 That's working, yeah. It's been a terrific conference, and Ross is getting the computer hooked up. And the panelists, I think I don't need any introduction really since they've all spoken, and since their names are clearly labeled right there. And while I'm waiting for a, don't you have some kind of high throughput robot that could do that faster, Ross? Never mind. So since this is the penultimate talk in this meeting, I'd like to take the opportunity to thank NHGRI. Eric probably won't thank NHGRI at the end. So in 2003, we were studying a lot of gene expression in the testis, and we were seeing all kinds of transcripts that weren't annotated. And if the slides come up, they're not up there, though. And we were finding a lot of unannotated transcripts, and this is a cartoon that I was using back in that period of time. And we would look at the annotation, which was a wonderful thing when the fly genome was first sequenced. And I went to complain, actually, to Francis Collins. And I said, isn't there something we can do about this? And he said, well, you need to go and talk to Peter Goode. And so I went and talked to Peter. And it turns out that plans were well underway in NHGRI for tackling exactly this kind of a problem that people recognize it. And as usual, I think, NHGRI was kind of ahead of the curve. And I would like to thank them for that. And I really look forward to seeing what happens nine years from now, given the kind of long incubation period for these projects, and then finally seeing them through to fruition like this one. And I also like to have a special shout out for Peter Goode and Elise, because in a lot of modern code meetings, you hear people referring to their job being herding cats. Well, Peter and Elise are the herders of the cat herders. So that's a doubly difficult job, I think. And they've really pulled off a fantastic project. And we're all really excited with the results. And they're going to be useful for many, many years to come. So I think we should give them a round of applause before we move on. What happened? I see it here. We're going to go through the questions that the panelists are posing. And then we're going to take all the questions and have a nice, hopefully good discussion at the end. So my question is here. I mean, like a lot of people, I've just been absolutely stunned by all of these GWAS studies. They're incredible advantages to working on humans. Like you don't have to screen for mutants. They walk into the clinic. And the kind of pie throughput sequencing that we can do now has just opened up all kinds of exciting avenues for research. And pretty much this, I think, is a typical kind of a GWAS result that I've highlighted here, where we know that a particular trait or disease susceptibility is highly heritable. There are few Mendelian kind of things that are easy to get at. And after that, it gets messy really, really fast, where you have a lot of minor, small effect variants. And then you have to get into all kinds of complex combinations and network kind of analyses to kind of get close to getting to finding all of the heritability in these. And so my question, which I have some, at least, potential answers to, which I won't give, because I want to see what other people think, is what can modern code do both now and in the future to help inform these GWAS studies and try and make sense out of these oligogenic phenotypes that are much like what Ross spoke about. So that's my question. And Jason, you're up next, and your slide is the next one. I can just run it, I guess. Is it, I don't know how it's organized. I gave him my, OK, that was it briefly. Hopefully it'll come back. All of the slides are on the same talk. So we were asked to come up with provocative questions. So my question is a very simple one, which do genomic technologies diminish the utility of model organisms? So we heard a little bit, I was really glad to hear Ross's talk earlier today, and actually Eric mentioned at the beginning of his talk that genomic technologies have really increased the accessibility of mammalian cells and human biology to inquiry. And one could think of this, if one thinks of it as a zero-sum game, then I guess you would say that maybe the utility of model organisms is diminished accordingly. So anyway, so I put up some yes and some yes arguments to this question. So with decreasing sequencing costs, many of the advantages that these models were originally chosen for, for example, smaller genome, become less important, right? I mean, if the genome's two orders of magnitude are smaller and the sequencing cost has gone down by 10 orders of magnitude, does it matter as much that the genome is small? Snips and mutations can be readily identified now. And so traditional genetic mapping and screening, in many cases, can be obviated. Improved transgenic methods allow controlled single copy integration into mammalian cells. There's improved knockdown methods for interrogating loss of function in mammalian cell cells. And also barcoding and cell sorting technologies make possible cell-type specificity and highly parallel functional assays. So all of these things are really democratizing to the study of human cells. And the question is, does that make things that we've traditionally done these types of studies in models, right? So the no answers are, there's a lot of obvious no answers. One is you can't control a lot of critical variables in human studies and I wonder whether, you know, whole organism human studies, you can actually control any variable at all. And so that's one, you know, and obviously control of the experimental conditions is a prerequisite of science altogether. Human studies are expensive. Mutants are not often available. Some mutants can't walk into the clinic, right? And the relevance of cultured cells is always a question. So people have these questions all the time about how relevant cultured cells are. And actually this is the one I was happy about with that Ross address, which is that the integration of cell biology, so even if you can do these experiments in cell culture and biochemistry, if you can do experiments in extracts of human cells, it's very difficult to integrate those results with whole organism biology in humans. And models are really good for that, right? Because you can easily, that's where the generation time and things are still extremely relevant is integrating those results. You might get at a cellular level or biochemical level with the whole, how the whole organism behaves. And then even invertebrate models like mice, whole animal studies are slow and difficult. So anyway, these are the two sides of the argument. And I think it may seem like everyone in this room is saying, oh, well, of course not. Models are really important, but this is the debate. I mean, we hear a lot about an emphasis on translational studies, or why can't we look at this directly in humans now? So I think it's something worth explicitly discussing. You wanna do it from there here, it's up to you. Hi. Okay. So I think in a similar way, I think maybe picking up on both Brian's and Jason's components, which is some of the challenges that we're hitting up against in mapping human genes are perhaps more easily addressed in the model organisms. And I think there's both advantages and disadvantages of actually moving to the models. I wanna first show sort of a little bit of that and then sort of talk a little bit about the challenges ahead. So the advantage of the flies, as we heard very eloquently in Ross's talk, is that you can actually map the interactions. In other words, having these very short generation times is something that genomic technologies cannot get around. In other words, you can make the double and the triple and the quadruple knockouts and make crosses specifically based on your hypothesis of interest. So you can go to higher order disease models, which is a lot of where these, a sort of the technology of being able to study these regulatory networks gets us to in the sense that it exhibits sort of these interactions between multiple loci. And also it's also where the challenges of these single locus and additive models of human disease are getting us at in the sense that a lot of the misinheritability might actually be sitting at the interactions. So basically that all of these higher order aspects might actually be more easily addressed in the model organisms. So again, the rapid crosses to actually isolate the genetic contributions in a controlled environment, the rich information of the regulatory network, but again, this is something where the human side of things is very rapidly catching up and then the ability to dissect these rare variants in different contexts. Now some of the disadvantages is that the genes that underlie the disease might actually not always be conserved. The networks themselves, i.e. the interactions, even if the primary genes are conserved, the ways that they interact might actually be different. And number three, the phenotypes themselves might not be available in the fly. So basically constructing these models for human disease is a challenge. The second aspect is we're struggling a lot with regulatory mutations, namely mutations that fall in non-coding regions that are not immediately sort of associated with a gene, or that are certainly associated with a gene, but in a complex way that depends on the specific regulatory element. So mapping these regulatory regions to mouse can be certainly possible, and within mammals, most of the regulatory elements are in fact conserved. However, if you go beyond mammals, if you go to flies and worms, all the regulatory components, the specific regulatory elements, are actually not conserved. So I think that's a big disadvantage in the sense that sort of testing the specific regulatory functions of these non-coding variants might actually not be possible. And then the interactors themselves can be distinct. So again, we have many challenges ahead, many opportunities to address some of these questions of whether the genes, the networks, and the phenotypes are conserved. I think the modern code, the current phase of modern code has a lot of opportunity. And to that effect, we can, for example, look at, we can basically ask, are the genes themselves that are associated with disease that come from, for example, OMIM, or from the NHGRI GWAS catalog, or from the cancer gene sensors, how many of them have actually orthologous genes? And you can see that a small fraction of them do, actually substantial fraction, I should say. And then how many of the diseases themselves have orthologous relationships? So I think that the current mapping of orthologous genes at least gives us some opportunity that this may be relevant. Number two, if you actually ask for the functional enrichments, are the functions themselves conserved, then what we're finding is that for a large number of categories, what we're actually finding is that the diseases indeed have functional enrichments associated with the fly genes. And again, that's very encouraging. And lastly, if you actually look at the genetic interaction network of fly, many of these diseases actually are sort of revealing other diseases that they're connected to. Again, giving us a handle for actually studying relationships between these diseases. So what I want to, I guess, end with is that there are many challenges ahead, but I think a lot of the work that Modern Code is setting up right now sort of can sort of perhaps bridge some of these challenges. But again, I think the discussion will cover a lot of that. Oh, I remember my question. My question was, and I think yours is gonna be similar. What's the difference in going after disease between basic approaches to, or basic research approaches to disease versus translational or therapeutic approaches? And I think the best way I can illustrate this is in an example. Because there isn't, I think there's an underlying assumption in most of the meetings I go to, including this one, that, as I mentioned, identifying mechanism means identifying therapeutics. And I think that's not always the case. And there are many, this is a discussion to go for a long time, but let me explain the difference. For example, my dad had quadruple bypass surgery about 20 years ago, almost 20 years ago, and he just got back from a cruise from Alaska, okay? So what causes heart disease? We really still don't understand the mechanisms that drive disease, and that's why we have difficulty, even identifying risk factors for it. But it doesn't matter because you can bypass it and you can get on with life, all right? And so that's an empirical approach to the disease and it works, and in fact, this past year, heart disease is now the second leading killer of Americans because of bypass surgery, whereas cancer has now taken its place as number one. So in other words, if you're a basic researcher and you identify mechanism, you win. If you're a translational researcher and you identify mechanism and you don't have a therapeutic, you lose, and if you identify a therapeutic and you have no idea what the mechanism is, you win. And that is a mindset that I think really is quite different from what we're used to. I certainly had to move towards that in the things that we're interested in. And in going to the modeling and so on, I would actually turn the case on its head and say not our model organisms being marginalized by network analysis and so on. I'd actually turn it around and say to date what has network analysis done to treat major diseases that had market success on rare diseases, and of course being a little harsher than I anticipated here. But the truth is, for example, with the drugs we've been developing, our assumption is not that those drugs are even addressing the tumor directly, all of the activity. So I showed you some of the pathways that it's acting on, but I actually suspect that some of the effects of the drugs we're developing are on other, either neighboring tissues, on the metabolism of the fly, and so on. The network analysts that I've worked with to date have really thought of these networks as sort of standalone networks, and they have not considered, because at the moment it's essentially not possible to consider the complexity of the animal. And while it's easy to put that aside and say, well, that's something to sort of deal with, again, if I go back empirically and ask, rational drug discovery, how well has it worked as opposed to phenotypic drug discovery? We have spent a lot of money on rational drug discovery, but the outcome has been mixed, harsher than I expected, but whatever. Well, I guess I would take quite a different perspective because I'm not willing to say that we're just gonna do medicine empirically for forever, we're never gonna understand, and I really think it's a matter of perspective. If you go back a set of five or 10 years, 20 or 30 years, I mean, could you imagine if we didn't know that DNA was a genetic material now? I mean, fundamental research has a massive impact on medicine over longer periods of time. You can't get away from that. It's inevitably over a long-term driven, most of the advances in medicine. And we mostly, we're just at the very, very beginning of understanding metazolean organisms, so this process is just gonna repeat itself again and again. I think the biggest problem right now is that people think they can understand metazones without getting into the full complexity of tissues, their 3D structure, all the different cell types. You can't just grind these things up, you can't just put them in a dish, and I'm sorry, but in IPS cells, can it just be a useless model of disease in 95% of the diseases because they depend on structure and interactions of cells, all of which are lost when you just have one cell. And you see this in mutants. I mean, most trisophila mutants have non-autonomous interactions. You see it's studying stem cells, which I do. If you just have the stem cell and don't know about the niche, just forget about understanding it because it's a property of multiple interacting systems. So we just have to start loving the complexity, which is, I guess, one of Ross's terms, love metazone multicellular 3D complexity. And the sooner people get that in their heads and just start thinking that way and working that way and using our fantastic technology to now, in effect, have that kind of genome information on each one of those cells. And you know how they're all interacting and what's close to what and what's far. That's what's gonna be the wave of the future. And then, step by step, we'll fundamentally understand how groups of cells work and then how tissues work. And it won't be any harder looking at the human genome and seeing some changes to predict what the phenotype is than if somebody brought you in a yeast karyotype and there was a hiss mutation and you'd be able to predict what that phenotype is. There's, it's not impossible to understand metazones by fundamental biology. So we'd like to open things up now on that very strong last set of comments. So if you wanna go to the microphones and while people are doing that, I mean, one of the things that I think came across in all of these questions was this idea of complexity and that we have to embrace it and we're being faced with that continually. And I think that as more genomic medicine starts making its way into the mainstream that that's gonna be increasingly seen as important to embrace that complexity. Yeah, to continue to echo some of the remarks that have been made here, I want to go back to little older days and see what the reaction today is. You will recall not too far away and quite recently just yesterday, there was this adage that the animal models really tell us about the possibilities but not the actualities because we might be able to study a phenomena polygenically as two people mentioned today. But in real life, it is not just polygenic phenomena. Phenomena is polyfactorial as you just mentioned. Are we anywhere away from that kind of adage that the animal model really, really tell us what can happen will really not tell us what really happened. So we will have to go back to ourselves, clinical trials or what our experiments with ourselves to find the arrived at the conclusion which affects us. Comments? So I guess I can start. But so clearly there are all kinds of factors, right? I mean, so there are people that smoke cigarettes for their whole life and don't get lung cancer. There's a lot of complexity and not only in the genome but in the interaction between the genome and the environment. And we really don't have very good models for that even in the model organisms right now. Although I think that that's changing, that especially now that we can sequence so many different genotypes, we can start to set up especially designed populations of model organisms with that you could seed a population cage for example with some of Ross's potential tumor promoting mutations and essentially select for a tumor phenotype or then try and use these populations to screen for efficacy of some of the drugs and kind of really build models for personalized medicine using these model organisms. And I don't see any other place you can do that other than using cell lines. And I completely agree that cell lines while they have a lot of wonderful properties, I mean they're not organisms and it completely ignores physiology. So if you- If you let me hang myself to the wall and you with me, you made the comment that we don't have good models. Let me make the proposition that we will never have good models which would be able to give us real prediction about what is happening to us. So that's one point you can remark on that. What is missing, what is often called the reality check is that the environment within which we live cannot be reproduced in models. Maybe tomorrow we'll be very smart about system biology but I am personally not hopeful that we will not soon end the limitation of our ability to reproduce what occurs in nature. So if I can just step in on this. Okay, and again there's many threads going here and I'm sort of choosing which one to address but to address the no good models comment. I think that was your comment. So for example in mouse, mouse metabolism and we've all been told this but in stepping into the diabetes field I've really come to appreciate it. Mouse metabolism is utterly unlike human metabolism in just astonishing and fundamental ways. For example, if you feed a person or even a fly a high sugar diet they'll become diabetic. That's not true in a mouse. You have to go to a high fat diet and then if you take a rat and feed it a high sugar diet then it's fine. So the difference is even in fundamental processes even across mammals is striking, okay. And then of course as you go down the chain it's gonna get even more distant. But to follow up on the second part of your earlier point which is the, so that's one thing is that at the end of the day it has to be tried in people and the best you can do with model systems is to increase the odds that you'll have success in the clinics. You'll never know until it goes into a person and then to sort of riff off that point and what you started with. And I completely agree with Alan. I'm 100% behind the idea that eventually our ability to model things more and more sophisticated in ways and understanding the mechanisms behind things is where it all has to go, okay. To me that's a tomorrow versus a today issue but we don't disagree otherwise. And I think Alan would agree with that. But let me just- I agree with you. I think we are making great progress. I'm very hopeful. But I think we need to be increasingly, as we advance we need to be increasingly more critical. Right, but exactly and the point I wanna bring out is this, yes I agree. And the point I wanted to bring out because this is a point that we often don't see in our basic research world, is that for example in the 1980s, spending per year on R&D by pharma was in the range of $2 to $3 billion total, okay. Last year it was in excess of $80 billion. So now let's review the success of that which is the only thing that matters. Everything else is talk, right. It comes down to new composition of matter, drugs and related things that go into the clinics and actually help people live longer, right. And as you may know, cancer for example is a terrible example of this because five year survival rates for cancer from 1950 to today has moved only slightly which is disturbing to think about. If you look at, so in 1996 which is an echo of drug discovery done in the 80s there were 53 new drugs introduced into the marketplace based on that level of spending. In the past decade that number has hovered around typically between 10 and 15. New drugs introduced in the marketplace per year despite the fact that spending has jumped probably 15 fold. The difference between the 80s and then the 90s onward and we all remember when all our buddies went into pharmaceutical companies and did rational drug discovery, right. The idea is you're gonna go into drug companies, you're gonna identify the mechanism of the disease and then you're gonna hit that with femtomolar specificity and the disease was gonna yield to this. That's the outcome of that result, okay. And a lot of it is exactly what Alan is saying which is we were probably naive in thinking that we could model things to a useful level. I don't care if it's a complete level, I don't care, the only matter is if it's a useful level and clearly we have not been able to do that and that has cost the pharmaceutical companies dearly to the point where they're all facing patent cliffs and they're all merging and so on and so forth and it's because the PhDs went in there and essentially changed the paradigm from phenotypic screening to rational drug discovery. So I think a bigger question would be and it goes to something you were talking about is are we ready for rational drug discovery or do we need to just do the basic research, move along those ways and in terms of actually generating treatments, emphasize phenotypic screening. It will only increase our odds, it's not gonna be perfect, fly models, for example, clearly are gonna have failures but increase those odds is we can't get worse than we have now and that's really the point that I wanted to bring forward. This is a point that we don't see, we just sort of think that by studying mechanism we're helping people today and that's simply not been proven to be the case yet. I think the burden of proof is on us. So I did, I mean one of the things that I think is interesting about model organisms is that it gives us an idea of how certain mechanisms have evolved over time and the impact that those have in the biology, particularly, I mean cancer is an obvious example maybe of that is how those evolutionary changes that may be functional in one organism may be mechanisms that are selected for for certain types of diseases. So I think understanding those mechanisms in another organism is important because it gives insight to what mechanisms may later on be selective or advantage, some type of advantage for later diseases. So one can think of evolutionary change in many different scales. So you can think of it at the level of sort of fly versus human but you can also think of it at the level of sort of within mammals or even within primates or within human and I think each of these levels can tell you different things. For example, looking within the human population you can find loci that have been selected across the history of humans and very frequently they have been selected for one thing and they're now associated with the disease associated with something else and I think these connections, as you say, are indeed very useful. So in mapping genes within mammals or between flies, worms and human what we're finding is one to many and many to many paralogous relationships that are indeed connecting pathways that were previously isolated. We saw a beautiful example at the time course yesterday when you compare fly and worm, the developmental time course where you see these sort of almost duplication of late embryo into pupa which again sort of may suggest connections there between pathways. So I agree this is something that can be exploited to a level that goes beyond just drug discovery but also very much into basic mechanism of pathways that were previously isolated or disconnected and connected that way. Can I just say something too? I mean I think you don't have to look at the problem just at the level of genes. I think the mammalian tissues that are very poorly understood even at the cellular level until recently there were not really accurate knowledge even of this where the stem cells are, how those tissues were new and in fact there are actually many cell types in mammalian tissues and human tissues that are not yet described and model organisms can really help you in finding these things. So for example like my lab has worked on the fruit fly ovary where in the course of we and others found a number of cell types that were not classically described by histologists. We can now look into the mouse ovary where the somatic cells are very, very poorly characterized and so we don't even know the cell of origin of a lot of ovarian cancers. By correct, just by correctly identifying the somatic cell types and what they actually do in this very parallel process that you see in a mammalian ovary or a fly ovary you can get tremendously useful information like the origin of cancer. There's just an awful lot of straightforward things that can be done with model organisms to advance our understanding of mammals. I think another aspect to touch upon is just this extreme pleiotropy of many different genes. The fact that you may be studying one gene in one cell type that's associated with a particular function and then if you look at how that gene behaves in a different cell type that function can be very, very different. So when we're looking at genome-wide association studies that map organismal phenotypes to loci, I think a step of translation of that into not only the relevant pathways but also the relevant cell types that actually might be involved in different pathways I think is another component of complexity that you're bringing up. Okay, maybe speak up a word for the mouse as a halfway house, old platitudes. It's hard to study the nervous system or immunity and yeast and it's even hard to study it in Drosophila. For example, the marvelous things we've heard about about cancer chemotherapy on the basis of genetic networks, really beautiful but one of the newest, most exciting cancer therapies came out of the study of mouse immune system and inhibition of inhibitors of T cells. Similarly, the patterning of neurons as studied for instance in the retina classic studies is coming out of studies of mouse and vertebrate models and talking about complexity depending on whom you believe they're between 40 and 80 different kinds of amocrine neurons in the retina period and what they are. I don't think you could find that out from Drosophila or even worm, thanks. Well, there was this kind of old concept that for any particular question that you might have there's a perfect model organism out there someplace that you can study it in. And I think that all of these kinds of discussions about what can and can't be done in a mouse or in a fruit fly or in a rat, what that really means is that we have to kind of have a collection of model organisms that we can pose these questions to. So we need flies and we need worms and we need zebrafish and we need mouse and perhaps we need some to develop some of the emerging models that look particularly promising. Hopefully there's in tough budget times there's still enough money around to occasionally take a particularly surprising result that an investigator has discovered a new system for something and develop that to see if it can't kind of join the mainstream of the model organism community. I guess I actually had a different point to make but first I have to strongly disagree with Sherm's point which is even for things that may seem to be more complex in mice and humans, I mean the basic rules as we've discovered over and over and over again are often still applicable to the model organisms and more easy to discover in the model organisms. So though I agree there's more complexity in mammals, it's not that you can't learn significant and important things about those pathways in humans. In the model organisms of course just to raise the somewhat humorous example which is a few months ago learning that flies that don't get enough sex drink more. It's probably not for exactly the same reasons but you know. And I'm sure you're not talking about yourself personally. No, no. But just to amplify your point though and you are right we have to and I think I had mentioned this in my talk you have to be smart about what and maybe this is your point about matching problem to organism there are plenty of problems that are not well modeled in flies and I think that's an important point to make and you pointed out some of them. You're right, I'm always surprised to see some of the ones you can but flies are limited in for example looking at nervous system details and so on and a lot of times it's the details that matter. So I actually would agree with that. No I think there's certainly. But the point I wanted to make or question I wanted to raise at least I don't know from the last couple of days but mostly many years of thinking about these issues and this really expands on Allen's point and others about complexity. For me if we look at yesterday's panel discussion there was, we were talking about how really it's for monocode and specifically in code and the focus of these projects which is the nucleus. It's the same thing, it's the same point that Allen was making with respect to tissues. It's not just about the 2D linear order of genes and what binds where. It's about the 3D dynamics of the processes that we're studying. So you can take the same principle, go down to the molecular level of it's about how proteins, how atoms interact and then continue to move up. It's about how the molecules interact. It's about how nuclear organization works and how things 3D structure and the endemics at the cell level, at the tissue level and backed all the way up to environment organisms. And for me at least the complexity of those so many levels of interactions absolutely requires model organism studies because we will not distinguish. And I think, I can't remember who it was who raised the issue about it. I mean in human studies it's great except that everyone including like for example my genetically identical twins who are epigenetically different. There is no control. I think that was Jason's point. Even in the case of twins, genetically identical twins it's not a controlled situation because each has experienced a different environment. So I think my point is just that I don't really see how we can solve a lot of these problems just in terms of human biology and human disease without really thinking holistically about that whole spectrum of structure, architecture and complex interactions. So I think everybody agrees with that. Does anybody disagree with that? Yeah, very good. So I've been told we have time for one more question. So please. I'm here from FlyBase and we've done a lot of work over the years to make the information generated by the fly community more accessible to the fly community and we are starting to think about and be strongly concerned with how do we make all of this model organism information more accessible to the larger biomedical community? What sort of approach should we be taking? I can actually press a little bit of that. So the question is how can we make FlyBase accessible to the general community? First I have to say I love FlyBase. It is really beautifully designed. It's simple and so on. And on that point, many of my colleagues who are not fly people have actually gone onto FlyBase, looked up their gene of interest and then come to me and said, oh, I found this and so on and so forth. So usually a million disease workers and mouse workers and so on, they have a question about a particular gene as do flies have it and so on. And I have to say they've apparently been able to easily go on there and navigate through. So I think it's actually well designed. That's easier. Go ahead, you were about to follow up, I think. Oh, okay. So I think that FlyBase and mod and code data in general, speaking also of worms, that it ought to be incredibly useful to people that are in the clinic. And a lot of times, I think that they can go and find what they want in these databases. But a lot of times they have some difficulty. And I think that when we've gone through the effort, for example, of making, Manolis making these and also Mark making all of these lists of orthologs, if you could have a portal into some of these model organism datasets from some of the ones that people in the clinic are used to looking at, that would be great. So if you could go from whatever site you happen to be in and be able to essentially query FlyBase or to query the mod and code data in the new encode DCC, which hopefully will happen, that you would be able to enter an ortholog list and get some of the model organism data kind of spewed back at you. I mean, I think that that would be kind of a minimal step that we need to make sure happens for making it easier for clinicians to get it at model organism data. I think what you need to do is go beyond just the genes as the point of comparison. So for example, the gene data would still be there, but that's fairly low down in the hierarchy and genes do a whole variety of things in different contexts, which often doesn't come across very well in these databases. So maybe if you had more cell type and tissue type organization and show what the gene, like show what is an insulin producing cell in a fly? What major pathways and genes does it use? And that to a clinician or to someone in another system who's interested in that kind of a cell, that gives them a much more useful entree into what's similar or different with their system than if you just say this one gene, that this sulfonial real receptor is conserved between mammals and Frosophila, which it is by the way, very highly conserved. I think that's a really good point. And I believe FlyBase is actually working on something like that to be able to query FlyBase by human disease, right? We are, we are working on that. So I mean those kinds of things I think could be incredibly useful and maybe in the future if you could do something like being able to query by some kind of a sub-network model or something like that, that would be great. I guess we're finished. So thanks everybody.