 Well, I too am honored to have an opportunity to speak to you today. And I think certainly the notion that cancer is a disease of the genome has been around for a long time. In fact, it was first proposed by, it was, you know what, I'm going to use a keyboard. It was first proposed by Theodore Bovery, who was studying cells under the microscope and various stains that were available at that time basically a century ago, and proposed that defects in chromosomes would lead to abnormal cell proliferation and that could underlie cancer. So this idea, although we're capitalizing on it now, this idea was proposed a century ago, but it got waylaid for good reason. And there were some competing theories out there, one of which was that cancer maybe really was caused by viruses and not by an abnormal genome or abnormal chromosomes. And in fact, you could argue that Varmas and Bishop, when they sort of set out to study why Rouss Sarcoma virus caused cancer, were expecting to understand what about the virus caused cancer. So the observation that, in fact, the cancer-causing agent Sark was actually present in the normal genome. It was a perturbed version of the normal genome, really gave new life to the idea that it's really about the genome having gone wrong in cells that cancer arose. And even after that observation, there was a flurry of discoveries that really by the early 1980s gave us the basic language for thinking about somatic genomic alterations in cancer. And it was known by basically 1982 that there are three major categories of somatic alterations. Base mutations, this is codon 12 of RAS shown here, probably the most mutated codon in all of cancer. There are chromosomal copy number alterations, either amplifications or copy gains that give you too much genetic material, or deletions, copy losses that inappropriately take away material that would keep the breaks on cancer. So this is another important category, and there are well-known oncogenes, MIC and ERB2, for example, that are amplified, and then, of course, tumor suppressors, RB, etc. deleted. And then finally, there are translocations or DNA rearrangements, and BCR-ABLE was the classic one, but now we recognize many of these. So the language, the genomic language, although we didn't use the term genomics at the time, for speaking about cancer genetic alterations was all in place by the early 80s. But it took multiple sort of revolutions, both technological and experimental, to really give, to breathe fire into the equations, as it were, of really unlock how important genomic derangement and how extensive genomic derangement was in cancer. So what I'd like to do is give a couple of vignettes about the insights that have come now from the human genome era, and that are continuing to accumulate in the present day with the explosion of genome information. And I'd like to give some examples of what we learned biologically, and then I'll move to the implications for precision medicine, which you've already heard some introductions to. And, you know, this is one of these areas, particularly in the zero-sum game world that we now live in, of research funding, where there has been an active conversation, some would even say an argument, about the relative benefits of large-scale, very expensive government-funded projects that arguably take steer money away from traditional R01-funded research. And I'm not going to sort of try to argue exclusively for one or the other, but I think certainly it's been the case that these large-scale projects have taken us in directions that are crucially important for cancer biology and therapy that many of us were not thinking about at all before these projects started. So let me just give a few high-level examples. So one of the early fundamental insights that came out of sequencing an entire exome in cancer was emerged from a study led by the group at Johns Hopkins, and actually this was a sequencing study of glioblastoma that was published around the same time as the TCGA study, and they sequenced a very small number of cancers, but completely, meaning all of the protein-coding gene exons at the time, which was a huge feat, and they discovered that there is a gene called isocytrate dehydrogenase that was mutated very commonly in glioblastoma, brain cancers. Now, isocytrate dehydrogenase is a member of the Krebs cycle, the TCGA cycle. So any of you guys who took college biochemistry, which I'm sure is a large fraction of the audience, would have had to memorize this and then take the test and forget six weeks later, and it was a very forgettable enzyme, but lo and behold, it's mutated in a large fraction of brain cancers and AML, and it'll lower frequency in some other cancer. But what was even more remarkable was that the mutations, so initially one thought, OK, well maybe this is disrupting the production of alpha-ketoglutarate, but it turns out the mutation was a, it was neomorphic. It gave a new function to this enzyme. So it caused alpha-ketoglutarate to become a, what we call an oncometabolite, two-hydroxyglutarate. And among the things that two-hydroxyglutarate does is essentially interfere with a whole series of enzymes that are normally controlled by alpha-ketoglutarate. For example, histone-methyl transferases, the tetenzymes, histone-demethylases, agolin enzymes which are involved in responsive cell to hypoxia. So this spectrum of alterations has an entire series of effects, one of which is major dysregulation of DNA methylation in genomes. And it has given rise to an entire field, and in fact there are now therapeutics being developed against mutated IDH1 and IDH2. So this is a whole field that now exists all traced to this one discovery from a bold project sequencing 11 or 12 or so glioblastomas, which was thought to be far too few to learn anything. It probably was, but lo and behold, this popped out. And it has now forged a durable link between the notion of genomic alterations and deranged cancer metabolism. So the Warburg effect and the importance of cancer metabolism has really been galvanized in many ways by this discovery. Now, another fundamental insight has been really the genome telling us that we need to look beyond the genome. So I guess I don't know the art world very well, I'm sure there's an impressionist analogy that I could use for that. But what we can see here is a whole series of enzymes, for example, that modify the histone tail and then regulate chromatin compaction. This is the switch-sniff complex. And what you can see is that these are now decorated with genes that are recurrently mutated across many, many cancers. So the derangements of chromatin and epigenetic modifiers and the extent to which this is relevant and selected for in cancers, many, many cancers now, is probably one of the most groundbreaking discoveries in aggregate of cancer genome sequencing. And now what it also has done is, so this is a hugely important discovery in aggregate, set of discoveries, but it's also reminded us of how ignorant we are because in most cases we don't know what the chromatin targets or target genes that are dysregulated in concert by these various mechanisms, but clearly it's important and we have to get a handle on it. In some cases there are already therapeutics that are being developed and even in the clinic in some cases to target some of these enzymes. A third enzyme, a third insight, which really I would say took the field by a complete surprise, and this is only two years old or so, was the importance of mutations in mRNA splicing. So mRNA splicing, I mean of course we all know that it's important to regulate gene expression, post-transit, et cetera, but I don't think anybody had realized how important somatic mutations in splicing would become. And in fact in malignancies like myelodysplasia and chronic lymphocytic leukemia and even in some solid tumors like lung cancer and breast cancer, several of these splicing factors are mutated recurrently. And in fact some of these mutations appear to be gain-a-function events. And it turns out that there's even a drug in, that has been tested in clinical trials that is known to affect the splices on the SF3B complex. Now it was never of course deployed against patients who had splicing mutations in the tumors, but now of course the idea of targeting the splicing machinery has new life when in fact I would argue there was hardly any life for that at all before this set of discoveries. And then the final one I'll just mention right now is that in squamous malignancies, so head and neck squamous, carcinoma, lung squamous, carcinoma, cervical squamous, carcinoma, it turns out that there are a whole series of mutations often in 30 to 40% of cases you can argue there's at least one of these that disrupt squamous differentiation. So the concept of maturation arrest is actually a very old one in cancer. In fact most of the translocations that were discovered in leukemia by keratopic analysis affect core binding factors or transcription factors that arrest the maturation of that particular hematopoietic lineage. But now the idea of maturation arrest is gaining traction in the solid tumor arena and the squamous malignancies are one area in large part because of cancer genome sequencing has really put this on the map in a way that I think was not appreciated before. So these are four cross cutting categorical insights that have really awakened whole subsets of field of biomedical research and now I think you can argue center a lot of hypothesis driven work that can be focused in these areas. Now I'm going to mention what I think could be another important insight and I'm just going to shamelessly point out that this coming from our group and it's being published I think just today in cell. But I want to describe one more insight that we would not have known about had we not done whole, not even whole exome but whole genome sequencing in a subset of cancer. So this is a figure from an older paper that we published a couple years ago showing a complex rearrangement pattern that we discovered in prostitutes. I'm not going to walk you through and you can already look at this and say well it's a pretty figure but it's kind of complicated. That's basically where we were but there is actually a method to the complexity. And when we sort of step back and thought about what the rearrangements look like and what they were telling us it actually gave us a view into what could be a very important cellular process that happens and then goes wrong in cancer. So let me walk you through what we think is happening and how we think it's happening. So what we initially call them closed chains of rearrangements and they're in a subset of prostate cancers that have recurrent rearrangements in the ex transcription factor family. And so what we think is happening is that, so let me just back up and point out to you that transcription happens in many cells that appears to happen in a physically localized region of the nucleus. So they've been named transcriptional hubs. So this is a little bit different. I used to, when I was in college anyway, I learned about transcription as thinking about the DNA as a rod and there were cis-acting elements and transcription factors would come sit down on that rod and sort of send transcription to the right of the screen but in fact, obviously DNA is not a rod. It's floppy and exists as a fractal in the genome and in fact, different disparate reasons of the genome can migrate to transcription hubs. And what we think happens is that these come from different chromosomes or different reasons of a chromosome and for some reason that we don't understand errors happen, breaks in the DNA happen, but they get erroneously repaired. So rather than being repaired back to their partners, they can sort of propagate themselves and be misrepared. And if that happens, you get, well this is why we call it sort of closed chains because you can see that there's a series of breaks but they kind of come back to the beginning. So A goes to B, B goes to C, C goes to D and D goes back to A. So it's a closed chain. Chain is not circular, it's sort of a propagation event of errors that kind of comes back to where it started after those errors are made. So this is basically what we were seeing in some of the prostate cancers. Now, this is all well and good, it's all fascinating and it sort of gave us a perspective that yes, transcription is dynamic and this could be one, you know, transcriptional dysregulation could be one way that rearrangements form but we hadn't realized how important this insight could be until one of the MBPHG students in the lab, Sylvan Baca, realized that these breaks that were happening could be more complicated. So what can happen, so what we initially had seen was that we sort of said these are single, these are clear breaks and there's no deletion or only micro deletion and they get repaired but in fact, what often can happen is that deletions can begin. So there can be deletion bridges that occur after the breaks happen. So when that happens, various reasons of the DNA, so not only is there a rejoining but these reasons that are deleted are now gone and so then the errors happen and so now what you can do is say, we're not only gonna look for change, we're also gonna look for DNA that's missing at the break points. And this sort of is rather subtle and complicated but this recognition that there can be deletion bridges and not just clean breaks and rejoining led us to, I can't take any credit for this, but Sylvan who turned out, although he didn't tell me this when he came to the lab, turned out as a genius and he said, oh, this is just a graph theory problem. All we have to do is take the break points of the nodes and the deletion bridges are the edges and we're just gonna come up with an algorithm that applies graph theory, finds these events and actually tells us how often we see these change and the bottom line when he did this and he applied this now we had a much larger set of prostate cancer genomes thanks to a center initiated project funded by the NHGRI so we had about 57 whole genomes by that time and what he found is that these events are incredibly common. So 84% of primary prostate cancers have at least one chain so actually we decided by this time that we needed to give it a new name, we thought the Sanger Center can't have all the fun naming, the Chromo event so we gave this a name Chromoplexy and so now these are circus plots where the different color codes are different chains that you can detect within an individual genome. So 84% of primary prostate cancers have at least one chain and 2 thirds have at least two chains and so this is prevalent enough that you can imagine that maybe the ones that don't have one are simply because we didn't sequence deeply enough of course there's a lot of stromal lab mixture so these may be incredibly common and some of these chains exhibit subclonality though these are each important points in and of themselves but I'm gonna tell you why we think they're even more important when you put them all together. Now one of the reasons why they're important is because known cantigens are often they often are disrupted genetically in the context of Chromoplexy so this is a nice pictorial representation of a well-known tumor suppressor the P10 locus where you can see there's all kinds of disruption occurring in a region this happens to be a Chromoplexy event and P10 is involved but actually if you go and tabulate cantigens you find that there are a whole series of them that are recurrently involved in Chromoplexy including very importantly these ets fusion rearrangements so ets either ERG or ETV-1 these ets factors are the subject of rearrangement in at least 40 to 50% of prostate cancers but the majority of them occurred in the context of Chromoplexy which means that Chromoplexy the initiating event for Chromoplexy happens upstream of the ets rearrangements and there are a whole bunch of other cancer genes that had this phenomenon the other thing was that we looked across other lineage for which we had whole genome data and we found evidence of Chromoplexy in several additional cancer types head and neck cancer, breast lung cancers and melanoma so we think although it's incredibly prevalent in prostate cancer it may well be have some prevalence and importance in other lineage as well so this is why we are maybe precociously but we are labeling this fundamental inside number five because we think it's telling us something about tumor evolution that we maybe hadn't fully appreciated so here on the left is kind of your classical way of depicting tumor evolution which is that every cell division you get errors in the DNA and they accumulate and over time you get more and more mutations at some point you cross the threshold where you get just the wrong collection of mutations and you're a cancer so this is the classical Darwinist view of evolution now a few years ago Phil Stevens and the Sanger Institute taught us that there can be an opposite extreme where you get a catastrophic event that has massive pulverizations of focal reasons of chromosome this was called chromothripsis and so catastrophe can sort of be the opposite end of the spectrum whereas gradualism is on one end catastrophe is on the other what we think that chromoplexy is doing is sort of it's in the middle because these are not as catastrophic but they're clearly more impactful than single nucleotide substitution so we think that this is sort of a a question to punctuated equilibrium from a la Steven J. Gould so this is punctuated evolution and there's evidence for this because there's clearly punctuation within prostate cancers because you can see multiple chain events you can see examples of subclonality and importantly as I mentioned cancer genes tend to be dysregulated by these events so we think that this is an important modality of evolution that had not been sort of appreciated before and the other I'll just sort of throw out a speculation that if we could understand why these events happen so what is the error that happens when these transcriptional hubs or it may not always be transcriptional maybe it's chromatin what is that could we reduce that a little bit if we had a way to reduce that error rate even just lowering the slope a little bit perhaps that would be a chemo prevention mechanism maybe fewer prostate cancers would occur I realize it's total speculation but it's certainly something we can speculate about because we have whole cancer genome data for the first time another area that caught us our group largely by surprise was an insight that recently we had into the dark matter of the cancer genome so we genomicists call dark matter regions of the genome that we cannot easily interpret these are as you heard from Ewan for example regulatory regions, intergenic regions some of the repeat-rich DNA and in cancer we often think about the aneuploidy that's not focal we don't really understand always what it's actually doing is it multi-factorial or is it just noise or what have you so we asked a question we were in a study of melanoma we had a certain critical mass of melanoma genomes and we asked the question are there any nucleotide substitution somatic mutations that occur at the same nucleotide more than once which is actually if you think about it a statistically unlikely event unless there's particularly in a non-genic or regulatory region and what we saw when we looked at the initial set of melanomas that we had enough coverage to assess was that 17 of 19 had one of two mutations within just upstream of the transcriptional start site of the telomerase promoter now this is melanoma it was actually tantalizing because these were both CDT transitions which occur in the context of UV damage and they were mutually exclusive now this is one of these things where like Ewan described in his talk we saw this actually we published sort of an initial paper on these genomes and we had seen this event but we didn't mention it because I was skeptical I said this can't possibly rewrite there's no way there is a somatic mutation that's happening at the same nucleotide at this frequency that's got to be about so we're not reporting this because if it's wrong I'm gonna look stupid and I don't wanna look stupid so we sat on it and actually it turned out to be a bit of a pain to validate because the region was kind of GC rich and you know, sequinome didn't work and so finally we resorted to old fashioned Sanger sequencing and that worked and it turned out that in a actually relatively small validation set the numbers held up one or the other of these mutations was present in 70% of melanomas and what was even more tantalizing was that when you looked at the DNA context around these mutations they both as a result of the mutation you ended up with an 11 nucleotide stretch but in the center was a consensus X factor binding site and X transcription factor binding site so this obviously suggested that these are biologically consequential and in particular they might bring an X factor and turn on transcription from the Tert promoter so the first experiment to do to test this idea worked nicely so these were cloned these mutations were cloned upstream within the context of the Tert promoter as you know Luciferace and you can see in a whole series of cell lines that each mutation was able with a range of two to four fold or thereabouts to increase expression from the Tert promoter so the mutations appeared to be functionally consequential we don't know for sure that it's an X factor that's binding but certainly they made sense and then we through another project that was funded by Novartis we had whole genome sequencing data from about 150 cell lines and so we looked for one or the other variant across that collection and indeed they started popping up three out of three bladder cancer lines a thyroid cancer line again in melanoma liver four out of five hepaticellular lines also you can see there are three out of four CNS lines and this is another example of the PI in the lab being too conservative we didn't want to we didn't want to make such a big deal about this because the numbers were small but now a group from North Carolina in collaboration with Bert Vogelsen has looked at this locus across about 1200 tumors and indeed these exact tumor types pop up as highly recurrently having one or the other of these mutations so it really means that this non-coding these mutations together are among the most commonly somatic mutated nucleotides now in the genome so it's a remarkable thing tells us that the dark matter may be hiding some surprises many other surprises that we didn't know about so in the last few minutes of the talk I want to move from what I would argue or have been very gratifying biological insights into the potential importance of the cancer genome in precision medicine and there are three underlying principles that I just want to drive home about the genetics and precision medicine the first one is that molecular pathways involved in tumor survival and progression are often activated by genetic alterations now one of the disappointments if you will of sequencing cancer genomes was that when this all started you know there was the discovery of BRAF and 50% of melanomas and the discovery of PIFU kind of mutations in colon cancer and EGFR and lung cancer and all these kinases would be mutated and kinases are druggable so the promise was oh we just have to sequence more cancers and all these kinases that are druggable are going to pop out and pretty soon cancer is going to be a chronic disease if you can treat with them kinase inhibitors but in fact it turned out that there aren't a whole lot of kinase mutations that we found when we started going more deeply but nonetheless if you step back and you start to catalog several of the major common solid tumors you know lung cancers breast cancer colorectal cancer melanoma head and neck et cetera and then you do bar graph form for glioblastoma and ovarian and this is using TCGA data and some other data and if you just annotate those collections of cancers for the presence of mutations that are plausibly actionable now by actionable we don't necessarily mean there's an FDA approved drug out there but we might well mean that there's a clinical trial or there's something that we should not do there's a drug we should not use now that we know the mutation there if you aggregate and you do kind of a back of the envelope calculation you realize that in several major tumor types 40 to 60 percent however at least one genetic alteration that affects an actionable proliferation or survival mechanism that is a very reasonable number so even though yes we didn't it turns out that that you know there are every major cancer type doesn't have 50 percent of one or another kinase mutated there is a lot that we could do clinically with the information that we already understand and that says nothing about the chromatin alterations and metabolism that we don't yet have a weighted target so the second principle is that we are now living in a special time in history which is that for the first time there are many many anti-cancer agents that target all of the classical oncogenic pathways in clinical trials so here's a graph that sort of shows this schematically these are some of the well-known pathway effectors the MAP kinase pathway the PR3 kinase pathway and some of the others here's some chromatin factors these are receptor receptor tyrosine kinases and if the box or a square has a solid line around it that means there's at least one drug and usually there's more than one drug targeting that in the clinic so this is the first time where not only do we have a knowledge of a large swath of potentially actionable events but we have tools we have experimental compounds or sometimes FDA approved compounds that could hit those altered pathways and finally as you know I'll have to slide out well the third point was that we have the technology but of course you already know that we have master the parallel sequencing and it can be used in the clinic so basically actually let me skip ahead here because I realized that I had made a this is the problem of making last minute changes you forgot we actually put in so here's the third point which is that we now have the ability in the clinical arena to do robust and reasonably comprehensive genomic profiling and here's another area where NHGRI I think was really forward-looking because in early 2011 they put out a request for applications about clinical sequencing exploratory research it's come to be known as CSER and we put a grant out and we were very fortunate with the help also of the NCI to be awarded one of the first grants in cancer and so we now have a project ongoing at Dana-Farber together with the Broad Institute that we call CANSEE obviously for cancer sequencing the project that's NHGRI funded is focusing on lung and colon cancer but we've been able to expand to other cancers as well and it's a very gratifying the enrollment we are barely able to keep up with the patients that are being enrolled to get this and what this study by the way is whole exome sequencing so we are doing whole exome sequencing and we're figuring out how to interpret the entire exome if possible and make that palatable to busy clinicians and there's a whole lot behind that now one of the things that you've heard from previous speakers is the cost of sequencing is falling but the data points that are coming because the costs are falling are skyrocketing so if you look at clinical data points per patient that a doc has to go through so up until about you know six, seven years ago there are you know in the dozens of laboratory values that a clinician together with a physical exam and other points of the history there are a few dozen data points that they have to collect to kind of come up with a differential diagnosis and decide how to treat well you know once we sort of started using kind of the early pass of cancer genetic information which was maybe a few a few dozen genes and a few hundred mutations by genotyping technology already we could see a bump to around maybe a hundred data points that you had to consider but now as we're doing whole exome sequencing and you consider transcriptome and other things we're now skyrocketing so the data points for patient of course are an inverse to Moore's law and this is a challenge now Ellie Van Allen who's a medical oncologist in our group but also has a background in computer science from Stanford turns out to be the perfect blend to attack this problem so he's come up with an algorithm that he calls file for short for precision heuristics for interpreting the alteration landscape which takes the output of exomes and then splays them out on he calls them sort of a virtual gel where at the top are the most clinically actionable variants and then you kind of go down and say well these aren't technically actionable but we should look at the biological relevance because they might be hitting a pathway in some way that could be actionable and then you kind of go down the list but I'm not sure why this is happening but the so file is kind of our first pass at the exome level of trying to get this kind of information now the so file itself you guys those of you like me are Lord of the Rings nerds will recognize that this is this comes from the Lord of the Rings the file of Galadriel was given to Frodo at a really mission critical time where everything looked lost and the quote was may it be a light to you in dark places when all other lights go out and we actually thought this is a great name because these are for metastatic cancer patients they are getting their exome sequence when they have run out of standard of care options so the question is can we shed a little bit of light give maybe a glimmer of additional set of options when there's much not much else going for them at that point and then Ellie is also cobbled this together and a web based tool that reads out the alterations we also have a we have a group a tumor board of genetics germline and somatic experts who go through this and we have an ability provisionally to make a report that can go to clinic to clinicians so briefly this is all great use genetics to identify dependency give rational therapeutics we get responses but resistance is a major problem to overcome those we're gonna have to make combinations they give us long-term control so the other area that we're actively working on is to sequence tumors that have not only prior to treatment but after a great response following relapse this is a patient who had metastatic melanoma and this patient happened to have a mech 1 mutation which as you can see in these kill curves here will shift the GI 50 meaning they confer resistance to RAF inhibitors and mech inhibitors there are other mech mutations that have been discovered like this fortunately though urc inhibitors which hit one more module downstream appear to overcome this so this is the kind of data that we hope to achieve that will teach us about resistance now I'm just gonna briefly mention this was an RNAI screening study which looked for loss of function mechanisms genes that when silence confer resistance and what came out of this study this was in melanoma as well was that NF1 was the top resistance gene clearly working to cause resistance the reason why I wanted to bring this up was twofold one it's important to think about loss of function mechanisms for resistance as well as gain of function the other reason is that the last time I was in this auditorium I talked about NF1 and neglected to mention that Francis Collins had discovered it and somebody gently pointed that out to me so I vowed that if I ever came back to this auditorium I would not make that mistake twice so NF1 is a resistance gene in melanoma and when we sequenced the exome so here's another little subtlety we found several patients that have NF1 mutations one of them is a standard stop codon which obviously disrupts the protein but the other three and one of them is clearly somatic are occurring in splice enhancer regions or so either splice motifs or splice enhancer regions so again outside of the coding region can be an important place to look so let me just, I'm gonna skip ahead a little bit because I just wanna close by pointing out that what genomics is now doing for us in cancer is putting forth a new vision for how we need to think not just about yet about routine practice but how are we gonna do clinical trials so the thinking is that patients who come in the door maybe they have a fresh biopsy maybe you use archival tissue but you're going to generate some kind of an omic profile and I think the point about starting with genetics is important but it's not all that's gonna be needed there needs to be a way to interpret the data make a decision, monitor response there may need to be even in the clinical trial realm a second biopsy to understand is the patient responding is the pathway being hit using genetics and transcriptomics to help with that and then finally at the point of resistance maybe doing another biopsy to inform you the salvage therapy or in the future combinations that could be given to patients going forward so this is my last slide sometimes in these kinds of meetings we tend to take a historical look because of how much it's been accomplished and sometimes that makes it feel like well maybe this is all done this is all great, we've done all these great things but we are only just getting started the Atlas, I think maybe it looks this is sort of a whatever 17th or 18th century Atlas or maybe 16th century or we didn't really understand what North America looked like we knew that there was a new world but we didn't really have any detail we need to get more detail by completing the mutational axis expanding the axis across disease, metastasis obviously following relapse to therapy we need to annotate the genome and understand mechanistically how all these alterations contribute to cancer and then of course systematic clinical implementation in a way that can test the utility and of course the ability to share this data worldwide is gonna be an important issue so I'll close there there are many people who have contributed to the work that I showed the most important ones are listed here and if there's time I'll take questions. So one of the consequences of our speakers having so much to say is that we are running behind so I think we won't take questions now and you can talk to Levi separately and now I'd like to introduce Dan Rodin from Vanderbilt who is gonna be talking about engineering the healthcare system to deliver genomic medicine.