 Can you give us some examples of the 2 to 4% null alleles that you found? Yeah, I should have put the table on. You know, 15 minutes, it's hard to do the slides. Everything from autosomal dominant, non-syndromic, sensory neural hearing loss, hox mutations with subtle limb anomalies, factor 11 deficiency, name and organ system, there's something there. And that's what I think is so interesting is the incredible breadth and heterogeneity, which means that it would be impossible pre-hoc to gather the phenotypic data that you might want to have to answer the downstream questions. That's why I think we have to do it post-hoc, just wildly heterogeneous. Yeah, Gell and Lessa, great presentations. I agree mostly. One concern is you both presented a case where you found an actionable item. And I think one of the issues is in education. So how often do we hear a medical student misinterpret a negative result as meaning no future risk for something for a patient? And so I'm wondering, I think there's a need to balance the presentations when teaching and wondered if you had any comments. I think that's really an important point and I will say one of the work that every site did and then we worked across site was what the reports look like for sequencing in particular with respect to what the report is negative for because a likely pathogenic mutation in BRCA is not going to be reported, only pathogenic as an incidental finding, right? So to parse exactly the language to say, hey, we looked at these genes but we didn't look that hard or we didn't tell you some things that we might know and then how to look again, I think are critical questions. The education is not just medical students, unfortunately. I think we have our best shot with medical students. It's the practicing physicians that we have trouble with and I think all of us in clinical genetics have had patients who had a VUS in BRCA 1 or 2 and had a surgeon say, look, there's breast cancer in your family, you have a change in that gene, let's just do mastectomies and they get done and so I actually now when I talk to the patients I say like, and if the surgeon says to you, you know, then you say, Dr. Jarvik says no, you know, because they will walk out and VUSs are very poorly understood. People assume it's a changed gene, it must mean something and that's one of the reasons we don't report VUSs for incidental findings but the education piece I think is going to be much easier with medical students than practicing clinicians. I agree with that. I use medical students because that's where we do have our biggest impact right now and where we can educate. But my other comment is we also have to be careful that we don't educate students to the point where they think it's all in the test and that clinical evaluation or physical exam isn't important. You know, we run the risk of no one knows how to do a cardiac exam anymore because they just say, well, we'll do an echo. Right. And actually I think in some cases as Les has pointed out that you can have the converse that the test actually suggests where you should go back and do your exam more carefully and so we had someone who had a very disturbing looking in silico variant for tuberous sclerosis. So we actually went back to our IRB because we weren't returning VUSs even if they were likely pathogenic which just fell into that category and said, gee, we'd like to examine this patient with this in mind and I think your hawks had a toe abnormality and had never been really picked. The family knew they had weird toes but, you know, the clinicians didn't really pay any attention to it so that we will learn things by looping that way through. I should have mentioned that CSER just formed a physician education workgroup. We've taken all the CME activities across the sites and posted even national resources for physician CME and we're just starting to work with American Society of Human Genetics where there's a big push for physician education for genomics. I think Val, your question falls into the larger realm of a discussion that we rarely have in medicine which is an honest discussion about which mistake do you want to make and it's something we mostly avoid doing because people love to sit in rooms and scream at each other. One side says apples are red and the other side says, no, you're completely wrong, oranges are orange. We have a lot of these kind of discussions and we don't really get to what you're asking about which is what mistake could you make doing a hypothesis generating approach versus what mistakes are we currently making in medicine with the hypothesis testing approach which is what we currently mostly use which is a differential diagnosis based on a presenting complaint. We're evaluating a presented hypothesis there. That's what we're used to and I would suggest that we are mostly desensitized to the horrific limitations of that not that it should be supplanted or discarded because I completely agree with you, it must not be but when does that work and when does a hypothesis generating approach work? We figured that out for newborn screening and we had that discussion with newborn screening and we made a deliberate discussion, deliberate decision that we wanted a huge number of false positives with very high sensitivity. Those, so we said we want to work up lots of babies for newborn screening even though they don't have a metabolic disorder. That's what we decided and that was a good decision. Now my question is what mistake do we want to make here? I would say the trade-off might be a little different in adults and with genomics and it probably will be different for each phenotype. Malignant hyperthermia and gastric cancer I have a completely different mindset about what mistake I want to make in those two phenotypes and so I think we need to have those discussions and then that'll tell us when this will work and when we shouldn't use it. So I think it is the key question. Eric? So Les and Gail, thank you though, great. Gail, I want to compliment the CSER program for tackling cost-benefit analysis. I think it's sorely missing in this field. And I was very interested that it becomes quite favorable when you reach out and begin to contact relatives. One I wanted to call your attention in the 90s, WHO led an enormous effort called MEDPED for anybody who carried a recessive disease to try to find out all your family members who may have that. I'm not sure what happened to it, but I think the great Roger Williams led that program. But second, are you grappling with the LC aspects of contacting family members who have a consensus for research and how that's best done? Who's the right person to do that? We've been grappling that in a number of our studies. How do you phone someone and who's best to do that? Yeah, so one of the things that's been nice about this program is we've been able to sort of bring our IRB along with us. And so things, they weren't happy about three years ago as the program's been successful. They've, you know, so they didn't want us returning BUS as three years ago, but now they're willing to let us return to BUS. They didn't want us directly contacting relatives three years ago, but now they see we have something to say to those people. So that's coming along. We are interviewing both physicians and family members. It's going to be really hard to interest the family even when there's a good result. And we've definitely had that problem locally in Seattle. I can't speak for all the sites of you have a pathogenic variant and you still can't get the family to come in. You know, that's a clinical problem that we have as well. So, you know, or they're not talking to the side of the family or whatever. I mean, families are a really interesting thing. But I agree with you, we need to crack this nut. We need to figure out how to engage the families. You know, we've been talking like, let's go talk to Ancestry.com and see if we can get genomic information posted there because they don't do that. And I've actually had a meeting with them about that. Let's find different ways that people are engaged in social media and try and use that to... You know, it's really interesting because we have big HIPAA controls, but people go home and they post their BRCA results on their Facebook. So, you know, we should be able to use that to get the communication to the family members. You know, your aunt may not be on Facebook. So, there are definitely tools that need to be studied there. And I actually... I think that study, Carlos did a really nice job identifying how important that is to the cost-effectiveness. You want to add? Because you've collected more families over and more time. I agree with everything you said. Yeah. Can you talk a little bit, both of you, about who the patients are who get into the... who gets into your study less are they, you know, sort of asymptomatic, middle-aged nothing people or do they have a strong family history or what motivates them to be part of this? Because I think it's sort of, you know, we were talking about this before. Is this theorem based? Yes. So, we've actually done some work querying by interviews and focus groups with our participants to ask them that exact question. What are you in this for? What do you expect to get out of it? Improvement. Newspaper ads, word of mouth, whatever would work. So, it's a very heterogeneous ascertainment group, about 20% with atherosclerotic heart disease, 80% without, but agnostic to every other phenotype. And so, they are very heterogeneous and what they expect out of it was about equal doses of benefit, for themselves or their family of learning some genomic information that could be useful for their health care along with almost percentage-wise exactly equal altruism. They want to do something to contribute to the wider knowledge of how genetics and genes contribute to health. So, those two things, which I think makes this cohort very workable for the things that are coming on, and that's why I think what motivates them. So, for the rest of CSER, for the most part, the studies are phenotype-based. So, people with cardiomyopathy at one site, colon cancer at our site. But, Katrina Goddard's project is preconception counseling. So, these are women who present for, you know, preconception genetic screening as part of prenatal care and then they're asked if they want to enroll in her study. Robert Green's project does healthy people and they actually have the physicians return the results and they record that to see how the primary care physicians to see how that worked. Which is really an interesting project, right? Jim, I know you go out into rural communities and so you're not relying just on people coming to medical centers. So, there's a very built in purposeful diversity of the patient populations to look for, you know, what works for everyone, what doesn't work for everyone. Thank you both. I'm clearly biased because I'm part of CSER and I think it has a tremendous amount of benefit but I'd just be curious from both your perspective, Stale from sort of within CSER and Les from sort of someone on the outside but doing a similar study. These projects were intended to sort of push us down the pipeline right into genomic medicine and what do you think have been the biggest contributions of the CSER program and these types of studies to the field of genomics and kind of what do you see as the major next step? So, I think if you asked anyone in CSER what the biggest contributions were, you would get a lot of different answers. I do think CSER, just because of the timing and when the program started, was very, very important in pushing back on the ACMGs mandatory testing for the 56 incidental finding genes which was originally said like if they get a genomic test they get those results back and want it or not and so the LC group in particular there's a community in responding to that. I think that the impact on the ACMG variant classification criteria has been substantial and will be more substantial. I think that the cost effectiveness studies are some of the most important studies being done but little things how do you write the genomic report so everyone understands it. That's critical and so some of these things are not very sexy but they are essential to doing good medical care. I think that it has been a very spread out across all the work groups have done really important consensus papers and the sites have done important individual papers. So I don't know about that. So I'll stick with the larger question and not the small one. I would say that what I think that we have done in ClinSeq we set out to do three things initially. Number one was just to develop infrastructure in 2006 we couldn't even imagine what sort of an infrastructure you would need to do clinical genomics. Nobody had a clue and the sequencing instruments weren't even available then. So we thought we needed to build that which I think was correct and true. We wanted to pilot how you could use genomics in a clinical research context as I live in a total research context I have none of the Caesar sites I don't have any retail medicine exposure and so that's my domain. I want to see how that could work. And the third thing was to discover relationship between genes and atherosclerotic heart disease. And I think we succeeded in the first two and we failed miserably and relatively quickly in the first one which is I think exactly what we were supposed to do. Try things, fail quickly get on to something else and what happened is the story I related to you which is we looked at these exomes and we said for goodness sakes why are we limiting ourselves here to do the phenotype that we set out to look for in the first place. We should use it as broadly as we can use it and I think that's our big challenge and I think that having that opportunity to pilot this have some failures and have a success that we didn't even anticipate we would be working on is exactly the kind of research milieu that we need to support whether it's intramural or extramural I think that's where we need to be in the direction that's taken us not where we expected to be today but it's I think interesting useful and it will be fruitful for the field. We get a lot of queries across about setting up the genomic sequencing pipeline and results return informatics pipelines and as we were leaders and people are following we get a lot of people coming to visit to see how it was done. So 2-4% was higher than you thought and higher than Gail's number of those 2-4% do you have a sense of how many of those are important for their health care? Weird toes probably doesn't count as important for your health care unless it means you sort of don't have special shoes or something. Sorry those are two different numbers the 2-4% is when we wrote the ACMG recommendations for incidental findings that was our predicted yield for those 56 genes the 3% yield was the genome-wide null variant yield so those are but they're in the same range yeah but they're above the 3% how many are Let me just say the difference between his numbers and my numbers really falls to the Ashkenazis which he had represented and I didn't and that makes a big difference Go ahead You were trying to ask a question there now The 3% includes the funny toes Yeah What proportion of the 3% are funny toes and what proportion are bad cardiomyopathy So here's a really interesting question is where is the boundary between a disease and a trait? I have no idea where that is We drew in the papers and press I'll be happy to send it to you we drew what we thought was a reasonable boundary admitting that we were in absolutely gray squishy territory and you could very well argue it should be farther left or farther right You should talk about the hearing impairment patient who didn't know he was hearing impairment Hearing loss is that's a clear one cancer susceptibility etc there are some that are easy there are some that are PTC tasting and earwax consistency is clearly way over here where you draw that boundary I have no idea If I'm remembering correctly you had a not young adult who you predicted would have hearing impairment who said his hearing was normal and when tested that's why I like girls so much because the guy is my age his wife probably knew about his hearing but he didn't so all of the above is the answer Jay you had a question I was going to say things along the same lines or ask about similar things but the your study in particular does that lead us to any insights with respect to what fraction of people with Mendelian with Mendelian disorders have diagnoses like is that 1% of live births get diagnoses and you're coming up with 2 or 3% does that imply that 2 thirds of Mendelian diagnoses are undiagnosed so far is to put a fraction our sample size with a thousand people leaves us when we start you know we end up with 80 null variants half of which had a phenotype and half didn't so you know better than I what the confidence limits are they're big down there so I'm not going to put a 2 thirds number on it but your point I think is correct that I would suggest that there's significantly more out there than I think is out there and my first case that I started with the 66 year old lady with combined methameronic and methameronic acidemia there she is at 66 years of age bouncing around in her primary care docs office with these vague symptoms she hasn't been dealing with disease we never would have thought of that five years ago the genome told us we should think about that and I think there's going to be a fair number of those examples and it will vary enormously from different genes I got in a furious argument without sure at the genetics of common disease meeting last year and he said your results speaking to me are impossible it cannot be I said it was between 3 and 6 percent was my estimate he said that's not possible and he gave all the reasons why I thought it was not possible and I you know respectfully disagree and I think it's probably not that high for diabetes I think for other things probably neurologic diseases I'll bet you it's much higher than we think it is but not a new problem in clinical genetics since I saw a patient more than 20 years ago as an adult with lung disease and I said to the referring pulmonologist I think this patient has cystic fibrosis and the pulmonologist said if he had cystic fibrosis he'd be dead well he had cystic fibrosis but he had cystic fibrosis we see the tip of the iceberg we don't know a disease but we only know who's walking in complaining about that disease we don't know the people who are out there and actually I think that's one of the real power is to understand what's under the tip of that iceberg what's under the water Eric so Lessingale we treat for example cholesterol or blood pressure because of the risk of future disease given your two talks should we be treating genotype now so I was trying to be diplomatic about that Eric by saying it's very controversial and hard to push this out into the clinical community but you put a pretty fine point on that there was a wonderful report from the IOM a few years back it was called Towards Precision Medicine that was important and then the subtitle is A New Taxonomy of Human Disease and that's they are posing your question which is to say should we reorganize our taxonomic understanding of human pathophysiology with a primarily molecular focus as opposed to a primarily hemotypical and that is a question that makes most people's hair catch on fire or their heads explode because they hate it I was taught in medical school do not treat the test treat the patient that's a weird talk and what I'm actually suggesting is a little bit of treat the test I like that I can use that I'll fine tune that and I'll say if you have genes that predict blood pressure or genes that predict cholesterol I want to measure those intervening things but if you have a gene that predicts heart disease without some intervening thing that you can measure or that is useful and I think we can generalize that if you have the intervening trait and you can measure it that's always going to be the best data because it's going to take us a really long time to be clear you know we're talking about average risks so individual risks are a totally different thing as you will know awesome question Bob in the back that the payment system reimbursement system is really good for people who get treated symptomatically and not very good for diagnosis unless you're failing symptomatic treatment so there's a large number of people who bump along being treated symptomatically who never get diagnosed that's a good point thank you that's the work that we have to do is to build the evidence base to convince people that it should be paid for I agree with you I just wanted to back up on something you said earlier last about figuring it out with newborn screening that's not quite true we're still figuring out particularly with expanded newborn screen some things we're finding we're finding out they're not really disorders and so again I think that needs to be taught in addition I really like your concept of hypothesis driven but that can't spill over into the clinical testing of an individual what you create when you test an individual without a hypothesis or a differential diagnosis you create a huge Bonferroni situation where you have all these VUS's and you don't know what to pursue it's a great question and obviously there is enormous potential for mayhem doing that in an uncontrolled manner there's no argument about that but what I was actually surprised about was that my yield was 50% that the disease was I thought it was going to be much lower than that because if they don't have it the variant shouldn't be that powerful but it was powerful and in fact it's more powerful than we think it is because if you do the Bayesian analysis it's telling us that actually in spite of the Bonferroni problem the yield's 50% that's pretty high it's pretty high and so it's not 100 we all understand that but you know some of these people you change their odds of having Burt Hogg to Bay syndrome from 1 in 50,000 to 1 in 2 that's a huge change in odds even though the first one wasn't 0 and the second one isn't 1 I'll concede you that that's a big change and as a clinician I can use that change probability to direct the care of that patient in a fruitful way better than I can without that so I would say it is more useful than you might think so I'll condition and say actually maybe more useful than what happens to that you know 70 year old person is knowing the range of implications of that variant so when you have a baby with that variant you don't say 90% chance you're going to have some terrible thing and it turns out that there is a big range that we just don't know about so we have a very poor understanding of the penetrance of you know even the most common traits and the approach less is taking tells us at least the range it doesn't give us the whole curve but it tells us the range of penetrance and I think that's crucial information that we need clinically thank you both very much particularly Gail for coming so far he doesn't get reimbursement alright let's go back to concept clearance Tino you want to come up next on the docket is presentation from Tina Gatlin she's going to describe data analysis and coordinating center for the training and career development programs at NHGRI