Founder Stories: Randy Scott, Founder of Incyte and Invitae

I really enjoy the opportunity to come here today. I'm going to focus probably a little bit more businessy than you might normally at a conference like this, but it truly has been my passion for the last 20+ years now. How do we really bring the science and technology to the real world of clinical practice and medicine? And like Brad mentioned, you know, in the early '90s I was involved in the race to sequence all the expressed genes in the human genome and incredibly science— interesting, fascinating science and technology. And then in 1999, I had a really close personal friend who was diagnosed with late-stage colon cancer. And for the first time, I, I went to an oncologist with her and her husband. And we were, you know, close friends, you know, had small kids in school together, etc. And obviously the closest thing they knew to anybody who knew anything about cancer. Um, and just sitting in there in my office and sort of realizing she has stage 4 colon cancer So she only survived about a year and a half after this sort of story. But listening to the oncologist talk about how they're going to treat her— same drugs have been treating stage 4 colon cancer for the last 20 to 30 years. And I just realized that this is crazy. Here I am, at that time I was the chief scientist of a company that had— we had probably close to 1,000 people, 200 or 300 people in bioinformatics. We were even back then doing 10,000 gene microarrays on individual tumors, but really just to build big databases for pharmaceutical companies for research with the idea that maybe 10 or 15 years from now a drug will develop, you know, out of that. And here's my friend dying of cancer. And for me it was just like that sort of shock that I think anybody who— once you get into your 40s, unfortunately many younger, where you have these first individuals close to you in your life starting to pass away from cancer or some sort of disease. So people always ask me, you know, what's the definition of personalized medicine, I say, well, it's when disease happens to you. That's when it really gets personal, and that's when you throw all the statistics out the window. And so since then, through 2000 to 2013, I was at Genomic Health and now Envitae. It's really been my passion to say, how do we take the science and technology and bring it to the marketplace? And by the marketplace, I mean in a way that helps real physicians and real patients improve the quality of care. And just like that experience in 1999 with my friend. Since then, over the last decade, you know, I remember even starting Genomic Health, I always thought that my family was kind of one of those families that we must have really great genes. All the grandparents lived to, you know, healthy old ages, and so genetic disease, genetic conditions, that's all about somebody else, right? And then in the early 2000s, I had a nephew who had a little girl who was born with galactosemia, rare metabolic disorder. It's kind of, hmm, your first instance is, well, it must be my brother-in-law's fault. He must be the carrier because it couldn't be our family, right? And then you sort of think about it and realize, well, that's a really rare condition. People are carriers for those diseases, and people are carriers for lots of diseases, and in fact never, never know it. And then a few years later, a much more tragic case, actually the same sister, so the same family, but she had an adopted son who died at the age of 28 of hypertrophic cardiomyopathy, which is a rare heart condition that you hear about this all the time of that athlete who's in their 20s that dies of a rare condition. Well, because he was adopted, they have no family history and no medical input, and at the time had a 6-month-old son who also carries the genetic condition. So that was kind of the awakening for me to go, it's like, Wow, since then I've had one more event on my wife's side of the family of a child who was born severely autistic, seizures, all sorts of health problems, in an otherwise completely normal family. We've been able to sequence their exome through some collaborations and actually find a mutation that may be linked. So I've gotten more and more interested over the last decade of this whole concept of rare disease and rare genetic diseases with kind of a stunning realization that, you know, first of all, there's no such thing as a common disease. Common diseases are really common because they have multiple molecular etiologies. In fact, rare diseases in aggregate are quite common. And so we need to transform our thinking of how we think about genetic disease and rare conditions, because in fact everybody in this room is probably carrying 50 to 100 mutations that affect protein function Many of which are probably carrier conditions for known characterized Mendelian inherited disorders, and yet we won't know it until we get an extended enough family to where that sort of rare event starts to show up once you get to 20, 30, 50, 100 individuals in the extended family. And so the goal of Invitae, which is actually— it's a 3-year-old company, but I've been involved with it on the board of directors, and we invested from Genomic Health. I left last summer because I'm so excited about this. I just, I couldn't sit on the sidelines and watch this happen. I had to be involved. The goal, I'd say, is really kind of ridiculous. We want to bring comprehensive genetic testing to a billion people on the planet. And people keep telling me, you can't say billion, nobody's going to believe billion. But, but you know, everybody on the planet has a genome, and everybody's genome affects their health and their livelihood. And in fact, we have the technology now. But I would almost say this is going to come to a billion people on the planet over the course of the next several decades, and the sooner the better. So how are we going to do that? Our approach is a little bit different. That's what I want to tell you about. And as we want to aggregate all the world's genetic tests into a single assay with better quality, faster turnaround time, and a lower price than most single gene tests today. So the idea is, let's actually kind of go backwards. Let's take all the known Mendelian inheritance, all those things that are already already used in clinical practice. So rather than say we're going to sequence your genome and then we're going to put an army of bioinformatics people sort of analyzing— that's, that is awesome for discovery and a lot of good is coming out of that. But if you think about we want to scale this to a billion people, it's like, well, we've got to come up with a way to do it faster, better, cheaper, in a high quality control, quality control manner. And so what we've done is pull together a team, and this is not going to be a cheap effort. We We raised about $37 million to date. It took us almost $200 million to bring Genomic Health through all the way to profitability because it's just so expensive to develop tests. It's so expensive then to get them to market and to just the whole commercialization because this field is completely new. And the medical community largely is not ready for the genomic revolution that's about to come. It's going to take a lot of time and energy to teach, but no better way to teach people than to get them activated. Because very few people want to learn about something until they have a case that's actually sitting in front of them that's helped by genetic information, and then they realize how important genetic information is to their future education. So we pulled together a really good team. It always takes great people. And interesting, but we pulled people from a wide variety of disciplines. Like our commercial person is not out of the healthcare field really at all. She was involved in helping us in the early days of Genomic Health, but actually has a background in the internet and at Disney. Because we're thinking, you know, if you want to bring this to a billion people, you have to find ways to think about how you're going to use viral marketing, internet. You can't put a sales rep, you know, sort of out there that's going to drive this thing. And one of our other business people, Rick Pompein— Rick, he's an old guy like me, but he was actually coming out of Stanford. He was one of the early product managers for the first Macintosh back in 1984. And so he had the great experience of watching the success of that and going on to start some other companies. So, and the rest of us of course are all sort of biotechies. You may recognize some names, some really good people from Fate Genomics, Jill Higginbotham, Steve Ligon, and others. But the point is it's going to take a village. It's going to take a lot of people and it's not going to be just about industry. It's got to be industry and academia and the clinical community all gathering together. And so some of the things I'll talk about are kind of core business principles that we think we're all going to have to sort of hold on to if we're going to be successful at bringing genetic information into routine medical practice. So not a pipe dream. We're actually well along the way. Today we have 264 genetic tests on the market available from our CLIA lab. We're in an early access program. We've already done a couple hundred samples. 264 genetic tests for $1,500. Now contrast that with the sort of average single gene test today is generally around $1,000. You can get maybe a few common mutations for $100, $200, but if you're doing whole gene sequencing, like for BRCA1, BRCA2 analysis, or a complete gene sequence for Lynch syndrome, for example, it's more typically in the $1,000 to $1,500 range. But with 264 genetic tests, and I'll show you the list in a second, it actually figures out to just under $6,000. $26 per test. And I'll show you how we're going about doing that, but the idea is that we are just going to keep piling genetic tests into this list because in the new world of massively parallel DNA sequencing, we can go through, curate individual Mendelian inherited genetic conditions, look at all the genes, all of the clinical variants that are known to exist, that are used in clinical practice today. So we're not doing speculative things that are non-discovery. These are things that are well published to be involved in everything from, you know, Bardet-Biddle syndrome to Charcot-Marie-Tooth or whatever the case may be. And we plan every 6 months or so to have a new release in which we'll just keep putting out more tests with the goal that by the end of 2014 we want to have over 1,000 genetic tests for less than $1,000 or effectively less than $1 per gene. This starts to change a little bit about the way you think about genetics, and the first component is that I'd say really 2013, 2014, we're leaving what we've been in historically, which is the world of genetic scarcity. Genetic information has always been scarce, it's been expensive, we didn't have very much of it, so we had to select, we had to make decisions like, well, we'll just look at the most common genes, or we'll just look at the most common mutations because we can't afford to do anything else. Well, the Moore's Law effect in DNA sequencing is dramatically driving down sequencing costs. This is our internal projections looking out over the next 20 years. A lot of people, you know, show the sequencing curves of kind of the last 10 years, and you know how fast those have moved. I don't actually think they're always going to move that fast. We're not going to see, you know, threefold sequencing drop in cost a year like we have probably for the last 4 years. So we project something much more modest. About 30% a year. Here's shown, you know, whole genome sequencing, whole exome sequencing, or in our case, you know, we have our targeted panel. And so if you just project— and who knows what the real number is going to be— but this just projects about a 30% increase— or excuse me, decrease in actual retail cost. What would it cost out in the retail setting to an individual patient or physician over the next 10 years? We're starting to see some interesting things. And if you put on top of this kind of the current retail market price points of genetic testing today, you can get today, if you wanted to get a standard cystic fibrosis test, which is the most frequent CF mutations, something on the order of a couple hundred dollars to do a CF test. Single genes, again, around $1,000. And then multi-gene conditions start getting up into the $3,000. And so there we're starting to see whole exome sequencing actually be more attractive. If you know you've got a condition that has 4 or 5, 10 or more genes, it's just cheaper, easier just to do an exome. Well, the problem with exomes, of course, is that you have lots of gaps in sequencing, you know, areas that don't read as well, and so you also have gaps in your report out. So we're particularly excited about the targeted panel approach where we can target and go after large numbers of genes. And our projection is that within the next 4 to 5 years, this this is actually going to be less than $500. In fact, I would argue we're going to have the entire Mendelian— Mendelian— over 3,000-plus Mendelian inherited genetic disorders for probably less than $500 within 5 years. That's certainly our target. So that just starts to become a game changer in terms of what you can do clinically. So this is a list. We have it kind of crudely broken up, and everybody has big fights about what belongs in neurology versus carrier versus cardiac and so forth. But this is a gene list in our 264-gene panel. We're particularly strong in cancer today where we have what's probably one of the better hereditary cancer panels for $1,500. But also a lot of other areas. And the point is if you were testing for any one gene or any one condition, the reality is the rest of this this information rides along for free. So that also becomes very interesting. A lot of things you wouldn't normally test for, you say, well, but if I could get it for free, it may be useful information. And the more things we can pack in, the greater the probability that we'll find sort of secondary clinical findings for any given patient that are actually quite meaningful. And I'd love to talk earlier about pharmacogenetics because we haven't done a whole lot of pharmacogenetics yet because it's really hard.— and how you interpret it is really hard. You want to be very careful because you're talking about how people are being treated with disease. But this is a big area for the future. Now here's what it looks like if you put it into kind of the clinical framework. Instead of listing it by gene, if you list it by condition— and if you ever want sort of fun, just take this list and go to Wikipedia and you can spend hours just looking at all these rare genetic diseases you've never heard of and sort of astonishing fashion. But the fact that you can get all of this in a single assay now for $1,500 really is amazing technology. And of course we put it into groupings, but most of these things actually split off and there are multiple different conditions that all sort of feed into cancer and may feed both cancer and other aspects of neurological and others. So we even have to think differently about how we characterize and catalog these diseases, which also reflects the symptoms that some of these patients are coming in with. And so we're going to learn as much from this information about how it affects different diseases and different symptoms as we currently know in the literature. So people ask me, you know, why don't we focus on rare diseases? There's no real market for that. What, you know, what are you thinking about? This is kind of crazy. But in fact, we think there's an enormous market here, because rare diseases in aggregate are actually quite common. Here's some numbers that you can take a look at, which I find astonishing. It's probably roughly 5 to 10% of the population has some genetic condition. Many are just diagnosed, and there are syndromes that are reasonably common— 1 in 500 to 1 in 1,000 patients. We of course understand that all of the patients all the population is carrying multiple rare genetic variants. All of us have genetic variation that affects our predisposition to response to certain drugs and effects. For those of us that have some type of common disease, it's likely affected as well by genetic differences. So the classic approach has been to say, well, we've got this incredible new science and technology of massively parallel DNA sequencing, so let's put it to work. And let's start doing whole genome sequencing, whole exome sequencing, and there's some amazing discovery that's coming out of this, but at the clinical level, it also gives you some problems. One is you generate really large datasets with large signal-to-noise problems. How do you analyze all of that data in a meaningful clinical way? And it's effectively like trying to find a needle in the haystack. So we're essentially just going backwards. Let's start with the known clinical variants that cause disease Let's curate those. So systematically go through every individual disease, find all the genes involved, all the clinical variants, and then let's design our next-gen platform to capture all of that, which is really powerful because now you can optimize to catch each of those different types of variants. So now instead of trying to find the needle in the haystack, you're just trying to move a little bit of hay from the needle stack because everything you're looking at is something of clinical value. So the way that we think about this internally is that it really involves kind of 4 simple steps, but in fact they're not so simple. You know, the first thing you need to do, and I would say across all society, is we need to go really organize all the clinically relevant genetic information. You would think that would have been done long ago, but in fact it suffers from the tragedy of the commons, which is because all this information is just put out into the public domain, actually nobody's really incentivized to curate, annotate, find errors, clean the errors up. And we've systematically gone through, as we go through these diseases, and you can find as much as 20% error rates in most diseases where simply, you know, it's just annotation errors or, you know, mutation at, you know, position number, you know, 35 was really position number 350 or whatever the case may be. And so first thing the company did is is to put a team of scientists that's been working for 3 years now to work organizing all these clinically relevant genetic variations. Then we can move into designing industrial-scale genetic testing. If we want to bring this to a billion people on the planet, we're going to have to make it really cheap. It's got to have really rapid turnaround times and so forth. And of course, once you do that, the obvious thing is that now we're going to generate far more information information for any given patient than we've ever done before. And there's really not a lot of systems for how you're going to manage that information. I don't mean just from a bioinformatics or software standpoint, but how do you manage it in a clinical context? And so that's something that we've got to build collectively as a society. And then we're going to find that a lot of that information may be incredibly valuable to other people in the healthcare system, but not even necessarily the patient. The fact that you don't have a particular of disease becomes really important when you show up with a clinical variant that was thought to be pathogenic for that disease. Now you're an incredibly important reference for scientists that are studying that disease, but because we almost always sequence only those patients that have symptoms, we actually don't have a really large population of controls. And in those public population controls will actually turn out to be some of the answers to solving that problem. Why is it they have the same variant that Patient A is pathogenic, but it's non-patient pathogenic in this particular patient. And now we can begin to explore and look at that. The way to do that is to create a global marketplace to allow everybody to share their genetic information. And I'll talk a little bit later, you know, obviously under patient control. So I want to walk through each of these 4 steps and tell you a little bit about what we're doing and where we are. So today we're building what I think is going to be one of the world's best databases that just curates Clinical Variants, where at over 450 genetic conditions curated. Now, this is just hard work. This is— and we have a staff of about 10 to 14 scientists, a few part-time, that this is all they do. This is systematically going through all these different genetic conditions, reading all the literature. So at the end of the day, we now have a database of over 30,000 clinically relevant variants, every one of which is scientifically referenced. We know exactly the publications and papers where that data came from. And we put bioinformatics and software folks to work. One of the great things about being in Silicon Valley is we've got incredible software teams around. And probably one of the more fun things about the company is I would guess for every scientist in the company, we probably have one full-time software developer. So think of it from the lab scientists to the curation scientists to geneticist and team, you know, it's not like they have their personal software developer because everybody's on projects, etc. But this is actually a big part of what it's going to require for success, and this is one of the most underinvested areas probably in most companies clinically today. They just don't have the resources to put a lot of people into developing software tools to make this fast, easy, accessible. But by doing that, we're really putting an automated data analysis and infrastructure infrastructure in place so that from any set of sequence data, we can rapidly run it through this pipeline and within a few minutes determine if it has any one of those clinically relevant variants. So instead of a weeks- or month-long process to screen and go through data, we can do that in a very automated fashion, very fast, for the known clinical variants. But then we recognize that this is all dependent on the literature, and so we actually need We need a broad clinical network to say, well, what about all those clinical variants that are out there in clinical labs all over the world that have never yet been published? And maybe they won't be published for several years, or maybe they're never going to be published, but somebody sequenced that gene in a patient 2 years ago, found a pathogenic variant that was novel. They know it in their lab, but nobody else has that data or information. So together with a variety of groups, including Heidi Rehm and Bob Nussbaum and others, we're sort of joining and becoming part of a kind of a worldwide free the data effort to get clinical labs to actually share clinical variants worldwide so that we can all create a common database and that labs can help each other and help their patients much more rapidly, and then that can actually feed back into this dataset. So the goal over the next several years is to improve the tools enough that will allow us to do all 3,000-plus mendelian inherited genetic conditions and have them into a single assay at low cost. Now this shows the complexity for just one condition. This is cardiac electrophysiology disorders, and even in the literature it goes by many, many different names from long QT to acquired QT, Brugada syndrome, etc. Some overlap, some not overlap. Today clinically, you know, physicians have their own bias. Well, I think it's this condition. Somebody else else thinks it's that condition. For this condition, I look at these 3 genes. For that condition, you look at those 4 genes. Now in the world of genetic abundance, we don't have to look and say, well, let's kind of narrow down and just select 3. We can say, hey, in one single assay, you can actually look at all those things simultaneously. So we're going systematically through this kind of long tail, and you have some diseases where the number of variants can be enormous, up in the up in the thousands, others where it's more, you know, rare. And the fact of the matter is there's been a fair amount of public work on those sort of things that have a lot of— you know, we probably know that there's a lot of variants only because they've been common conditions that people have discovered. And down here in this long tail, there's a tremendous amount of work to do. This is another example, colon cancer, a very complex disease that our team has curated over 16 thousand values for colorectal cancer across all the different genes, screened all these articles to narrow down to a database of clinically relevant information for colon cancer. So once you've got these sort of curated databases, the next step in the pipeline is start feeding those things into the lab. And so we have a very good lab team and technologists now working on this. We have a clear research lab and development lab that's now releasing data to clinicians in an early access program, the capacity to handle over 20,000 tests per year with 264 genetic conditions that I mentioned. But we're also cognizant that in the world of regulation, in the world of intellectual property, we have to make tough decisions about what are you going to do about all the gene patents out there. So, you know, one of the approaches, you know, I always like gravity, you know, water flows flows downhill. Well, innovation goes to the path of least resistance, so we're actually setting up a lab in South America that's going to be a mirror image of our lab in San Francisco because across most of South America there's essentially no gene patents. And so by running the same test in South America, we can actually report out any gene we want worldwide on a country-by-country basis depending on the local intellectual property. And we like to say, look, we don't make up the intellectual property rules. Every individual country does, but if you want the best test possible, you're going to go to the country that allows you to have everything for a much lower cost. So we think this is kind of an interesting approach to solve some of those problems, and we'll talk a little bit later about the patent landscape as well. Now having said that, sometimes you can make this all sound a little too easy. Anybody— and I think there's a really sophisticated group you know it's actually not that easy, right? This is really hard. You have to have really good scientists and bioinformatics people on the curation and annotation side, cleaning up those errors. And of course you just have to make calls. We have good clinical collaborators and people we're working with. The reality is, over time, this has got to be more of an industry-wide or, you know, society-wide program, not just us. It's everybody contributing to help decide— you know, how we set clear condition boundaries, what we decide is pathogenic, et cetera. And likewise on the sequencing side, you know, one of the first things we can do is we can get all the easy stuff first and then more difficult stuff we add on as time goes on. So, for example, this summer we're going to actually add in the ability to do copy number analysis on all 264 genes in our condition based on the sequencing data. So we have a team that's been working very hard on doing doing that because for some conditions, if you don't have copy number, it's like you're really missing half of the variation in that particular condition and/or disease or more, and those can be really important. And then, of course, as time goes on, we'll get to more and more difficult sequencing in that area as well. But the advantage of the targeted approach versus, for example, whole exome or genome sequences, we know what we're going after. So for any given gene, gene, we can target to pull down the genes. We actually sequence all the exons, all the intron-exon boundaries, and then we can also set the target pull-down menu to go into regions like introns or UTRs that would be otherwise missed so that we get the most comprehensive view of all the clinical variants that we know of for that particular gene. So what this looks like in the commercial process is that a physician can actually go online. You can order one or all 264 conditions for the same price. It's at one price, it's $1,500. So if you're a physician, say, look, I don't want all the rest of that data. You know, I have a patient, they're here because I'm interested whether they have Lynch syndrome. I just want to look at X number of genes that are involved in Lynch syndrome. Don't give me all the rest of that data. That's great. In fact, we found that that's even a barrier to some physicians. Physicians, it's like, I don't want to know. I don't know what to do with all of this clinical data or variation. But now that we have it, we can also go back to the patient and say, you know, by the way, we actually are running 264 conditions. If you have other physicians or your general practitioner or medical geneticist might be interested in this, or you're interested, you can actually avail yourself of that additional information. So, you can order 1 or 264 conditions. It goes into a variety of We do automated report analysis and generation, and then we have a clinical sign-out team. We have 2 medical pathologists that are experts in genomics on staff. We're about to add 2 more, and I would say this is actually going to turn out to be one of the biggest areas for cost driver of the whole system, especially as you start looking at 264 conditions and variants simultaneously, and then it's ready to go out to the— to the reporting physician. So this just tells you where we're going in the future, copy number by mid-year. Also working on a lot of the hard-to-do assays, and we're hoping to have some of those up by year-end as well. But the point is we really want to take this and turn it into a heavy engineer bioinformatics software development process. So even the sign-out team, the medical pathologists and geneticists, have the most sophisticated tools we can get so they can say, hey, I want to drill down here, I want to look at this a little closer, this doesn't, you know, this patient, the symptoms they have, we don't think quite fits with that, so I want to look at the data a little closer, etc., and build all those tools into really elegant workstations for the clinical group to then work with the physician. So this is an example of the medical scientific advisory board that's working with us on both, you know, what we need to do with those tools. It really is an outstanding, outstanding group. So, the third level, this is where I think it gets really interesting, though. So, you know, 1 and 2, we're well on our way. We're curating, you know, clinically relevant variants. We're getting them into assay format. We're now live, sort of offering the services to clinicians. But now we're generating, in some of these cases, a lot of information for an individual have to start thinking about how are we going to manage that information. But it's also kind of the excitement of, well, what can we do now that we have all of this information as well? And I would say there's really two stages to that. One, this one is just a simple, okay, like every other genetic testing lab, we're simply running genetic tests. But now we wanted to— we want to move into a world of genome management. So you can kind of split these two apart. And as we look to the future, I think there's some exciting opportunities. I mentioned the Free the Data program, which is in our clinical lab. Every time we find a new clinical variant, we're going to join up with ClinVar and others to make the anonymized clinical variants of just that specific allele available broadly and publicly so that other people can look and compare it. It'll all be annotated so people can call up a lab and say, wow, you know, I just found this variant and I see that you've seen the same variant. Tell me more about the patient or the treating physician or something Is there something you know about this patient? Did you call it pathogenic? If so, why? If not, why not, etc. But we also look at the world of genome management in which, because we're generating a large amount of data and information, it's actually quite valuable to be able to get patients involved in this genome management to say, would you be willing to consent your sample, your information for research? You may not even want to see it as an individual, but that information may actually they help other sort of patients, patients like me or patients like mine for the doctor. And so building those kind of consent mechanisms into this I think is also exciting. And if I had a long-term dream, it would be instead of separating, say, research over here, then development here, then there's clinical, it's like those things are going to completely merge over the course of the next decade. That clinical becomes research, research becomes clinical. And I think there are actually good guidelines that we can use to separate here's when we have a clear clinical result that is something that is actionable, that a physician can do something with. Along for free came all this other information that we want to try to get consented to be able to use in a research setting in ways that can actually accelerate the cure. And one of the classic examples of this is the VUS, the variants of unknown significance, Jill Hagenkord, our medical director, likes to say, "Everybody hates the bus but us." So, you know, our whole philosophy is, well, let's get on the bus, bus, bus, take those variants of unknown significance, and let's do something with them. Let's figure out what do they mean. Are they clinically significant? Sometimes you get buses that are actually really clear. You've got a stop codon in the middle of a gene, and you have other stop codons in the middle of that gene that are known clinical pathogenic variants, so they're Therefore, it's highly likely that this is a pathogenic variant. Other times it's not so clear, or it's simply a modest variant that may be meaningless. The best way to solve these variants of unknown significance is actually to share them in an anonymized format that allows people to do research and find out more about those variants. The other thing is it allows us to really start thinking about patient management differently. And I've chosen one example here where— because probably about half of our ordering right now is for hereditary cancer. Everything we do is hereditary. We're not looking at somatic mutations. But let's say a patient presents with cancer, has an extensive family history of cancer. The doctor wants to know, for various different reasons— maybe the hereditary form of cancer should be treated slightly differently, or maybe they want to use it for screening other family members. So a great example is Lynch syndrome. Lynch syndrome. Somebody comes in with colon cancer, you want to know if it's hereditary or sporadic. Even today, if you just run a 2-gene test for Lynch syndrome with a couple of the most common genes, if you do copy number variants, roughly $1,500 or so. Well, now for $1,500 you can get sort of 50 cancer genes in a hereditary panel. But by the way, along comes a whole bunch of additional information. And over time, we can look at it and say, well, if you're thinking about the whole patient management. It's just not do they have hereditary cancers, they're about to go into surgery. Well, do you know if they have any genetic predisposition to reactions to anesthesia? Do you know if they have any blood clotting disorders? A lot of blood clotting disorders that are in the kind of 1 in 500 frequency rate, so probably no healthcare system is going to say I want to go pay for a 1 in 500 risk that somebody's got a blood clotting disorder and this is their first surgery and we don't want to find it out on the surgical table. On the other hand, now this information comes along for free, along with information about cardiomyopathy. So I love the pharmacogenetics presentation earlier, it's right to this point. Maybe we can identify a pharmacogenetic profile, again, that might not have looked for by itself, that fits. And of course all of the carrier information and the additional genetic information. So even within the cancer sort of field, it's pretty well understood. But not really practiced, that, you know, like in colon cancer, there's actually multiple genes involved in Lynch syndrome, not just MLH1 and MSH2. There's a lot of genes that cause hereditary colon cancer. Because it's not cost-effective today to test for everything, we tend to just test for a couple of things. In hereditary breast cancer, BRCA only accounts for, I think, roughly half of hereditary breast cancer. So that's the most common analysis, but in fact, there's a lot of genes including the Lynch syndrome family that can also affect breast cancer patients. And so now we no longer have to choose. So you get the concept. And then lastly, I just want to say that in order to really drive this forward, we've got to think about this whole world. So if we're generating this information, there are multiple ways to use it, but there's also multiple ways to use it to affect research in the future. And so we think that it's created a global marketplace for genetic information. And to do that, first I think we're going to have to do some disruption. We have to make this really cheap, really inexpensive, drive down the cost. So we've got to get this type of genetic testing down to a few hundred dollars per patient. And the fact that it's potentially one time over the course of your life, so you're amortizing that cost to the healthcare system, makes it dramatically different. On the flip side, I think people could say, well, no, no, no, this is way too early, we need a decade of study or thinking. My point would be, look, the technology is going to go there, and anybody who's got access to FedEx and a computer is going to be able to order a genetic test from somewhere in the world very cheap, very soon. And so as a society, we're going to have to learn how to deal with that information, and it's going to be disruptive across the kind of reimbursement, legal, and regulatory landscape. Certainly in the U.S., the thinking has always been about, you know, that everybody's trying to gouge the system by raising prices too high. So we have all sorts of legislation that assumes that if you drive prices down, that you're somehow trying to induce patients or physicians to buy, and you can be guilty of fraud. So we're actually worried in some of the cases that because you can get the same test for, you know, 1 gene for $1,500 or, you know, 260 genes for $1,500, we could actually be accused fraud under the Medicare statutes, because they say, well, if you can do 60, then it ought to be $5 for one surgery. Well, that's not how the technology works. So we've got to do a lot of work with the regulatory system, with the legal systems, to figure that out and create new models for sharing genetic information. So I'll just kind of leave you with some of our sort of underlying business principles that I think are going to be important. One is at the end of the day, we think that people have a right to self- self-knowledge for their own genetic data, their own genomic information, and that they should control that genetic information. I think that's very important to put control in the hands of the individual patients. But secondly, we're going to require physicians to order interpretive genetic tests, because our view is that going direct to consumer I don't think is necessarily a bad thing, but in order to really apply genetic information innovation into radically improving the quality of health, we've got to bring physicians along. And the best way to do that is to make the whole system dependent on their education. But if we can start landing a lot of genetic— very inexpensive genetic information into the laps of physicians, there's no better way to educate than, hey, I have a patient in the door, and if I don't learn how to deal with this genetic information, the hospital across the street or the physician across the street is. And so therefore, Or, you know, I've really got to turn it up a notch in terms of my understanding of how this information can help. The other is this whole concept that genetic information is no longer scarce, it's abundant. And just like in the internet, people begin to realize we're better off with whole new business models that instead of hoarding information, you share information. I think the history of the biotech industry is all about hoarding information. I've got my database, I don't want to share it with anybody, and I think just good business practice is going to dictate that we're now crossing the Rubicon. And in the future, those companies that share information broadly are going to be rewarded for it. The companies that don't are slowly going to become, really, outside of the system. So we want to encourage genetic information sharing at the patient's discretion, so they should control. And we also support the elimination of DNA sequence patents altogether. We should just— their day has come and gone. Should just be rid of DNA sequence patents, and this information should be broadly available to patients with the ideas we can drive down cost. You're really going to start to see an increase in clinical and personal utility. Today we're just getting to the edge of this having real-world clinical utility, but it's going to grow and grow over time. So I think I'll stop there, but I really appreciate your time and the chance to kind of present a little bit of our vision of the future. Thanks.