 Why don't you turn them into questions and generate discussion? So can we back up one slide? Actually, can we go back a slide? Yeah, we can. I couldn't even do it. How do you do that? I did it. Excellent. Well, so the discussion is open. And we want to predispose your opinions. So since I pushed Dan into that, let me ask a question. I think throughout this, we need to be asking ourselves, what is the role of NIH and NHGRI? I mean, obviously, there are bigger questions, but I'm most interested in what our role should be as a research funding organization. So this is an issue for laboratories, obviously, and lots of other people. But it seems that there is a knowledge base that's needed here. And we're in the business of generating knowledge bases. We could say that all of the programs that we have so far are going to generate much of that knowledge base. And we just encourage everybody to put their data into ClinVar, and that solves the problem. But I have the feeling that's not the solution, or the only solution. So could you comment, maybe? I'll comment. I mean, I think actually one potential role of NHGRI is to do research that asks the question, what is the best structure for this knowledge base? What makes it the most useful? What has the highest public acceptance? And what is the right funding model for it, even? Are there laws that need to be changed? And I think NHGRI could support those basic questions of asking, what are the next steps? I'll also add, one of the biggest barriers to laboratories sharing their data is not actually their willingness, but the resources required to get their data out and to submit it. And it's a lot of effort. Most of the large labs that have submitted their data are ones that we are partially funding off of the ClinGen grants. So I think starting to think about how we can make this easier with better infrastructure. And I'll also say that ClinVar is really a variant-centric database. It is not case-centric. So very few of the labs are actually submitting their evidence that supports the classifications. And so I think what we ideally need is an infrastructure that allows both a variant-centric database that is tied to evidence that we can help share and display and allow us to build off of the knowledge that we're seeing in there that will help build the evidence for how we think about variants. Mark. So if I can also weigh in on that. I mean, I come back to the slide that you presented early on that shows the five domains. And I think tend to think of those as moving from left to right. But I wonder if there's at least an opportunity here to reverse the direction of that arrow and say if there are particular issues relating to genes and or variants that really come into the knowledge about biology and that sort of thing, is there a mechanism by which we could feed that back to the programs that are being funded through NHGRI to prioritize those areas where we think they might have the biggest clinical impact and actually make that more of a constructive interactive loop as opposed to a unidirectional arrow. And I might ask Eric to comment as well on that aspect of going, not backwards, but really I think what Rex has described as a virtuous cycle. So say we identify something clinically, we then sort of need to present it to our basic colleagues and I might ask some of our basic colleagues who are on the council. How do you get people interested in that? I mean, they may say, well, that's fine, but that's not my area and so I don't want to pursue it. So any thoughts on that? We have examples of that. I mean, certainly in the, I mean, I can think of common fund examples where there are sub programs specifically designed. I mean, the UDN has something that's trying to stimulate model organism work to try to get people who happen to be interested working on those genes to pick them up and pursue them in greater detail. I mean, and there are other examples we can either come up with or others can come up with. Well, or maybe, and while Carol's coming to the microphone, we can also think about the programs that we already have in place. So we have ENCODE, for example. Is there a way and maybe Elise would like to comment. You know, we're trying to promote collaborations and we have, I think, a very fruitful one going on between ENCODE and EMERGE. We had something very nice that developed between ENCODE and our Geneva program, which was genome-wide association studies. But suppose, you know, we just come up with something on a, you know, almost on a one-off level. How do we do that back and forth to try to drop them into a program and get them investigated quickly and, you know, divert resources from the things that they were gonna do? How do we have criteria for doing that? So I think part of it is having more representatives from those communities come to these kinds of meetings to hear what the problems are. And Cindy and Vence and I were talking about that just before this session started about the fact that there are new model systems. There's new technologies for better modeling of human variants in model systems to get at the fundamental biology. Those things are emerging and changing as rapidly as genome technologies are. And so it seems like it is a good opportunity now for us to bring those communities together to address some of these issues about what these variants of unknown significance are. So I think there's a lot of opportunity here that we just haven't leveraged to date. So I think that's a good point. But I do wanna emphasize that the challenge with this is not what we can modify in the animals to do this, but it's, will the physician be willing to write that in the medical record based on the data? And I think there's a huge gap between what we can do in basic research and then what we're willing to call clinically. And I see this over and over again. So I think that's another whole area that needs to be explored. And I know as part of the criteria that Heidi was showing was that there's different categories for this, but that's a big jump between what people will really use. And so I think we do have a major gap between people are looking for clinical evidence in clinical trials in order to show this variance is causal. So there's a gigantic gap in these VUSs. I think, I mean, I would disagree with that a little bit in the sense that I think that the practicing clinicians are need to have their educational levels brought up a little bit. So if the lab reports something as pathogenic, that's what they're gonna act on period. They don't critically look at any of the data sets. And as a consumer of this kind of information as well as a generator, that's what we see. And I see Sharon vigorously nodding her head. So I'm gonna stop talking and let her say the same thing. But I think that that, I mean, so there's the problem of the common variant and whether you act on a sip variant if you're gonna give a particular drug and that's sort of a huge area of controversy in cardiovascular and other kinds of medicine. But the rare variants, the disease-associated cardiomyopathy, cancer susceptibility, what have you, all of that stuff, I think clinicians rely on what the lab reports. While I have the microphone open, I just wanna throw one other thing into the mix. And that is that it seems to me that one thing that this field needs is large data sets. Large data sets of genotyped and phenotyped genotyped and phenotyped patients. And we're doing okay or we're sort of making a little bit of headway in white people, but I'm not sure we are getting anywhere with anyone who's not of European ancestry or not making progress as fast. So it's sort of to state the obvious. And I think that in this particular space, if you sort of wanna spend a little bit of time wandering through the XX server, it's really pretty interesting. You find variants that are cosmopolitan and you find variants that are clearly confined to a single ancestry mostly. And my own field has made the mistake over and over again of assigning something as pathogenic when in fact it turns out to be pretty common in another ancestry. So I think that at the very least, we have to figure out ways of generating those kinds of large data sets. I'm not sure what that will involve. The genotypes are getting there, the phenotypes are behind. So Gail, Heidi, Pierre, I don't know. I don't know, is this an international effort? Irwin, Irwin. I'm on, okay. I think there's another aspect to that can bridge the gap between the basic science and the functional interpretation of variants and the laboratory's difficulty to interpret it just from the laboratory's perspective and that for a good number of phenotypes there are intermediate biochemical markers. I mean, I just wanna mention things like endocrine disorders, et cetera, where you can measure hormones. And one, I think systematic approach could be to actually as from the laboratory's perspective to feed back to the groups that deposit and say, was this measured? You can think of parathyroid hormone, things like that, where in some instances this might help clarify that there are abnormalities in clinical laboratories that might help as intermediate phenotypes. Sharon? Well, I think the other reason we need to bring the biologist to these meetings is that I think one of the fundamental problems has been that many of the published functional assays are not quote unquote clinical grade. Laboratories often will take one or two known pathogenic mutations, maybe one or two known benign mutations and they'll say see it works and then they'll make predictions. So Fergus Couch actually was invited and gave a wonderful presentation at ClinGen about the work he's done funded by NCI to do functional assays for BRCA one and two and he showed all of the very large number of controls they've used both on the positive and the negative side and you really had a much better idea whether you could get intermediate values or not which were relevant. And many people in the field are looking at ways of incorporating those kinds of assays but we really need the functional assay experimentalists to realize how complicated this problem is and that you really need a large data set. The other thing is there's some very, very good functional assays that do not actually correlate with clinical activity. So the protein may have a function but that may not actually be the important function for the disorder. And so again, you really do need the people doing these assays, talking with the people familiar with the diseases and the laboratories to really develop a robust enough system that it can be incorporated into the classification. Are we talked out? No. So I think another thing is the importance of the longitudinal data, especially in the pediatric population where these variants are being identified and a lot of the outcomes won't be known for many, many years and just taking advantage of that now to follow these kids. Well, and maybe I might just know one of the challenges that I think we face in any funding organization or in any study is, you know, when should it end? And we do face this at NHGRI. We need to close down some things because the budget isn't infinite in order to start new things. And so when do you decide, okay, that's enough? Or can we put systems in place that will allow that kind of outcome to be collected at relatively low cost? And I don't know how feasible that is. I mean, one of the challenges related to that, Teri, is that I think we've noted in a merge that the children's hospitals then lose those children at 18 and that there are very few healthcare providers. There are people like Kaiser and Group Health that do, that a lot of children will stay into adulthood. They'll even, when they get their own insurance, it'll end up being the same insurance. But a lot of the studies that are centered at children's hospital then lose those children at 18 and can't follow them in the same way. And so that's a challenge. And I'll also add some of the, I think one of the critical things is the engaging a patient cohort in a much deeper way. I saw an example that 23andMe had done where there was a colon, a VUS and a colon cancer gene. And they literally sent an email out to over 10,000 people and within a matter of a few days had the phenotyping they needed to resolve the variant. So some of these things are gonna require re-phenotyping and ongoing collection of data and even engagement of patients I think could be not only highly useful but incredibly cost effective where you just send an email and get a reply back in minutes. So I think we have to think really differently about how we engage the public and cohorts to try to help define phenotypes that would help resolve variants. I guess I would just echo the need for this type of data and think about how do we bring 100,000 UK genomes, 150,000 Saudi genomes, the Geisinger Project, the Personal Precision Medicine Initiative, a million genomes theoretically to bear on this question and other national initiatives. It seems that the community working together is a theme that we've talked about but this is a real area where there could really be some high value coming out of it. Howard. And then Steve after me. One of the things on the other end of the spectrum is the consistency of interpretation for the non-expert groups. And in particular I'm thinking of CAP and AMP and all that and yes, those are expert groups but they're not as expert as some of the folks that are in your group and in terms of interpreting this. And I'm wondering whether just as we need to engage with the basic folks for trying to look at mechanism, one way to normalize things would be to put more effort on making sure that CAP and AACC and AMP and some places that maybe have no business being in this area are doing it a better because they're gonna be there whether we want them to or not. And some of you are members of these organizations but CAP in particular with the way they send out the blood samples for you to see how bad you are, there's some great opportunity to really improve things there, drive it in some way. Sounds like a good question for you. Heidi, and then you. So you're a member of CAP, right? I am a member of both CAP and AMP and ACMG but I think the challenge, so the proficiency testing is incredibly useful for laboratories but the interpretive side of that is extraordinarily small. And so I think the way that we've approached it in ClinGen is that in the clinical and main working groups and the expert panels there is a requirement that you engage the whole spectrum of different perspectives and that includes both the basic science researchers, the clinicians, the clinical laboratory staff, even sometimes patients but all of those perspectives which I think are incredibly important, those people do represent those different and it's been interesting I will say that particularly in one of our cardiovascular domain working group, each of those perspectives does see the world in a very different way and it has been challenging to bring them all together and think about it the same but also those different perspectives I think has enriched the conversation because they see things different way but I think, I don't know that CAP would have its own working group necessarily in a clinical domain, it's more getting the people that are in the clinical laboratory that are part of CAP on these work groups, does that make sense Howard? Yeah I think getting them involved and there's a point where you have too many people involved but I think you're certainly with the CPIC, I've noticed out in the wild that it's starting to get some traction not because they want to be experts but because they don't want to be experts but they also don't want to give up their patients and their business model and so they've engaged the guidelines because it helps them be better without having to be better if you see what I mean and so I think there's some opportunities here for us to help the same thing. I think that we need to look for opportunities in that space, I mean the fact is clinically 10 years ago if I had P10 testing I sent it to Karis Eng and if I had OI I sent it to Peter Byers and there was really the expert and so now in the era of panels and incidental findings which is not even what the labs necessarily want to be looking for, every lab has to interpret basically everything, the pediatric labs have to interpret the adult cancer genes and so it's a very hard change to implement to go from those experts who know those genes inside out to people who don't and people who won't know what the functional assays for that gene mean if they're important or not because they just don't know the literature so we really have to look for opportunities to standardize because everyone is now working in a space they were not working in five years ago. And actually I'll just add Howard because as I think about your comments with respect to CPIC I think we're engaging the clinical communities in a very specific and targeted way is around the guideline development and the implementation side of it. So for instance the work that CPIC is doing and coming up with standardized nomenclature and I think there's an amp or cap committee that's trying to help with that nomenclature question amp, so what's that? Yes, tumors but they've also been doing it for pharmacogenetics and I think having those guidelines in the same way that our guidelines came out of ACMG not ClinGen I think helps with the implementation side of that and so ensuring that you're working with the clinical professional societies in trying to take the work that an expert group may do but to get it really implemented effectively I do think requires that engagement in the professional societies, maybe that makes sense. Yeah, I wanted to go back to a question Terry raised earlier in the discussion which is the role of the genome institute in funding infrastructures to meet these clinical needs and to encourage us to think about infrastructures as learning healthcare infrastructures that support both clinical and research needs because I don't think they're mutually exclusive and I think that there are ways to design infrastructures that really meet both sets of needs at the same time and I have some models of mine in my experience in pediatric oncology and forms my thinking here but how can we do this? There are these really important needs on both sides and they can be complimentary rather than mutually exclusive. If I could follow up on the comment in the pediatric oncology group, I mean to me this is really one of the most important demonstrations of how to do it correctly. Now I think there's some historical interest in war work and I were having an opportunity to talk about this at the break but we talk about the fact that in this country payers don't pay for research and they don't pay for gathering data and that sort of thing but the reality is is that every payer pays for pediatric oncology care and 90 plus percent of the kids in this country that are being treated for pediatric cancer are on a protocol. So in some sense there is a seamless coverage of the outcomes acquisition and that. Now some of it's a historical accident because this group I think it was initially the children's cancer group or whatever it was assembled before health insurance really became the huge industry that it is now and so I don't know if it's possible to engineer something at the present time that would look like that but I think it's really a powerful model to examine how we might be able to do this because if you look at when that group was started when the mortality rate of pediatric cancer was essentially 100% to now where it's 80% cure I mean it's spectacular. So Mark that makes me think of something that we've tried to do in other fields in terms of research that CMS would support that they would only pay for something if people are part of a study. I've forgotten the name but. Coverage with evidence development. Thank you Mark coverage with evidence development. So CMS mainly deals with people who are age 65 and when we've talked with them about genetic and genomic research they sort of say you know for the large part it's not our role. However I think we have been pretty effective in these meetings and in other settings and many of you in at least opening dialogues with the major health payers in this country and so wouldn't it be cool? And we start a lot of neat discussions at NHGRI and lots of big programs with the wouldn't it be cool phrase but wouldn't it be cool if they were interested in doing the coverage with evidence development work with us and basically say we'll reimburse a test if they are part of X study. And I wonder at those of you who have experience with those groups and unfortunately we don't have them around the table. How feasible might that be? Actually you have some experience Wendy. Well I mentioned this to Jeff at the break but I spoke with a molecular pathologist who, well I guess because I'm part of some of the NIH CMS meetings and the NHGRI CMS meetings, it came up that genetics I guess as a whole maybe me, maybe not ought to have a liaison with the Palmetto group which Elaine Jeter runs. And so I met with somebody in their group who is very keyed into these issues as a molecular pathologist. And my sense is that they're very open and realize that they need to do coverage with evidence development studies but I think that they're grappling with where does this information go? How are they gonna structure it? So I guess my feeling was that some of the information is result of tests and variations that need to be deposited and they're quite willing to say okay this needs to go into a certain place as a requirement of this type of study coverage with evidence development. And we talked about whether it would go into a proprietary database that's all these are emerging at the same time ClinVar is trying to gather data but they realize they would need to put this into a database that is going to be supported over the long term and have public access. So I think that part has been smoothed out if the conversations will continue. I think where they really need help is you know there's lots of data, it's unstructured and how are they going to get help structuring it? So you know the idea is how would that fit in with some of the studies that are happening here where they're already being structured and maybe do it in an area that they're very concerned about and you're very concerned about at least is a one or two demonstration projects. Maybe. Peer isn't it? Okay. Thanks Dan. Just before Heidi mentioned 23 and me I was just going to ask if there were any potentials for collaboration with industry partners on some of this because it seems to be a lot of potential value but I know lots of potential issues as well but I was wondering if that door was open. So one thing I'll say is that through ClinGen there is a fairly strong engagement of the commercial space in part because they do play a major role in clinical testing. So a lot of the data that's being submitted is in fact coming from commercial laboratories that are sharing that data. I think a challenge that I see is how we think about the infrastructure side. The amount of data we could capture if everybody used the same system would be extraordinarily large but should NIH actually contract the development of laboratory information systems so that it can then help labs structure that data and capture it or leave that to the commercial industry competition? So I had another thought with respect to the NHGRI portfolio specifically or sort of thinking about that. It seems to me that we've developed evidence or we are developing evidence using sort of two different models. One is sort of capturing genomic variation in populations and then going back to phenotypes and then the other is to start with specific phenotypes and understand genomic variation across large sets and one question for all of us in this room I guess is what are, number one, are there other ways to do it and number two, what's the right balance between those two now and going forward? I don't have an answer to that but it just seems to me that we've sort of conceptual, we're generating evidence in two specific or two different kinds of ways and they'll converge eventually but balancing the portfolio is gonna be a challenge. Robert Green. Well I was nodding vigorously because that's a, I think a key point and I think it's a point we're gonna talk about in our session afterwards but just to say that I think moving, this is in part echoing some of the points that Les B. Sicker and others have made is that moving from variant, presumed variant pathogenicity to both phenotypes and intermediate phenotypes is a very fertile area for us to explore and I think that really should be part of the clinical data gathering in these large scale projects starting with the ones we have and moving into the other ones that Jeff mentioned. Well if there's no other discussion then I think I'll just get the slides up for the summary, so these are summary thoughts that were written down before we started this conversation and I'm not totally sure that they're gonna be perfect but maybe I'll read from Jonathan's laptop. No, I think I can see it from here. So the summary recommendations, number one, there are, as you've heard, critical knowledge gaps, lots and lots of variants that are BUSs and the problem is that while we're getting good at collecting genomic variation data, we're not that good at collecting phenotypic variation data and that seems to me to be something that the working group and people who set priorities for genomic medicine across the institute and the institutes need to think about. There are inconsistencies in the systems and there are costs associated with all of that next. You have the slides. So tools to, one recommendation is better tools to support variant assessment. Heidi alluded to this idea of a web-based environment or engaging patients using open source models and then some way to easily access the data. I think XAC is an interesting model for that and I think the 23andMe anecdote is an interesting model for that and whether those can be exploited or whether those are dead ends, I don't know. Heidi wants the publication process to require submission, that's in bold letters, of interpreted variants to ClinVar and supporting evidence into accessible databases and that's a recommendation that I'm willing to stand behind for now and then the whole issue of how electronic medical record systems get incorporated into this, it is relatively straightforward, although cumbersome to genotype people who have electronic medical records. One of the things that we're discovering is that the electronic medical record is an incomplete phenotyping tool and I think that's an interesting observation and if we could then figure out mechanisms to go back to patients and providers and say you have this interesting genetic variant, now go and phenotype yourself is one thought, go ahead. Yeah, there are huge training needs across the spectrum of providers and participants in this activity sort of, so from medical school to genetics programs to postdoctoral programs in basic sciences, including genetics and beyond genetics, residency programs and then the whole issue of educating healthcare providers, I come back to the idea that healthcare providers need to understand the nuanced nature of genetic information that sometimes you get an answer and sometimes you get a maybe answer and they need to understand that the maybe answer maybe doesn't deserve the same kind of intervention or activity that the real answer does. This is a particular problem in I think cancer prevention and it's a particular problem in the cardiomyopathy and arrhythmia susceptibility spaces where the interventions can be pretty life-disturbing. Next. And then so this idea of a virtuous cycle so we need high throughput approaches, the relationship between basic and clinical sciences we've talked about and again the idea is that collecting just genotypes won't get us very far, we need genotypes with phenotypes and re-phenotyping systems and those are all words that are written down there, I think that was the last slide. Yeah, okay, so other comments, that's our starting position, I don't think it's changed much over the conversation, I have some notes and we'll flesh that out later. So maybe just to ask the question, we hear over and over again we need to get these data submitted to various databases and I think Heidi has pointed out some of the barriers to those and maybe we can address whether some of those are either researchable questions or places where NIH can be of help, so some of the barriers to those obviously include consent and hippoprivacy rules and that sort of thing. I don't know that we're doing a lot in terms of ClinVar per se for that, I mean we're obviously doing a lot of research and a lot of churn, frankly, in this area. But how much of a barrier Heidi, do you see that as being in terms of healthcare systems and others being willing to share these kinds of data? So the question is related to what kind of consent has or has not been captured in order to facilitate the sharing, is that, so I think this is a great question and a great area for some work and one of the things that we're working to do within ClinGen, we've drafted a consent form for use in the clinical context, so a lot of effort has gone into consent language in research studies to enable data sharing, but very little effort has gone into during routine clinical practice when a physician orders a genetic test, are you capturing simply the willingness of the patient to allow their results to be shared, even if they never enter a research study and never sign a research consent? And so that's something we're focused on is a very short, less than one page clinical consent language that could be inserted in the test requisition process to just simply allow that data to be shared and I think that has been a barrier, like my IRB will not allow me to share case level data into ClinVar or any other source unless the system is restricted access or I've gotten consent, so I think thinking about the language that could be utilized purely in the clinical workflow that would allow robust data sharing that doesn't require the long research, IRB approved consent forms, I think will become critical, as well as actively engaging the patients for them to understand why sharing their data is of incredible benefit that also allows re-contact, which for re-phenotyping continuous gathering of data will be critical. If I could just add on to that, I think what we're really talking about here is safe harbor, which is, this is critically important, so how do we define from a policy perspective what would constitute a safe harbor and this is of particular relevance to the, sort of what we're all anticipating, which is that at some point, the proposed rule for the common rule is going to be coming forward and of course in the advanced notice of proposed rule making, one of the things that was in that was that DNA is inherently identifiable, which would bring it under regulatory purview of things like HIPAA and that, which could dramatically impact this, so I think that there's a role here for conversations in the policy and regulation side about how can we balance the need to develop the evidence that we've so critically identified with the importance of protecting privacy and have those somehow synergized. In terms of the sort of leveraging existing resources, there is the newborn screening translational research network and I know that NSITE groups are developing common data elements to utilize and the amount of genome data that will be put into this is still under discussion but there are resources in terms of a consent, specific consent documents that are being incorporated into our projects and so that's a potential resource where more information on the genotype as well as phenotype can be used. Jeff. Yeah, I have two thoughts. One is going to come up, no surprise to Dan and others is that it wouldn't be nice if we had a three generation or more family history on everybody that was getting a sequence done so maybe that's an opportunity for Ignite and these programs to at least our program and others to collaborate. I think that would be really rich actually. The second is about 23 and me so has anybody had a conversation with them about their whole way of engaging the public and the patients in trusting them? Whether that trust still exists, I don't know but that's the low bar conversation. The high bar conversation would be about their data. They have an enormous amount of data associated with self-reported phenotypic information that seems to be useful in providing reproducible genetic associations by at least from the literature so is that an opportunity that's worth exploring? I just asked that as a question. Maybe some of you have already engaged them in those conversations. We haven't engaged but. The Stanford part of ClinGen, Carlos Bustamante has had discussions with 23andme and Ancestry.com with regard to use of their data particularly for allele frequency. I would just say though that I'm concerned that those types of direct to consumer databases still don't really reflect the United States population. They reflect people that have money to spend and so I do think we have to look at other ways of getting better population data that really represent our country. The phenotype data are what they are, it seems to me but the great potential of that model is this re-contact idea that Heidi alluded to that you sort of say well I wonder among the people who have this particular genotype is their phenotype X and then going back and reinterrogating patients. That I think has some value and I totally agree with Sharon's point about the representativeness or not of that particular cohort. This is a conversation that I think is going on in the Precision Medicine Initiative as well sort of how to design a cohort that is mHealth enabled and yet representative of the population anyway. Maybe one last comment, Alexa. Actually I was going to make a Precision Medicine comment because in fact we're talking about a new million person cohort and so this would give us an opportunity to actually test out some of these ideas. Question is how perspective is this cohort gonna be or are we putting together retrospective data? And I don't think we know the answer to that question yet. Thank you to all of you. Time, yeah, time's up so I appreciate it. How would Jacob be your lead minister? Maybe while Howard slides in he's going to the podium and that let me just ask maybe Sharon or Gail or others Heidi that those of you who go to insurers and ask them to pay for this kind of stuff. How much of a barrier would it be if they agreed with us that yes, if we're gonna pay for testing it has to go into a database. Would that be a huge barrier for you? Is it more of a challenge for them? I have to say that I've spent the last couple of months really being unbelievably dismayed by the lack of wanting to cover genetic testing at all. I mean retinoblastoma and things we've tested for for decades. I routinely now have to send letters and argue and so I'm not real optimistic. I think on the other hand I think the insurers would be fine if they were told it's a good idea to use a lot to tell laboratories that they'll only cover testing who submit data. I think that type of routine programmatic idea they would not have a problem with. It would be a barrier for you is the question and we have to go to Howard but maybe think about that. We can chat about it. Yeah.