre-existing conditions. And so I at least wanted to raise that question because it's been going Exclusively towards cohorts, then you you do raise that that problem of Excluding other groups and so you know and thinking about not the pie that we're going to make because pios have to mix everything up You know you cut up something that's made to look like a pie the pipe that we would want to put this through You know there's sort of Bernoulli's principle of how much fluid is going to come from which particular Viaduct I think I'd want to put on the table that there'd be some element from the EMRs of particularly interesting patient populations that may be very Very novel and give us information about modifiers in different ways of looking at it some things is a is some small percentage of that That's in disadvantage of making the slides before And I agree with you as I mentioned I'm a big fan of case cohort designs I think if we think hard about this we have the ability to bring in lower frequency case groups that would be Nor quote normally present in many of these cohort studies I think it's an ideal opportunity and probably that's the one big way I would modify the slide list is the ability to bring in I don't want to say rare case groups But but case groups that may not be represented at high enough power sample size Perhaps the other point is one that may not raised earlier about Focusing on individuals who are healthy despite having a high exposure to risk either environmental or genetic and that may need to be something That's captured explicitly Sure, but if there is a way of enriching for those samples Eric, could I so could you define what you mean by case cohort? These cases that's why There's some unresolved issues that I found that throughout the day bounce around One is families Once miles of your family's report They're there to take the role of them Right now Thank you unless we give it a little more thought You know, it's it is obvious that families have the ability to see another copy of a very rare allele That's that's an obvious application my my difficulty my problem with that is that that sibling for example Probably has all of the other variants that surround those that one rare allele and so It's difficult to disentangle. So I think number one is we need to do a better job of families number two last night I guess there seemed to be more enthusiasm than today about Integrating other ohmic technologies and maybe that's in the lumped in with other phenotypes But my own opinion is we shouldn't lose that whether RNA seek someone measured RNA seek date. I think would be very valuable I really would like us to see a mention of the ability to bring along other omics To my surprise one of the things I've learned the last 24 hours is I guess this would be for staff Yeah, or the investigators could meet We need to push continued development analytic frameworks We really need to push more analytic New analytic development and I think right and how the problem is that many of the people I know we're doing this work We're so overwhelmed with applications. We have data coming out of our ears There is we're not spending a lot of time on new analytic methods and the testing a new analytic methods So I hope I'm wrong with that statement, but I'm wondering if the time is right as part of this preparation for this project is pushing a Group to consider new analytic developments. Those are some unresolved issues. I see from this meeting I can show I Eric on the family issue. I think It's difficult. Well, we saw in GWAS that many of the cohorts did have families Within them. However in the analysis, they were kind of adjusted out. They were clustered out But so it's hard to think about combining an association and a family study but at NHLBI at least many of our cohorts do have families and if Those cohorts that had families built in could be represented in some way so that we could tap The power of the family study design perhaps in a second stage as Evan brought up That might be something to think about another Thing I wanted to throw out there is that NHLBI has an RFA out there to do sequencing in families at this point We're just starting out could be looked at as a pilot study Some of the groups are doing whole genome some are doing exome sit, but they're in smaller samples Other comments There are some Non-trivial sequencing studies under way involving families I think some in type 2 diabetes one of the types of diabetes products I'm not involved in is doing some work on large pedigrees Yeah Is it no one do we know enough to know? I mean that's potentially valuable information on informing these kinds of discussions, and I wonder what we know about how those studies have gone Results are just like the results for everything else so far Chris one other comment on families in terms of getting the most bang for your buck is you you can impute Unsequenced family members and actually gain more Power if you will by virtue of doing that we've done that in Framingham, and I'm sure it could be done as long as you have Substantial, you know first-degree relatives, etc Yeah, I mean I I think in the cancer world there You know 30 to 40 years of collection Within the NCI and a number of extramural groups that are deep into the exome and whole genome sequencing and it You know and some of them have actually looked at you know these data With f bat tests and like and seen some very interesting things So I think that there is there certainly is precedent and there's very large numbers of individuals there But one of the things it's really interesting about the family studies for younger Outcomes if you look in the literature for some of the cardiac outcomes and now autism Testicular cancer there are others where you see de novo mutations with larger structural events that are taking place that maybe germline mutations that are particularly interesting for you know a range of pediatric conditions That again family structure would allow you to get but you have to very specifically Hypothesize that you're doing that. I'm not I'm not sure that that would be a major goal But again this question of de novo events if you have family structure and knowing sort of what that Perhaps the other side of that is people who have these de novo events and live on and they're getting into the 60s and 70s without any problems You know one of the real advantages of the family is that it allows you to see a very rare mutation in Multiple individuals the larger the family the better and that's something that will be difficult across a population Some of them will be novel to the family, but they might tell us about a biologic pathway or gene of interest They have to say there's enthusiasm for thinking about how families could be part of part of this equation Let the course has risen up Just one question regarding the existing cohorts and the family information that's in them It seems to me that families are most useful when there's a high incidence of the disease in question within the family So how often is that the case? So It's it's hard to actually make the jump because what's worked well by putting the family consortia and Case-controller cohort together for GWAS doesn't apply that is it's very attractive when there's a big rich family Discovery group that can say I found something now walk it into the population setting or vice versa But for rare variants it wouldn't work that way, right? It has in melanoma the nature paper last fall You know they found it in families the MITF and then went into the general population found something with 1% and you know and you could see it in the general population very nice laboratory confirmation of disruption of simulation of The variant I mean it's a it's a nice example whether that's but that's what we do For things at 1% and above doesn't it by definition almost not work if they're if they're really idiosyncratic Not one offs, but it's I'm not sure it translates Because the same gene though might be that exact variant may not be but the same gene maybe so that focused you in on an interesting Biology, so that's I think would be the recommendation for your priorities is not that we try to put families in But that we make sure that wherever we sequence in a population group. It is Coordinated with an intense Family study and that only be done where there is in fact Yeah, but I don't think you can smush them