 So let me introduce someone who needs no introduction, Chris Austin, Director of NCADS, the Godfather of COMP. It's all yours. Thank you. Well, it's great to be here and it was great to be asked by Colin to do this. You know, I've been at NIH for almost 16 years now, but I look back on COMP as arguably the most important thing that I started since being here. And it was one of the – it's a singular pleasure for me, not only scientifically, but community-wise. I don't think you may realize how special the mouse genitus community is. And I know you have your fractiousness, and that came out in the original Banbury meeting. If you can imagine, you know, Alan Bradley and Brian Simbrovich and all these folks in the same room at the same time, you know. But still, in the end, it comes together, as it did in that meeting, and it allows you to do things which are quite remarkable, and I hope you value that. It's not always the case in the human world, unfortunately, the human world, whether it's – we're talking about human genetics or other kinds of human interaction. And I would just urge you when you – as you're thinking about the next stage to be as bold as we were back then in 2003, and I just went back and looked at the – at this paper again and looked at what we said in the abstract, and it's interesting to look at what Steve Slides are, because the community has done much of this. But it's hard to – it's easy to forget how unorganized and inefficient this whole process was back 15 years ago. And in this idea that this last sentence is time to harness the new technologies and efficiencies of production to mount a high throughput international effort to produce and phenotype knockouts for all mouse genes and place these resources into the public domain. At the time, that was audacious to the point of being foolhardy in some's view, but you've shown it not to be foolhardy and have shown over and over again that if you get a lot of smart people in the room that are committed to a really ambitious, important vision, these things can happen, and I look forward to being involved in whatever way I can and encouraging the work that you do in the next phase, and I'm excited about the directions you're going. So, what I decided I would do today, and Colin asked me to do, was just to give you some reflections on some things that are going on in the human world and the human genetics world. I'm not going to go into any detail because you've got fantastic people who are going to talk to you about individual programs, but I'll just tell you sort of where I think in some ways we are. So, this is how I often describe the problem that we're in. It's the problem that brought me to NIH back in 2002, but it's gotten arguably even more extreme since then, is that we live in an almost painfully bittersweet time where we know more about ourselves and health and disease than we ever have perhaps exemplified by the genome project and embryonic or induced pluripotent stem cells, that's what you're seeing on the bottom left, or ability to make mouse models of human genetic disease in remarkably rapid fashion and phenotype them, but if you ask how are people doing on the right side, this is a report from the Institute of Medicine now called the National Academy of Medicine. A few years ago, called Shorter Lives, Poor Health, sort of goes into depressing detail about the health of the American people or lack thereof, and the bottom right is a child with progeria, early aging rare disease, and so we are left with this really interesting question to the degree that what's on the right side is related to and built on what's on the left side. Why has the left side so outstripped what's on the right side? Those of you who are clinicians, or you think about the last time you were at a doctor, for the most part, unless you're a cancer doctor or a cancer patient, your care has really not changed very much in the last 30 years. My own clinical specialty is in neurology, and if you think about neurological diseases, the care of neurological and psychiatric diseases has really not changed very much in the last 30, 40, 50 years for the most part, despite these amazing advances in fundamental science, and so the question is how do we do this better? And believe it or not, I think what you're doing is absolutely key to this, so what do I mean by this? The reason I believe that this has remained such a problem is that science has never been brought to this process of translation. It's a remarkably empirical process which has many, many steps, increasing degrees of freedom and the likelihood of a many-step process with an increase in degrees of freedom getting successfully to beginning of the end if you have a trial-and-error approach, that aggregate success rate approach is zero, and that's exactly what you see, but the answer is not to buy more lottery tickets. The answer to this is to understand what the underlying science is that allows us to go from what we see in these more reductionist models, whether they're a genome sequence or a cell on the bottom left or a mouse to what's on the right side as a human, and I must say that for the most part, though we're making some headway, from a patient's point of view, we got a long way to go. I did a calculation a few years ago just to give you an audacious goal or a big problem to think about. I did a calculation a few years ago with the current rate, how many years will it be before every human disease, most of which are rare numerically, have a treatment, and that answer is 2,000 years at the current rate, 2,000 years before every human disease is treatable. If one looks at the cost function, probably the only thing increasing more rapidly than college tuition is drug prices, and no one will be able to afford these drugs at the current rate either, and that fundamentally comes from this relentless empiricism as well. The only way we're going to make this process more effective and efficient is if we change this from a system of essentially pulling the lottery to engineering that has rules that make this predictable, and that's really the challenge, I think, for the next decade or more. So the question is, how can company IPDC contribute this? Now this is a slide that Dave's shown, Francis has shown, I'm sure you've seen, and I showed this at all of my talks, to illustrate the fact that translation is not new, people have been trying to do this for thousands of years, but the opportunity is new, and the need is new, so when I graduated from medical school and I was a postdoc, this was in the mid-80s, actually before this graph starts, so these are the number of human conditions in OMIM that have a molecular basis. Now there are not all diseases, so the Duffy blood group antigens are in OMIM as well, but just take these as a proxy for human diseases. Studying for the boards back in the mid-80s was really easy because there were only about 10 diseases, human diseases, the molecular basis of which was known. Now it's well over 6,000, I don't know how people study for these things, it makes it really easy for David to come up with questions for the boards, you know, he's just like Professor Snape there, you know, he's asking about how to avoid any spots disease this year. Yeah, right, exactly, requires a lot of rubber bands, you know. But from a patient's point of view, you know, this is where the action is, and still, of all of those, the cumulative number of indications that have any treatment, FDA approved treatment is about 500, of rare diseases, I'll get to this, but the number is actually still a bit squishy, but it is known that about 300 have any treatment and that leaves about 95% or more that have no treatment, and some of these you've heard of, a lot of you haven't, honey disease, Alzheimer's or ALS, many other diseases that fit into this category, and some would say, well, gosh, you know, we all know that it takes 15, 20 years to have a therapy after a gene is discovered, sometimes much longer than that, I often remind people, and you will be glad to know that in every talk I give, I ask people what was the first disease, the molecular basis of which was identified. I got to tell you, I would say 98% of the time nobody has any idea, which is just terminally depressing. The answer is sickle cell in case you're, I hope you all knew that, but the point being that this is 1947 and there's still not a drug directed to that molecular pathogenesis, so just because you have a gene doesn't mean you're gonna have a drug anytime soon, but it certainly helps, but the question is, how we doing? Are we getting better? Are we getting worse? Well, during this, so you say, well, gosh, maybe what we need to do is just wait 15, 20 years and there will be a right shifted increase in the number of therapies that come from these discoveries, right? So that's a perfectly reasonable model to have. Unfortunately, you would be wrong, and the reason is that during this period of time, this has happened, so this is Moore's Law, for God's sake, if you don't know what Moore's Law is, don't tell anybody, because that's even more embarrassing than not knowing about sickle cell. After this, you can go Google it and Google, Moore's Law will allow you to Google it, but this is the increasing number of transistors that you can fit on a microprocessor that's revolutionized what we do in computing in our daily lives, but during the same period of time, a therapeutic development has done this. This is E Room's Law. This is Moore's Law spelled backwards. This is a real paper. If you haven't read it, please do. It is really depressing, so make sure you have some Prozac on or Scotch on board before you read this, but what's remarkable about this is this is showing dramatically, monotonically negative productivity growth in developing therapeutics since 1950, such that the number of new drugs that have been produced per billion dollars spent by the entire biomedical research community has gone down by 50% every nine years since 1950. I just think about gone down by 50% every nine years. I think of what's happened since 1950, invention of computers, recombinant DNA, knockout mice, CRISPR, gene therapy, a huge number of cloning, and none of this has had any effect on this decline in productivity. This is depressing in a number of ways. First is that if you draw out this line, if you extrapolate from this line, it goes essentially to zero in 2070. Now, it can't go to zero because it's a log plot, but it becomes asymptotic in 2070, which means that unless something changes, which has been the case since 1950, there will be no more drugs after 2070, or in other words, the cost will be infinity. The other is that NCAS was formed in part six years ago to bend this curve. We have to bend this curve in ways that invention of computers and recombinant DNA and all that have not. We need to take a fundamentally different approach to this. If I've given you two numbers so far, right? No drugs after 2070 if we keep going where we're going in 2000 years, even at the best rate before every disease is treatable. So if anything tells you that this requires a think different kind of approach, the same way as COMP and IMPC has been doing for the last 15 years, this is it. But this is in human genetics now. Okay, so into this Maustrum came NCAS and NCAS in many ways is like, well, you know, look at the landscape do for lots of other things what you all have done for mouse genetics. And so some of the things that we're doing, I got to go through all of them, of course, but we spend a lot of time worrying about the rare disease problem. And Dave is going to talk about this. He will give you a remarkable talk about this. But there is and but but is a connection, of course, between rare diseases and mouse genetics. But it's amazing how tenuous that is. And I believe and I know Dave believes that this, this is a really opportunity, great opportunity for us. The first number you all can really help with. So we all throw around this number about 7000 diseases, maybe it's 6000 diseases, that's the number in orphan that the number of guard, which is the genetic and rare disease database, but and approximately the number in OMIM. But that number is really squishy. When you when you start to push it, it's really squishy. And so one of things that that we're really interested in doing is to is to do. If I can use the analogy of a cot curve, those of you remember what a cot curve is, you know, that's it. That was what was done back in the old genome era. If you're trying to figure out how big the genome was that you were trying to analyze, we have got to know that for rare disease. We got to know what the universe of rare diseases is. And how do we define that? And it's it's for those of you who are or clinicians, this is a big lump or splitter problem, if you know what I mean by that. But but I think the mouse genetic community can really help us with this, because to the degree that human and mouse genetics are and physiology are conserved, then essential genes in one with the exception that Colin was talking about the human, you know, homozygous null people walking around. You know, who knows, I sometimes wonder whether everybody in this room, you know, is is homozygous null for something of some essential gene. And that's why we're here. I don't know. But but but but the question, the question is, how many genes in a species, which we can prospectively mutate, give us a phenotype? And it would stand to reason that that number ought to be about the same in mouse and in humans with the exception, perhaps of neuropsychiatric diseases, or other, if there's other, if you believe in the concept of human specific diseases. But currently, we think the number is about 7000. They're thought to be about 80% Mendelian, although that, as you know, that is not quite as crisp a definition as we all thought it was in high school. About 50, 60% have their onset in in children. So they're develop, either developmental, or they're, or they have their onset in in in childhood, that is before age 18. And something that Dave will talk about, which is truly frightening from the perspective of developing treatments, is that that if you look at this from a standpoint of syndrome ology, that these people identifying new syndromes through something like through things like the undiagnosed disease program or novel genotype phenotype correlations is about 250 new ones generated every year. And give you a sense of this, the number of rare diseases, which have a new therapy, I developed approved by the FDA for the first time, every year is around three, or maybe four. So we're, so we're moving three or four from the untreatable to the treatable category every year. And, and, and Dave is discovering 250 new ones every year. So either we got to speed up drug development for rare diseases, or we got to we got to defund Dave Valley. Now, I, you know, so I'm agnostic as to which which which is the better option. But but in Washington, believe it or not, that passes for logic, what I just said, which is truly frightening. Okay, so something that I think this community really needs to inculcate and something that we are really trying to change in the rare disease community is the word rare. The word rare is a real problem. Because as soon as you say rare, most people tune out, whether it's, you know, person in the, in the, in the, you know, in your neighborhood or a congressperson or whatever. But, but if you look at the over the cumulative prevalence, 7000 diseases times relatively low prevalence, official definition is 200,000 or below, but most are in the few thousand range. The population president prevalence is thought to be about 8%. Which is about the same prevalence of type two diabetes. And, and these patients account for a disproportionate, very disproportionate amount of healthcare spending, not even including the, the diagnostic odyssey. But we do have problems in the, in the human community. I know it will shock you that people don't even all agree from country to country on what a rare disease is, how they, how they define it. So we got that problem. But the important thing is that it is clear that less than 5%, 5% of them have any regulatory treatment. And I talked about the number of years. So, so Brendan Lee may talk about this later. I don't know, he's, he's one of the, one of the PI's, a PI of one of these networks. This is something that most aprecy is very involved in as well. This is one of our approaches to the rare disease problem. And the idea about this is to, to change the way we approach rare diseases from a, each disease is independent of each other. And these are 7,000 independent syndromes, which had nothing to do with each other biologically, which is of course absurd. But that's, that's the way it would take the extreme phenotype. That's the way this has been addressed to say, well, gosh, these are all related to each other. They're all, these are 7,000, these are, this is, this is a very large jigsaw puzzle with 7,000 pieces. They're all one color. So there's no pattern that we can see. But we got to approach this as a unitary problem and look for, and look for commonalities very prospectively. And so these, this network is funded based on some kind of commonality that the PI will identify. It could be all diseases of a biochemical pathway or an organelle or a cell type or an organ type or some sort of phenotype that the humans have. But, but they group them, the three to 40 diseases, studying them together in some sort of a common biological feature. And, and, and this is a, a natural place for you all to interact with. And I think Brendan, I hope Brendan will talk something about this when, when he talks later. The other thing, which I just want to emphasize here, which is really important for you all, which you've learned from the, from the work that you've done so far is, is, is this. That is that, that NCATs is to rare diseases as NICHD is to pediatrics. That is that everybody thinks of NICHD as the place that does all the pediatrics. And it does a plurality of the, of the pediatric work that NIH does, that it does more than any other institutes as far as I'm aware. I don't know, is that true, Melissa? I think, yeah. But but, but most institutes do work which is relevant to child health and human development. And, and rare disease is the same way, of course. And so the already CRN is run as a big consortium of, of, of over 10 ICs that just happens to be coordinated by us. The other thing we have, which I hope you'll take advantage of, is something called the rare disease, the trans NIH rare disease working group, which has, I don't know what this is, 8 plus 9, 17 ICs on it. This, this is, these are your people. You know, you should think about these people as the, these people think about problems from a human genetics and phenotype point of view in many of the same ways that, that you think about mouse genetics and phenotypes. And, and so I hope as you go forward, you will use this group and the, but we're the coordinator, as you can see with the little, the little hashtag there. And, and Pariser, who's our head of office of rare diseases, is our point person on this. But, but I hope you'll use this group as a contact. The asterisks are, are members of something called Erdirk. So, so what is, what is Erdirk? I think, I think there's some, I must have some response element, which, which, which makes me particularly attracted to large international consortia. Maybe it's because my mother kept telling me that many hands make light work and that kind of thing. It actually happens to be true, as you've, as you've demonstrated. So, about three years ago, I took on the chair of this group, which is just what it sounds like. It's a group of 60 organizations all over the world coordinating their work on rare diseases to try to make headway faster. It is in no way as cohesive and organized as, as COMP and IMPC are. But, but we're, we're getting there. So it's about 40 funders, all the biggest government funders and a number of nonprofit funders like Telethon, about 15 companies, or 20 companies, mainly small, but also some large companies, about 15 patient organizations. And this is something that, again, you should really think about taking advantage of. Most, most mouse phenotypes don't have mouse patient organizations. And that's, this is an advantage of the human world is that you, you do have parents or patients that you can talk to. And of course, a number of scientific programs as well. And, and, and I want to, I want to just read you the goals, because these will sound as absurd as the COMP goal was 15 years ago. And that's on purpose. So the first one, well, so first of all, I need to tell you that the current standard state of the art, if you can call it that, is the average child of a rare disease, a child with a rare disease, and maybe mainly their children, but person with a rare disease, will undergo an so-called diagnostic odyssey, wander from doctor to doctor to doctor for five to eight years before they have a diagnosis and end up with a medical chart about Ye Big. And, and, and all during this time, their quality of life and quantity of life is going down and their, their costs, their spending is going up, right? So, and, and these, so we asked ourselves, well, gosh, if all of these really have been defined in the medical literature before, why does it take eight years for this to happen? This is not a science problem. It's an operational problem. So we set this goal at the head of ourself for the next, in the next 10 years, that all patients come, all people with a rare suspected rare disease coming to medical attention will get a diagnosis within a year. And that's, that's about a log faster than it currently happens. And, and there's all kinds of issues, organizational issues, among other things associated with this, getting people to work together, etc. Very much the kind of things that you dealt with already. And that the, the, the patients who were undiagnosable, and there's after you, if you ask, how many people are there that if you, once you go through all the specialists, and then you do exomes, and then you do whole genomes, how many people remain undiagnosed? That number's around 30%, 40%, depending on who you ask. So there's still many patients who were undiagnosed. But, but and so what we're doing now is to put these in proactively into a globally coordinated diagnostic pipeline. So that within a year, within 10 years, we'll go from, you know, eight to 10 years for diagnosis to one. I actually think that we could do this a lot faster than that. But, but I couldn't get anybody to, I couldn't get them to go, go for a goal faster than a year. The second is 1000 new therapies for rare diseases, which is going to require a lot deeper understanding of pathophysiology of most of these diseases. And then a third one, which is important, but less relevant for this group probably is, it was important for us to articulate that just because we make diagnoses available or make diagnoses and have therapies that are developed. If the patients don't get them, they're not going to help them. And it's not, we can't assume that anybody is going to benefit from having a diagnosis or treatment unless we measure it. And so how do you measure that? And that's, that's really goal three. And so there are a bunch of papers of course that we wrote about this. This is my co-chair Hugh Dawkins who's in Australia looking at the first seven years of URDIC and now the next 10 years as well. And this has a number of people that you'll recognize, people like Kim Boycott and Gareth Baynham and Hans Lockmueller and Petra Kaufman and a number of other folks that you'll you'll recognize. The task forces, I'll just tell you about to give you a sense of the kind of thing we're doing. A lot of these you recognize, matchmaker exchange, the automatable access and discovery, being able to link records, model consent clauses, these are all coordinated with GA4GH of this, the solving the unsolved is what I mentioned before. You know, if you reach the end of the road with current technologies, what do you do? You know, do you look at proteomics? Do you look at metabolism? What do you do to try to figure out what's the wrong, what's, what's, what's going on with these, with these patients? So the, the human genotype phenotype problem is rather silly title, but I use it to illustrate two things what you're going on. One is, in case you missed it last week, the All of Us program released their or announced their awards to three genome centers. There are three of the usual suspects, their University of Washington Broad and Baylor to do sequencing for the All of Us program. The All of Us program, of course, is the hope for a million cohort of Americans who are getting themselves at least lightly phenotyped and some level of genome analysis yet to be determined, but hopefully some sort of sequencing. And, and so that is going to provide a another resource that, that this group needs to be aware of. The other is really interesting discussion going on in the human world, which is very, very much like the discussion that we had at Banbury 16 years ago, 15 years ago, and I'm sure you've had here which is, is it practical? And if so, how would you do it to phenotype humans with a variety of genetic or maybe non genetic abnormalities? And Gareth Fitzgerald at Penn is one of the people leading this, but you'll recognize some of these people, Rory Collins, David Botstein, Rob Califf. And, and so I frequently say to this group, have you talked to the mouse people because they've been doing systematic genotyping or phenotyping for 15 years. And, and, and the user, the reaction is, why would I talk to a mouse person? What can they tell me? So, you know, I feel you feel my pain and vice versa, I'm sure. But, but the point is that, that you notice this, this, this little quote here, deep phenotyping comes close to a human G knockout study. And, and so these folks want to do what you have already done. This is an enormous need in, in human disease for a very simple reason. Unlike mice, humans go to a specific specialist to phenotype and organ, right? So mice don't do that. But if you go to a cardiologist and you have something wrong with your brain or your eye or your kidney, unless it comes up in a buon and creatinine, they will never know because, because it's, it, they'll never look. It's just the nature of our health care system, for the most part. Kids are a little different because kids have generalists, right? They have pediatricians, but, but older, older adults do not. And so it's, it's a, and the other problems you run into is that insurance will only pay for minimum phenotyping, tier one phenotyping, right? And unless, for the most part, unless it's absolutely required. So it's, so there are all kinds of dynamics here, which are, which are a little different. But I'm not sure how different they are. I mean, if you think about, you know, tier one, tier two, tier three, however you talk about phenotyping in mice now, that it's a cost issue, just like it is in humans. So, so all the things that you've learned, I think would be tremendously valuable for human, the human rare disease and human phenotyping world to move forward. The Semat XL gene editing program, you're going to hear about from from Mary Perry, who's the common fund person, who's, who's in charge of this, PJ Brooks, who some of you probably know, who's in the Office of Rural Diseases is the program coordinator. I'm the IC director who's responsible for this. And the idea is the goal, as you can imagine, is to get more treatments out there based on gene editing. But, but what are the needs that are being addressed? So this comes from a workshop, you know, NIH always says workshops about this thing. And what the workshop, what the workshop identified, was development of error free editing machinery, including the non cutting base editors, if you're not up on David Lou's work, recently please do. It's really rather remarkable, looking at transitions and transversions, without cutting standardized assays for looking at genetic off target effects. And this again, things that you know a lot about. A bugaboo, which you may have done working, I'm not sure, is this problem that's gone back all the way back to antisense. That is, if you have a nucleic acid construct, how do you get it to the tissue of the cell of interest? How do you target it there? And that continues to be a problem. This one relevant, oh this says human and animal models, that should say non human primate, large animal and small animal models for preclinical testing. There's a bunch of RFAs that are out to do this as well. And then interestingly, ways to track these modified cells long term. FDA is very concerned that these genetically modified cells, which, unless they undergo a turnover, are going to live in the organism, live in the human forever. And they might go places in the body where you don't want them and do things that you don't want them. So how do you follow them? How do you mark them? How do you image them over a very long period of time? So there are a bunch of programs. These I took out the RFA numbers because they're not open anymore. They're all closed. But these are all programs that have funded grants that are just coming out now to address these. And the ones that you want to interact with, it's certainly are, of course, the small animal models, grantees, maybe some of those in this room, I'm not sure. And then the unattended biological effects and the genome engineering toolkit, highly relevant to what you do. Something which has been an issue over and over and over again, and has come up again, internally at NIH, I just want to mention to you briefly because it does influence the view of what you all do, rightly or wrongly. Root reproducibility issues in research with animals and animal models. The very fact that there is a National Academy Medicine report on this ought to give you pause. And I realize this group is about as OCD when it comes to doing obsessive compulsive disorder, when it comes to doing good science as I know of, but not everybody is. And so one of the things that I think you could get appropriately a lot of benefit from is to demonstrate how to do this right and provide some guidance to the field about how to do this right. And if you haven't read this, this is a paper that I was on, but lots of people from NIH were on, and about transparent reporting to optimize the predictive value of preclinical research. What preclinical in this case means is animal models, that's what it means. And all kinds of issues of blinding and power calculations and all kinds of things. But it's worth looking at this because it's an issue which has come up again. We're going to be talking about this and I've been pulled into this at the annual NIH IC Director's Retreat happening next month. So this is certainly on the minds of many people here. I'm just going to run through this. I'm not going to show you this. What I want to just finish with is another thing I'd like you to think about. So one of the many reasons that therapeutic development fails in this region for this downward slope is that even when a drug gets into people, into a so-called phase one trial, it still has an 80 to 90 percent likelihood of never being approved. That is, and it fails in humans. Things that look safe in animals are not safe in people. Things, drugs that looked efficacious in animals are not efficacious in humans. And that's the case, 80 to 90 percent of the time. So a big problem. Lots of reasons for this, but a big problem that we spend a lot of time working on. And among the approaches that we're taking, and I realize this may be a little, well, maybe, I don't know whether it's whether it's out of scope for you all or not, but I just wanted to have it in your head that what we and many others are doing is to try to ask, well, how would we identify potential drug better using more physiological systems short of a mouse? Because normally what we do is we do a screen down here in some very high throughput setting. And then we generally go right to a mouse. And the lack of any kind of intermediary testing system, whether it's a spheroid or an organoid or printed tissues or organ on a chip, has really created a problem. Eventually, perhaps, we'll be able to go right from an organ on a chip to a human. We're not there yet. But the thing I'd like to suggest is that you might think about doing this in mice. And I'll try to make clear why I think you should think about doing that. But this is the way to think about it. As you go from left to right, you go from a very high throughput compatibility in a 1536 well-plated down to a couple of compounds a day. These human on a chip models have the same throughput as a mouse model currently. And their cost is about the same as a mouse currently. That may come down over time. But that's the current rate. And so one of the things we're doing is to bioprint tissues. This is possible given 3D printing technology and IPS and primary cell technology. We're starting out with laminar tissues, retina, blood vessel, and skin. But a lot of interest in lots of other organs. The idea being that you would then use these to screen for potential compounds, which might be drugs. The Tissue Chip for Drug Screening program is even a more extreme phenotype where the idea here was, could we create a system, a microfluidic system, which represents the structural and functional elements of all human tissues as a way to test for safety and efficacy of novel therapeutics, using the convergence to use a key word these days, convergence of microfluidic stem cell technology, 3D printing, cell sensor technology, optogenetics, those kinds of things. And we started out with 10 tissues. These had to be in a modular, sort of modular platform that you could mix and match these tissues. And they had to be alive with that antibiotics for at least a month. And when we started this seven years ago, again, it was a pipe dream. I got to tell you, this has gone just like this project, has much faster than I ever thought it would for all the same reasons. It's a diverse group of people with different expertise all coming together working to common goals. And that's why it worked. But if you think about what these kinds of so-called microphysiological systems can do, they do all the things that a 2D system can't do. And as somebody who spent a lot of his career doing 2D systems, it gives me pause. But reconstitution of the microarchitecture of the organ, tissue-tissue interfaces, flow control, oxygenation, mechanical cues like stretch in the lung, spacial temporal gradients. Just to give you an example, it is very clear that most cells do not go terminal differentiation until you put them next to cells that they normally are next to in vivo. And if relevant, you stretch them, say, in a gut model or in a lung model. If you don't stretch them in and out the way, hopefully, all of you are doing now, breathing, the cells do not differentiate. And teleologic, you think, well, why would you differentiate? The person is dead. If they're not breathing, they're dead. Why do you care if the cells are differentiated? And so we see dramatic differences in these kinds of models versus 2D models. So if you're interested in two points about this, one, this was done as a massive collaboration, still is, between ourselves and a number of other NIH institutes, a large number of pharmaceutical companies, DARPA and the FDA. If you go to our website, this individual here whose name is Chip, sorry about that, we couldn't resist calling him Chip or her Chip because it's really both, you can click on any of Chip's organs here and see the state of the art of these individual Chips. But the other thing we've done to bring it back to rare diseases is to realize that these microbiological systems might be very good for modeling rare diseases or even common diseases. So we now have, in the next phase of the program, 12 grants of funding, a modeling of all these disorders. And if you look in red, is either the organ or the organ system or the disease that's being modeled. And so tremendous progress just since 2012 when we started this. OK, so some ideas. And I've thrown a lot of these at you already. And you know, per Collins instructions, I've tried to be provocative or set ambitious goals or audacious goals. Certainly, what you've already what you've already set on doing but something which would be amazingly useful is to have knockout mice for all disease, all genes which are designated as rare disease genes. Many of these have even the same gene has tremendous mutation or allelic heterogeneity in the phenotypes that they produce. Salamone is my favorite. Three different mutations in the same gene give three completely different phenotypes. We don't understand at all why that happens. As I'm sure you'll hear about later, knockout models from the Undiagnosed Disease Program, allelic series for understanding more common diseases like all of us in top med mouse models for gene editing effects both good and bad. We're already starting to do that, as I mentioned. Something that you could think about is mouse versus human 3D tissues. The reason is when we go to FDA and we say we think this is a better way of modeling disease, they say, well, the gold standard is the mouse. Show me it's as good as the mouse. And we say, well, the whole reason we're doing this is that animal models are not necessarily predictive. But they say, well, but the gold standard is the animal model. And that's the way regulators think. So give me a tissue chip in a mouse, including a knockout mouse of the disease that you're trying to treat or an allele that represents a rare disease. And then I can compare in vivo to in vitro within a species instead of going in vivo in vitro among species, because you're changing two things at once and you can't figure if there's a discordance why. This is something I've tried to get the tissue chip consortium interested in, but I get the same answer. Why would we do mice? Are you nuts? The whole reason is that we're trying to get out of doing animals, even when the regulators say, no, no, no, the kinds of mechanistic studies that you could do in a mouse you can't do in a human. I've been singularly unable despite pounding the podium to get the human 3D or tissue chip people to work on mice. And I just think they just don't understand the field very well, but you do. Providing guidance, as I mentioned, on the ongoing rigor and root disability of animal model research discussions. I think you could play a big role in that. And providing guidance also on these human deep phenotyping discussions. We are really good at humans in genotyping humans now. I mean, you can do it, well, you know this. How long does it take to do an exome or a whole genome sequence? But the phenotype side of the genotype phenotype has really lagged. And the methods to do it have really lagged. And I think you have a lot to teach on that as well. So I hope I've given you some ideas to think about. I, ironically, have to go and do, in an hour, I have to go and do a teleconference on Ergirk. But then I'm going to be back in the afternoon. And I look forward to the discussion in the afternoon. So it'd be good to be back. Thank you.