 One of the real questions is how do we apply these marvelous new genetic tools in our research? And I guess one perspective is that if you want to look forward in terms of use of this, it's probably medicine isn't the right place to look forward. Look at that, because the whole genetics revolution has already had a major impact on human health, and it's probably been through nutrition. If you look at rice yields, this year we're looking at the largest rice crop in the world, yield in the world, and obviously we're going to need it given the commodities issues, and a variety of things having to do with plant genetics, et cetera. Just to give you a little introduction, my own experience with this is tap roots. This is my vineyard in upstate New York. I have a vineyard on Cucal Lake, which is one of the finger lakes which grows most of the wine in upstate New York. Some of my grapes are these. These are good old Concord grapes. These are American grapes. Nobody's been fiddling with these. These are Native American grapes, but you don't see a whole lot of bottles of Concord wine on the shelf, at least not anymore, and ours goes to Welch's Grape Juice, and so those haven't been fiddled with, but if you're going to go into something more interesting, there are these all sorts of gene products, and most of that comes from the Cornell Experimental Station, which is in Geneva, New York, which is about 10 miles from my vineyard, and I like to go over there and watch them fiddle around with their grapes, which is practical genetics, and these are French bridles which are bred for cold hardiness, tannin structure, all sorts of things, and you can kind of pick out. So if you're into agriculture, you know that genetics has already had a huge impact and provides us with beautiful products. These are some seedless grapes, which are table grapes. They're obviously crossbreeds of crossbreeds of crossbreeds of crossbreeds. So genetics has already had a huge impact on human health, but we're into medicine and the clinical side of things, so what I wanted to talk about was, talk a little bit in finishing here about some of the inferences we can make, some of the variants are caused by disease, and kind of switching from what the usual evidence we want in epidemiology to the new context of the results of the tools we have, kind of a translation of those results, the inferences we make in epidemiology to the tools we have in genetics. Then really talk about personalized medicine a little bit. A lot of places are starting personalized medicine programs, and obviously many of these are based on genetic tools. This obviously is the whole course, but I want to just touch on two issues, genetic screening and pharmacogenetics, and then talk a little bit about, very briefly, some application, further applications to genetic tools to research and experimental studies. Well, I think the first thing that struck me as I retooled in genetics and took the Bar Harbor course and read in my various textbooks, et cetera, is that genome-wide association studies and the whole genetic genomic revolution already has had a tremendous impact on basic science. And one of the comments on Terry's last lecture was that as you go and look at these genome-wide association studies, I'd ask you how many epidemiology lectures have you seen recently, which right in the middle of the results will have a cell expression study or a histopathology analysis or a knockout model, a mouse model, in the middle of what in essence is an epidemiology study. This just totally blew me away, and they're the functional studies that Terry was talking about. They're very appropriate there, and so to some extent this is reverse translation. We're talking about translational science and going from bench to bedside. Well, this is actually going from population to the genome-wide models back to say, what is this? It's a snip in the middle of a gene desert. Okay, folks, go out and find out what's going on here. And the message is not to other epidemiologists oftentimes. It's back to our basic genetic scientists. So this is, I was asked at a conference, or somebody asked the audience at a conference when they tell me one thing that population science has done for the basic scientists. Well, I got up to the microphone and basically said, well, here it is. Here's a whole lot of work for a whole lot of people of insights that we didn't have before. So this whole insights into genome structure and function, you can see that just in the matter of a few years, the interest in some of these introns and regulatory elements obviously is renewed interest sparked in them. These novel mechanisms of disease obviously are these insights that we actually probably know very little about how diseases occur. From some of these come proteins that assess therapeutics, particularly in the area of clotting. But obviously some of these have been cloned proteins and these are useful. And then the other things, of course, was that as we come up with these novel mechanisms of disease, you have new drug targets. You find another receptor, another enzyme, another. And I've been very impressed with these ability to mass screen small molecule inhibitors where you can come up with a gene and then just run through hundreds of thousands of small proteins and get some hits to ones which will inhibit that receptor, etc. And then go on to animal work phase one. And so the rapidity and the magnitude of this is just mind boggling. But certainly the fuel for this are going to be these novel drug targets which obviously deal with novel mechanisms of disease. So if you, I think some of Eric Lander and some of the visionary people in this kind of cringe a little bit when you try to pin them down about how genome-wide association studies have affected clinical medicine. But no one's arguing about how much they've really opened our insights into basic science. But I'm a little more optimistic than that. So you're all familiar with the surgeon general's criteria. And so what I thought might be interesting in this kind of summary way is to say we're reading all of these genome-wide association studies. How can we be sure? In the same way, how can we be sure that a behavior or a physiologic measure or a laboratory measurement has to do with causality? And you're all familiar with this. But let's look at each of these in the context of the genome association issues and maybe make some comments. Well, Terry's already mentioned to say, well, the genes there, the temporal relationships, not an issue. And I guess that's true. But there obviously is a subgroup of individuals who are also interested in expression of the gene, not just its presence in the genome, but its expression, these expression arrays and a variety of things. And this is a temporal issue because what this suggests is that the disease occurs during which time a disease is not quiescent. It's being expressed in its acting. And so even this temporal relationship obviously I think could have some opportunities for studies which looking at not only that the gene is present, it's being expressed, and some of the functional studies that have been talked about deal with that is that these are expressed in the tissue of interest, for example. I wanted to comment on the strength of association again. The nice question came up before. But what you have is multiple SNPs and other gene variants and they all add to risk. And one of the real questions is what is the composite risk of all the variants known and unknown. And Terry's shown a little bit about how those might be explained. But I wondered if, well, and then we're going to talk a little bit about dose response is also relevant here. And in this context it would be the number of alleles. In other words, none, heterozygous, homozygous for the susceptibility will. So you can do dose response within this, obviously, and the recessive versus dominant. One of the things I've been thinking about was just to do some back of the envelope scratching out about this idea that was discussed about why you get such small odds ratios. And let's just do a little hypothetical study of a case control study with, say, 2,000 smokers and 2,000 people with a disease and 2,000 people without disease. So whatever disease you'd like, cancer, heart disease, whatever. And when you do the odds ratios, of course, you get an odds ratio of 2.25. And so this looks at the entire exposure by cigarette smoking. But what would you think if, in fact, you weren't smart enough to do all that together? And in fact, you had to dice this up into individual exposures within the same class. Let's assume that 10% of individuals smoke camel cigarettes. All you can do is measure camel cigarettes. Well, here what you see is 10% of the individuals or smokers. The denominator here is do not smoke. It's do not smoke camels. And in this instance, then, you get a much smaller odds ratio. Less than half is strong, really, in terms of risk above 1. And I think this is what's happening in many ways. I'm not even talking about gene-gene associations. But if, in fact, you put together Marlboro's and Winston's and Kent's and all the other coffin nails that we could come up with, you would be able to add all those together and come up with such. And I think that's just in plain terms kind of an illustration of what we're doing and trying to put all these risks together. And again, I think we may not know all of the culprits in this area, but just to try to put it into something that could be a little bit more quantitatively understandable. This is just this illustration of the alleles. Here's risk genes associated with breast cancer in the study by Easton. Here's the odds ratio per allele. And if you look within the data, the odds ratio associated with being heterozygous for these risk alleles or homozygous for this, you can see obviously there is this dose response. Perhaps not so impressively down here. But the point is, is that I think one can fulfill what we like to see, of course, in our studies as some dose response, although the dominant and recessive issues, I think, become more complicated. We just had a lecture on replication of finding. And I don't want to say more than that other than that the genetics community and particularly Steve Chanick and Terry's leadership I think have been really quite much more militant about requiring these replications than the epidemiology community has had in its previous studies. And I think this has been a very positive response to this alpha error issue. The biologic plausibility, of course, you've just heard what this is about are the functional studies about why this could be occurring. And these are an important part. One wonders if one needs in vivo studies, some of the knockout or knock-in models would be at what point, what are the standards of functional studies? You'll see tremendous heterogeneity. Some of them will be essentially a bibliographic look at gene location and possibly expression. Others will get into some of the other much more convincing issues of actually gene product measurement and tissue expression and animal models. So I haven't seen a lot about kind of what would be kind of your minimum requirements. The consideration of alternate explanations, obviously some of these models are very complex. And so when one finds an association one wonders how these all fit together. This attribution of genetic risk I think would be one approach to say, if we can understand much of the heritability or the familial association, we would get an idea that we've identified all of the parts of these complex models. And I think we're still away from that. So this whole heterogeneity obviously I showed this slide before in terms of the possible explanations and mechanisms of heterogeneity. And I think these we need to keep in mind. The various sites of course, again many of the GWAS studies have looked at SNPs related to exons and introns, but this whole area of regulatory elements and how these things interact on a single gene basis to say nothing about a multi-gene basis obviously is part of the complexity that, and so these alternate explanations unfortunately are virtually infinite. This is the Eastern study again and as part of that analysis, actually this is a second Eastern paper which specifically I think was looking at this issue of additional familial risk. So the point they're making is it's a known breast cancer loci such as a BRCA and such genes really only explain 25% or less of familial risk of breast cancer. Now there could be environmental and obviously but in general looking at the family histories of breast cancer and the risks that they had in parts only about 25% of that could be explained by the known genetic markers. So what they did was a two-stage study which essentially excluded individuals with these markers, 4,400 cases, 4,300 controls and within the replication of SNPs in these huge numbers of cases controls and found these five novel loci related to breast cancer at the 10 to the minus 7th and these novel loci then identified an additional 3.6% of risk again possibly on top of that 25%. But just to give you an idea of the complexity and the things ahead they also noted in one of those QQ plots that Terry showed you is that there were 1792 additional SNPs associated at the P less than .05 level where there are only about 343 expected suggesting about 450 additional SNPs in excess that would appear to be having something to do with susceptibility and maybe having a role in the complex nature. So this whole idea of alternate explanations and the ability to account for familial risk I think is something that's going to continue to be a challenge. This issue of cessation of exposure is interesting. Obviously this knock in or knock out these gene replacement therapy is not a topic I'm going to get into. Obviously it has its own levels of controversy. But we do have a number of interventions which replace defective gene products to suggest that and I'm going to give you one from Dr. Collins' laboratory but certainly familial hypercholesterolemia for example. The heterosegous FH we've been working on this for a while in essence of up regulating the LDL receptor to make up for the one that doesn't work. But I think we're going to see more with the insights of the receptors and the drug targets that they offer us. We're going to be seeing a lot more of this and there are several exciting ones rumbling around now in the development stage. The consistency with other knowledge obviously has to do with functional evidence again et cetera and animal models. And then finally the specificity of association obviously was kind of based on one gene, one protein but even that I think is being shaken a little bit by now a couple of examples and I'll show you one of these shared association of diseases with gene variants in which you would appear to have one variant which is related to do different diseases. And so this specificity which is one of the lower ranked levels of causative evidence may in fact not be so causative. This is from Dr. Francis Collins' lab. It's a new in the journal paper published recently on this progerious syndrome. You've probably seen cases of this. These are children who are born and undergo an accelerated aging process to the point that they die of cardiovascular disease. Interesting from the dropout of cells in their arterial media by about the age of 13 years. So obviously a severe disease. And now the defect has been found with modern genomic tools. It's this substitution of one glycine to the next in a codon which has been basically a cryptic splice donor that produces an abnormal protein. This obviously gets to be kind of an unusual mechanism and variant that leads to this abnormal protein laminate which has to do with a chunk of it missing so that it cannot release from a tether site on the nuclear membrane. And then as the cell tries to transcribe proteins this alters transcription and they have widespread growth, failure, etc. The idea here was then okay now we've got a gene and a gene target and once you understand all of these things which this new in journal paper describes then what you have is the opportunity to fill with some of the gene products and what's going on downstream and basically in animal models and in cell models inhibition of this enzyme prevents this anchoring of this abnormal protein which then thereafter can't release and buggers up the cells. So what essentially as Dr. Collins describes this this goes in from some of this descriptive work and some of this functional work rapidly into an open label clinical trial of an inhibitor of this to see if they can get these children to start growing. It just illustrates the idea of once you understand the gene targets and then can survey for these small molecules that will be able to affect those gene targets the opportunities that have never we've never had before of really doing something that otherwise we would have said it's just an untreatable genetic disease. And I think that's part of the excitement of the whole thing and there's a number there's a subgroup of the Marfan syndrome which has a similar kind of thing with actually a commonly available anti-hypertensive drug as the drug. So a lot of things that just really aren't so far-fetched even though it's pretty complicated stuff. This is one of these studies in which kind of looks at the specificity of this association of this SNP and this particularly confers risk for prostate cancer but it looks like it's protective of type 2 diabetes. And again there are many questions arise about how this single SNP could be doing this but in fact within the same study done in Iceland and then replicated in a larger group of patients here you see the cases of controls quite a substantial number of cases of controls. There's an odds ratio of 1.2 highly significant as a causative as a susceptibility factor for prostate cancer. But then within the same study again once you have your genes done in a group you can do cases of controls for a number of conditions and here you have for type 2 diabetes case control studies you have an odds ratio significantly less than 1 protective of type 2 diabetes. And in fact there apparently, I didn't know about it, but apparently there is a literature which shows an inverse relationship in populations of prostate cancer in type 2 diabetics. But to some extent the insight of that was heightened by this potential genetic basis for that. So some of these specificity of these genes may not be particularly having to do with their mechanism of action. Let's talk a little bit about personalized medicine. This is a very popular term probably means different things to different people but at least to the boss at NHGRI it means that personalized medicine refers to using information about a person's genetic makeup to tailor strategies for detection treatment and prevention of disease. And I think most of us who think about personalized medicine think about the human genome and the use of genetic markers. Each person's personalized individual signature that would have impact on the approach to their prevention, diagnosis and treatment of disease on again a very individualized basis. One of the things that has been pointed out is that family history starts out is perhaps the first step of a personalized medicine program. And unfortunately we're not so good at taking family histories. A book on Terry and my shelf is a physical diagnosis by DeGowan and DeGowan. And if you look in that book there's nothing on taking a family history at all. It just isn't in the book on how to take a history. And so maybe we shouldn't be surprised. The other thing is that once a pro band occurs in the midst it doesn't look like we're very good on acting about it at all. And this is a study I did with a medical student a while back. We had 5,620 consecutive patients admitted to 53 randomly selected hospitals around the country. Everybody who had standard criteria for coronary disease on entry. And when we reviewed these 5,620 discharge plans only 37 of them 0.7% identified a plan to screen the first degree relatives. And it actually didn't matter if the person was 35 with their heart attack or 75 with their heart attack there wasn't part of anybody's plans. And we followed these folks up six months after discharge and only about one out of six children had ever been screened regardless of the risk of risk factors in the pro band. So as we look at screening for genetic factors I might say well why don't we do the things like asking has your mother father sister brother had heart disease first before we do all the heart disease genes for which we have no evidence that we're actually doing. Okay because really aren't getting it there. So that the family history is part of this. So you know it's like you know it's right there. What are the what is the question? This is another rendition of the slide already shown by Terry. You know this is a person at genetic risk. It's not a you know a cow it's what is the question. So we should do that. Now one of the efforts and Alan Gutmacher of the ACRI is one of the one of the leaders of this. It's a multi-agency initiative in the Health and Human Services. It's the my the US Surgeon General's Family History Initiative and one of them one of the parts of this is my family health portrait. It's a web-based tool to collect and organize family history information. I've given you the website. You can actually get a printout to share with your healthcare providers. So from the previous slide kind of saying okay if you're not going to collect it I'll collect it and maybe you could do something about it. And of course Thanksgiving Day here in late November has been for the last several years the National Family History Day. So while you're sitting around carving up your turkey drawing a genotype, a pedigree on your wall or something encourage Americans to talk about right down health problems that run in their family. So the point is that I think this is really the first step and we're going to talk about some of the evidence there is is the US population or the US healthcare population that healthcare providers are really ready for some of the genetic and genomic information that is currently available commercially. And I think most of us would agree and there's been multiple editorials from multiple sources is that we're probably not. And there's a whole group at NHGRI that work on the many issues here. Well the other point of it is to say that genetic screening is something new is actually not really correct because of course there has been newborn screening for some years this is not necessarily DNA screening some of this probably should be followed up with DNA screening but we have been screening for genetic diseases for some time by far the most common is congenital hearing loss but you can see a variety of others and so we have been screening and putting into interaction the genetic counseling and the follow-up of these children for a long time. So there really is a precedent and something to build on and obviously some of these experiences we need to take forward. So an effective screening program obviously needs to have analytic validity that is, is that we know what we're measuring and there's good reproducibility, clinical validity in terms of it measuring what we want to measure. Obviously these are a variety of issues with some of our genomic ones but I really want to talk about clinical utility as we roll things out some of the genomic quality issues Terry's already covered up here. The condition should be frequent enough to justify the cost of screening the detection should be otherwise not, the detection would not otherwise occur at an early enough stage to perhaps prevent disease. Early treatment prevents morbidity, treatments available and family and personnel are available to perform the screening and form about results and institute the treatment. So the whole point is that we, in terms of the applications of genetic markers to clinical practice and this is from Greg Furrow from the NHGRI and one of the other members in the office of the director in this editorial recently in JAMA obviously identifies four barriers to really carrying this out. The first is really lack of information about how the prevalence and risk contribution of markers varies across populations. We have a lot of data in Europeans, perhaps North Americans, et cetera but other population groups obviously may have very different risk and certainly we have plenty of examples where the prevalence of a gene or its impact on disease varies tremendously from one population to the next. We have limited data on how the inheritance of multiple markers affects an individual's risk. We just talked about that, this whole gene-gene interaction, et cetera and so this whole risk assessment quality. There's little information about how most genetic risk factors interact with environmental factors, again the gene environment issue and then finally and maybe most desperate in terms of our research needs are that fewer studies, few studies in common diseases have tested the effectiveness of interventions on genetic risk factors really, again, identifying those markers, identifying the target targets for drugs and going ahead and testing them. One of the issues and we've had a lot of discussions and the Northwestern faculty have, Dr. Greenland and Lloyd Jones and others have commented on some of the biomarker issues as very appropriate to do but one of the points is you can look at a genome marker as just a new biomarker test and so you'll have patients at risk for disease, you can take people at general risk or perhaps people at high risk with family histories, et cetera but one of the real keys at the end of the day is to show that this is cost effective and the number of studies on any biomarker, I don't care what it is, being an imaging test or a serum marker or a gene marker, there are very few studies in which they've used randomized designs in which the new test then would reclassify individuals in low and high risk and treat them accordingly and look for the outcome and compare whether or not with usual care in which this marker wasn't used they just went ahead with treatment and no treatment and see what happens. If you look at the literature there's very few of those and one of the reasons there's very few of those is it from a commercial standpoint, once you develop a marker and are able to sell it, you have very little interest in this kind of a study because all the study could do is actually show low cost effectiveness and actually be a commercial disincentive for its use. So I think this is a place for governmental funding and we really need to know this kind of cost effectiveness research and I think the genome markers would fit in these as well. Well, I wish I had a Larsen cartoon about the elephant out of the barn but it certainly is. I've taken off all the commercial names of this but this is from a recent piece in JAMA about screening but you can obviously a variety of direct-to-consumer genetic testings are available. Certainly the whole genome, 23andMe, several companies and these are complex risk screenings based on steps discovered through their ongoing research. You can also get singular multiple trait testing for conditions or specific diseases using proprietary panels. There's obviously some studies available for paternity and family relationships particularly using mitochondrial and y-chromosome panels and there's some others. There's even some that you can get to tell you which diet to use and so many of these are available and so what's going to be increasingly happening is that if you're a practitioner someone's going to come with a printout of the polymorphisms that you have that your patient has and you're going to be basically asked so what do I do now? And obviously this is going to not be because we're already quite obvious that many of the simple family screenings taking a blood pressure in the child of a hypertensive patient not being done, wait until you get to this level of complexity and so there's been a variety of issues about this. So here we are, this is us. We're almost free, we've all figured out the genomics, we're about to... I think it's only about to get more complicated and I think there's going to be a flood. Now having said that there obviously are some examples of genetic markers which have been I think considered and used. These are the lifetime risks for cancer of the breast of the ovary of these two BRCA markers and you can see there are substantial lifetime risks of some very bad diseases and so the natural history of this particular marker is well worked out. Unfortunately, and this is from Wiley Burke from the Jackson Lab course, obviously points out that relatively little of the total incidence of breast cancer unfortunately is identified by the BRCA one and two testing and so the fallback position would be really to test the affected relatives with breast or ovarian cancer and then if she were positive, offer to affected relatives. This is not an unconsequential cost and I'm wondering if these are going to be coming down if you can get your whole genome for $1,000. But anyway, and obviously there are some things you could do about it, prophylactic surgery and mastectomy, tamoxifen, which may only be effective in some of the groups and obviously a much more attentive look at breast cancer screening maybe using breast MRI. So there are some examples to lead our way in terms of our approach to this so I don't think it's all just hopelessness. This review by Schweiner really looked at the delivery of genomic medicine for chronic diseases currently in the literature. There were essentially three or four areas, outcomes of genetic services, consumer information needs and barriers to integrations. The key findings were really a modest positive effects on anxiety, improvement, reduction of anxiety with the results of screening. Mixed results on behavior change in terms of people say following up with screening recommendations, etc. Very few studies with clinical outcomes. In terms of the clinical information needs some, I think, very convincing evidence that there are low levels of genetics knowledge but basically positive attitudes toward genetic testing but some great concern of the inadequacy of the primary care workforce to provide the counseling and referral, etc. In terms of barriers to integrations there had been some concern about the oversight of these testing, how these testing was going to be done and the quality of it. So many of our lab certification issues I think can address this. And there's a concern of concern about the use of these data, the privacy and in the case of positive finding discrimination of that person against them. And so this is kind of what's out there in terms of the use of genetic services and identify some of the issues we would have in terms of translating these into care. One of the positive step forwards is the Genetic Information Non-Discrimination Act which is on the president's desk and if you weren't marrying daughters apparently would have been signed already. This was, I think, the primary author on the health side of this was actually my congresswoman, Louise Slaughter from our district in New York. And this really is a necessary building block because one of, as you've seen, one of the literature, the studies results identifies concerns about discrimination if you do have your studies, genetic studies done and this law then fixes the problem at least in the areas of health insurance and employment. And prohibits health insurers from either requesting or requiring genetic information on individuals or their family members or using it for decisions on coverages, rates, et cetera. And this includes individuals who get genetic testing as part of research studies. So for those of us in research, obviously it has this important part as well. And it also prohibits employers from requesting or requiring information or using it in decisions, requiring hiring, firing in terms of employment. Now there's still many other issues involved but certainly I think this has been in the works for 10 years, right, Terry? 12 years. 12 years, yeah. So it's been in, but it like passed 490 to 5 in the house and I think there's only one senator that voted against it and I didn't understand that. So apparently this will be a law and obviously it's a step forward for the implementing and the practical aspects of genetic screening. I want to just say a couple of things about pharma called genetics. This field is the study of the differences in drug response to the lilic variation in genes affecting drug metabolism, efficacy and toxicity. It's an enormous, it's a course in itself. But just to say is that we know there's a number of the aspects of drug metabolism that are under genetic control. Particularly having to do with the systems that break down drugs and the lilic variations then being either the normal metabolizers, the poor metabolizers in which the gene product doesn't break down the drug as much or a fast metabolizers in which you're forced to try to keep up. Maybe you have to use a higher dose to get the same, let's say, blood level or therapeutic level. So there's obviously plenty to suggest that genes are very important here and we know that there's tremendous heterogeneity across populations. These are, if you're into tuberculosis control, the slowest etylator phenotype looking at INH metabolism obviously varies tremendously across populations. So that, for example, INH in this group could be a problem of not breaking down the drugs and in fact non-supervised INH, the classic studies in Baltimore in the 1950s, unsupervised INH turned out to be somewhat of a public health disaster as people with this high prevalence of slow acetylation obviously ended up with liver failure having to do with drug toxicity. You might also say that in other populations that this would not be a problem and maybe in some populations might even be a reason for needing higher doses of this with the Inuits, the Alaskan natives, et cetera, having a higher tuberculosis rate at least in the days before drug resistant TB and AIDS. So the point is there's a lot of variability. These are the five studies in the GWAS collection so far or at least in the first 109 GWAS that dealt with pharmacogenetics just to make the point that GWAS certainly are a very reasonable tool for looking at this issue, again in an agnostic, non-hypothesis driven. Two of them have to do with nicotine comparing this whole issue of nicotine dependency versus non-dependent and how nicotine would really its method of action really. One we refer to already by this biome, I think it's one of the better studies, response versus no response in multiple sclerosis therapy to a beta interferon. This Drex-Ramen inhibitor was really a drug toxicity study but could be a model for many other hepatotoxicity studies. And then this methamphetamine dependence, again another study of addiction versus control. So just the point is, even within the GWAS literature there's a number of genome-wide association studies looking at drug related issues. Really finally, I want to just, in terms of comment relative to research, the genome-wide association studies have now have a policy at NIH in terms of data sharing and I think some alluded to before is that once you do the whole genome there may be the opportunity to study many different phenotypes and this should be a repository of great national importance so that other investigators may look at issues involved. So the goal of this policy was to make available the genotype-phenotype databases as rapidly as possible to a wide range of scientific investigators. It includes depositing the data at the National Center for Biotechnology Information and the DV GAP registry that we talked about. Obviously, plans of the data submission and protection and obviously the lack of identification of information, personal information, how to access the data, the publication of results and there's some statements on intellectual policy within that policy as well and I think the driver of this really has been this open access which has really been a tremendously consistent theme at the National Human Genome Research Institute. It really has been and it goes to many others than just the GWAS but this high kind of opening up the idea that the human genome really isn't a commercial product. Let me tell you a story about this open access issue. I was at the Bar Harbor course on human genetics and one of the things you could do during that time is see patients. You had a patient day and I'm a cardiologist and so a couple other cardiology professors and I saw a patient and she was a tall, slim person with long arms, etc. and her doctor obviously was interested. There was an annual quote-unquote history in a parent and obviously the question was did this woman have Marfan syndrome? In fact, there are criteria, major and minor criteria for Marfan. She had one of the major criteria. You need two and she had about like five of the other criteria just a little bit below so she didn't quite make all the classic criteria for Marfans and so the decision was why don't we just go do her genotype? Well, the genotype was $2,500 and we actually calculated the time from when the patent and the genotype would expire and how much it would cost us to echocardiogram her every year and treat it with beta blockers was far cheaper to do it that way than to get this gene test because of the commercialization of this and obviously what we'd like to do is to say that the technologies and the tests should be patented and this is intellectual property but the gene itself should not be something that one could control its use and it's available to all of us in medicine and public health for the use of the public's good and I think all of us were just a little peeved by this whole thing that this is really a barrier to using science as we'd like to so that's just a little personal vignette but and the GY's policy and open access and these data are available is I think part of that philosophy and so for you investigators who might want to request and receive these data you need to submit a subscription, a description of proposed research project obviously this is a scientific research issue this is not for commercial use or other uses you submit a data access request as co-signed by your institutional official obviously there's some commitments to protect data confidentiality theoretically the genome is the ultimate personal identifier isn't it? There's really only one of them and so theoretically this is could be abusable but there are many provisions in here to protect this confidentiality and to ensure data security measures if there are any problems here to notify a data access committee about policy violations and then to provide annual reports so I think for our graduate students etc you know we may not have the exact genome wide association study in our own institution but with this policy provides opportunities for many many investigators both trainees and established investigators to be more active so I just want to conclude this evidence for the causal associations I think is still informative stages the whole gene polymorphisms obviously have a little bit different wrinkles to them but still I think fit well within the usual epidemiologic constructs for causality the application of products of genomics includes susceptibility assessment and pharmacogenomics obviously I think these are some of the low hanging fruit but still have some barriers as we describe unfortunately technologies are currently be marketed to consumers so that this is going to create a lot of tension particularly I think with the practice community and this is particularly so since evidence suggests a low level of genetic knowledge in consumers and low levels of skills to deal with that knowledge and providers so I think genome research really has been particularly benefitting the basic scientists I think the clinical investigators and epidemiologists I might say added to that should really ready themselves to participate in this developing field and I hope this course has helped to do that so any questions from mine and then we'll just close it up yes microphone we have some of our HEPA participants are involved in ongoing GWA gene environment study right now it's my understanding that the PI of the GWA study itself can specify what kinds of questions can be asked of the data and so for example in our situation our consent form for the DNA samples in the first place was very specific about what kinds of analyses can be done I had only so much room on my slide but thank you for the question and maybe Terry would want to comment on this and she's been involved with all that but the scenario you point out is in fact identified within the whole policy if you read it some of the informed consent set specifications about how the data could be not used and that's what the person signed in terms of consent and if some of those really basically suggested no other use or other use outside of this own study I think that's the answer unless you want to go back to them and re-consent them which obviously is a lot of work as we all know and so there is the ability if there is a limitation on the evidence an opportunity to cite what those limitations are and the reasons for those limitations but I think you can't just limit the whole thing so you just can't say I don't want to play or I don't want somebody to look at my favorite phenotype or whatever it's based on the consent but we don't look at the consent form so there is an element of trust here and obviously if we saw something that looked a little bit suspicious we'd ask someone the limitations that we tend to see are used for a specific disease so it tends to be if this was a schizophrenia study you only use it for schizophrenia and the other one is often non-commercial use so if you're a for-profit user there are some usually about 10% 5% of participants have issues with commercial use and drug companies and that and so those will be excluded then if you're a commercial user but you still can get the rest of the dataset this is potentially obviously an important paradigm in translation and certainly in our clinical translation of science awards I think the idea was consistent with this is to open up to a much larger community of investigators because a lot of people again would go in have their disease of interest ask their question and then say I'm done with it leaving 99.9% of the data really unanalyzed