 So, for those of you who have not heard of the Corel Personalized Medicine Collaborative, this is a longitudinal research study that started in 2007. And really the aim of the study is to understand the utility of receiving personal genome information. We are genotyping individuals, providing it back to, results back to participants, and doing a series of follow-up behavioral studies to determine what they're doing with that information. So are they changing their diet or lifestyle based on that information? Are they sharing the information with their health care provider, and is their health care provider making any recommendations, ordering any tests, changing their care or management in any way? In addition, we're looking to identify new genetic sites that are associated with common medical conditions and drug response. As I said, the study was launched in 2007, and at this point we have more than 6,000 participants that have enrolled. Just briefly how the study works, participants, the majority of our participants have been enrolled through group informed consent sessions. Individuals actually come to the Corel Institute or other group sessions that we've held. At the end of an informed consent presentation, they, if they choose, sign the consent form and spit into a tube, which is always nice in a group setting. Everybody loves that. After they have provided their sample, they then receive an online activation, and the remainder of the study is really conducted via the Internet. People create a personalized account, they fill out a series of required questionnaires, including family history, medical history, lifestyle, demographics. And once they have completed those required questionnaires, they're put in the queue for genotyping. We do our genotyping using the Affymetric 6.0 and DMAT chips currently. And then once results are generated based on those genotyping platforms, we are ready to release results. But here's the caveat. An important component of the consent process is that participants consent to only receiving results that are approved by our advisory group called the Inform Cohort Oversight Board. And I'll go into greater detail. So this is sort of a vetting process for how we will determine what results to report to participants. So that's going to be the crux of the talk, and let me just finish sort of the general outline of the study, and then I'll go into greater detail about that. So once that group approves, what results will be reported, reports are prepared, and they are pushed out to participants through this secure web portal. Participants get an email that says a new result is ready, but none of their results are communicated via email. They're always pushed to go to log on to the secure web portal to view their personal genomic information. After they view that information, they can contact us to speak with a genetic counselor. We also have a pharmacist coach network so that if they have questions about their pharmacogenomic results, they can speak with a pharmacist as well. And we follow up with a series of online questionnaires, again, asking them what they did with the information. Was it useful? Did it prompt any anxiety? Did they actually share the information with their health care provider because the information is going directly to a participant? So I just want to talk a little bit about, because the focus of this session is selecting variants for the return, I want to talk a little bit about how we go through that process. We, the Corial staff actually selects what variants we want to put forward to our advisory group. And there are a series of technical guidelines which I'll discuss that guide that selection process. We then prepare a report that goes to two different advisory groups, our ICOB and Form Cohort Oversight Board and our Pharmacogenomics Advisory Group. Only those drug gene pairs that would relate to drug metabolism will go to the Pharmacogenomics Advisory Group. Everything does go to the ICOB. The ICOB reviews and approves or disapproves, rejects what we've put forward to them. Obviously only those things that are approved go, moves forward to report development and then the deployment of the report to participants. So we have two sets of technical guidelines, one for health conditions and one for drug gene pairs. And it's a fairly in depth process. This was published by one of our colleagues, Cappy Stack, earlier this year. So we go through a literature search for all of the, looking at published GWAS. We will only select studies that, variants that have been replicated across multiple studies, moderately sized studies. And we do have minimum cutoffs for the number of participants that need to be included. And we're looking for variants that have been associated with common complex diseases, not traits. So we had actually originally considered putting forward things, variants that have been associated with things like height or eye color. And have since rejected that idea, those types of variants are not included in the study. Once we've identified disease associated variants, we go through a process of selecting a single variant per health condition currently. And looking at variant selection hierarchy, we rate each variant that we've identified as potentially worthy of inclusion, based on whether or not it's been looked at through a meta-analysis in multiple studies, replication in multiple independent studies, or replication in multiple cohorts in a single study. We take all of that data, compile it into a nice report, and summarize it for the informed cohort oversight board. And only, obviously, only those conditions with valid genetic associations are passed through that process. From the standpoint of pharmacogenomics, this is a slightly different process. We're looking to identify and select drugs and key genes that are associated. We do this by reviewing published literature as well as public databases, including the FDA table of drugs and biomarkers, PubMed, PharmGKB, the CIP 450 drug interaction table, national prescription data, and also our own population and the reported medication use. Obviously, we're not as interested in reporting on a drug gene pair where there's one person in our entire cohort that's on that particular medication. This wouldn't be of interest to the population. And as a result, setting utility wouldn't be very interesting. We also review published and public data drug metabolism pathways, PubMed, again, PharmGKB, and the CIP allele nomenclature database. Based on all that, we try to define key alleles and haplotypes and identify a minimum set of defining variants for those haplotypes, and then select haplotypes for inclusion based on evidence scoring. So the evidence scoring is a detailed process. We rate different studies based on the greatest to, obviously, the lowest evidence, based on whether or not there have been clinical outcome studies conducted, the available pharmacokinetic and pharmacodynamic studies, molecular and cellular functional studies, and genetic variant screening studies. So this goes into a complex scoring process that goes from 1 to 13. Variants that have been only seen in one study, such as a case report, will not be included. If there is functional data that suggests that the particular variant or haplotype or gene directly impacts the metabolism of the drug of interest that is rated more highly than if the gene is associated with a drug that is related, but not the one that we're particularly interested in. Obviously, this is a complex process that I won't go into in detail, but I'm happy to talk with anyone afterwards about the strength of evidence scoring system. So those are the technical guidelines that guide the selection of variants and drugs and health conditions that we put forward to the ICB and the PAG. But just want to briefly talk about that process itself. So we have the informed cohort oversight board. This is an external advisory board. There are no voting members from COREAL. It's composed of scientists, medical professionals, ethicists, and community members. This is modeled after CAHANA's ICOB concept that was published in Science in 2007, where an expert group would review drug gene pairs, and that would be used to determine what information could be reported back to participants. So COREAL's informed cohort oversight board consists of many notable scientists, some of whom are in the room. And actually, this group just met this past Monday. Their charge specifically is to determine whether each health condition or gene involved in drug metabolism is at minimum potentially actionable, and I know David just said he hates that term. It's hard to say what the cutoff should be. How do you decide what gets included? But because we want to look at the potential utility of this information, we felt like some level of actionability was needed, so we have provided the ICOB with some guidance with respect to that. In addition, they're also charged with determining whether or not the genetic associations that we're proposing are statistically valid. The Pharmacogenomics Advisory Group is a separate board, and they are specifically chosen based on their expertise in pharmacogenomics. They provide recommendations to the ICOB, so their decision alone does not determine what goes out to participants, and this group is composed of pharmacists, pharmacologists, ethicists, and clinicians. Again, we have members of this group in the room and certainly appreciate their service. Their charge is similar, but a little bit different, so it's to determine whether or not there's sufficient evidence to support the role of each gene in the metabolism of the proposed drug, whether the impact of one or more haplotypes is clinically relevant with respect to the proposed drug, and whether the drug pair is potentially actionable. Again, the workflow of the ICOB and the Pag is that although both review these submissions, the decisions of the Pag are then submitted to the ICOB for approval, and there are cases where the ICOB does overturn a recommendation or rejects a recommendation of the Pag. Just a list of the currently approved diseases and drug metabolism pairs that have been approved by the ICOB so far, and acknowledgments of the staff and ICOB and Pag that participate in this project. Thank you. Time for a couple of questions or comments now, and then the session is interrupted by lunch, but we have three speakers after lunch and then a longer discussion session at the end of the session. Howard? Just a quick point of clarification. You had indicated that a single variant would be chosen per condition. Do you mean that at a site where there are lots of different variants that are in the same haplotype block, you're only selecting one SNP? Or do you mean for a common complex disease, you will choose one and only one variant? So for example, in the case of diabetes, this gives me some concern given the plethora of potentially relevant variants that exist. So thank you. It's an important clarification. So for common complex diseases, we are currently only reporting on a single variant, and it certainly is concerning given the multigenic models that are coming out and the number of different variants that are known to contribute to the risk of these common complex diseases. At this point, we are working on coming up with a multigenic algorithm that would be approval by the ICOB, not at this past ICOB meeting, but at the previous one, we had put forward some multigenic models, and the problem that we ran into is that they're not replicated. And so although we all agree that there are multiple validated and replicated variants associated with each of these common diseases, how do you put forward, are we going to give people 50 different variants and make them process that? We'd prefer to give people a multigenic risk, but none of those models have been validated. Do you choose a model that has multiple validated variants within it? And so we're still trying to think through that process. In the meantime, we certainly see it as a limitation. We make it clear on our reports that it's a limitation. We're only providing you a very small insight into the many genetic risk factors that contribute to this, but at this point, we feel like it's one approach. I will say that for pharmacogenomics, we are using multiple variants and reporting on using the star nomenclature that's commonly accepted. Can you comment on, so this is great work, and it does seem like it takes a lot of effort. Can you comment on the amount of effort it takes and how this can scale? So for example, at different institutions that are tackling these same issues, typically you can get together a committee, but then it ends up, well, maybe we discuss one whatever is hot in that particular month. How can the scale, and for example, do you make your knowledge available to others to be able to use at other institutions? And if not, how do you sort of see this scaling beyond this particular initiative? So that's a great question, and is one that is discussed routinely by our advisory groups. The amount of preparation work that goes into each of the ICOB impact meetings is absolutely enormous on the part of the staff at Corel, because we really feel like as an advisory group, they shouldn't need to go out and do all of the legwork to try and determine this, but we should present it to them and they should be evaluating it with additional supplemental material if needed. Currently, we give them on average, I would say, four either health conditions or drug gene pairs to evaluate at each meeting. We would need an army of staff in order to scale this up, and so one possibility would be, as ideas like this pick up momentum at different institutions as having different groups of people working and sharing their knowledge across different institutions, I think we are happy to share our model. We have always said that we would make the decisions of the PAG transparent and we can certainly discuss sharing the submission materials with other institutions so that there is some collaboration and everyone isn't reinventing the wheel. I think we are going to have to move on to our next speaker. Thank you.