 While we are getting the slides ready, let me do a very brief introduction of a new good friend of NHGRI's, Josh Gordon. Josh is the relatively new director of the National Institute of Mental Health, or NIMH. And as all of you, I'm sure, know, NIMH is a lead federal agency for research on mental disorders. We just tell you a few biographical details about Dr. Gordon. He received a combined MD-Ph degree from the University of California in San Francisco. During his PhD thesis, he pioneered methods for studying brain plasticity in the mouse visual system. He then pursued psychiatry residency and research fellowship at Columbia University, where he studied the role of the hippocampus, the brain structure known to be important for memory and emotional processes associated with anxiety and depression. Then in 2004, he joined the faculty at Columbia University as an assistant professor in the Department of Psychiatry. Of course, moved up, as I'm going to tell you about in a minute. At Columbia University, Dr. Gordon's research focused on the analysis of neural activities and mice carrying mutations of relevance in psychiatric disease. His lab studied genetic models of these diseases from an integrative neuroscience perspective focused on understanding how a given disease mutation leads to behavioral phenotype across multiple levels of analysis. This research obviously has direct relevance to schizophrenia, anxiety disorders, and depression. In addition to his research, Dr. Gordon was an associate director of the Columbia University, New York State Psychiatric Institute Adult Psychiatry Residency Program, where he directed the neuroscience curriculum and administered research training programs for residents. He also maintained a general psychiatric practice caring for patients who suffer from the illnesses he studied in his laboratory at Columbia University. Needless to say, he earned many awards and honors in his still rather young career. But when it came time to recruit a new director for NIMH, he was an obvious target and we were delighted that he was willing to come to D.C. and take the helm of this very important institute. Actually, an institute that has a long history of having very good relationships with NHGRI. And it's quickly become apparent that that's going to continue to be the case under his leadership as well. So we wanted to get him in front of this council so you can all get to know him the way I'm getting to know him. And we can continue to explore ways to synergize the research agendas of the two institutes. So, Josh, I'll kick it over to you. Thanks, Eric, for that introduction. And thank you all for members of the council for committing your time and effort to ensuring that NHGRI is well steered, which I know it is under Eric's helm. I want to say at the outset that the presentation I'm going to give you is sort of a general update, slightly tailored to the interests or what I would assume would be the interests of NHGRI council members and public who would want to come to this meeting. I also should mention that it's pretty short. So if you feel like interrupting along the way, please don't hesitate. I really don't mind and there should be time for questions at the end if you don't. So like NHGRI I'm sure, NIMH has a mission. NIMH's mission, though, is tailored, of course, to the illnesses of the individuals who suffer from them, for which we're entrusted. We envision a world in which mental illnesses are prevented and cured. And that's our vision and our mission is to transform the understanding and treatment of mental illnesses through basic and clinical research, paving the way for that prevention, for recovery for those who currently suffer and for cure. And we take these words seriously and particularly I want to highlight the word transform. Because I think that's the way I approach the job. I think we need to really transform mental illness care in this country from multiple perspectives. As Eric mentioned, we're the lead institute at the NIH for research on mental illnesses. We could argue forever about what composes a mental illness compared to other neurologic disorders. But for now it's mostly a negotiation process. And NIMH supports more than 3,000 research grants and contracts at universities and other institutions across the country and overseas. And we also have a robust intramural research program that supports about 600 scientists, primarily on the Bethesda campus, but in other locations as well. This is my first year as institute director and I was fortunate enough to inherit an institute that was on sound footing. And so my primary goal in this first year is to listen and to learn. I'm, of course, interested in listening to learn from all constituencies, including this one. If you have suggestions or imperatives to give me, I'm happy to hear. And in that context, I guess this slide's a little bit old already. I have no immediate plans to launch rapid or dramatic changes in direction. Although as I approach my first anniversary in September, we'll see what might happen then. I take the helm of the NIMH at a very fortunate time, not just because it was running very well under my predecessor. But also because we have an astounding array of technologies at our hand to study the brain and to intervene in brain function. We have finally begun to get genetic clues. At this point they probably are just that, clues. But they are really important clues that are going to, I hope, enable us to break open the neurobiology that underlies much of psychiatric illness. And these technologies and these clues, I think, lead directly to approaches to characterize and observe the connections between circuits and behavior in the context of mental illness. That's been evident in my own work, which I won't go into today, but I'm happy to talk about if you'd like, and the work of many, many other investigators supported by the NIMH. That said, we face tremendous complex challenges. We have a limited knowledge base about what's really there in the neurobiology of psychiatric illness. That's really one of the main reasons for that is a diagnostic uncertainty which is actually that's not the right way to put it. We actually can be really good with our diagnoses, at least in terms of agreement across far flung doctors who've seen the same patients. But what we have is lack or uncertainty as to whether those diagnoses actually categorize mental illness in a way that adheres to the natural boundaries of illness. And therefore adheres to the underlying neurobiology. And although we have treatment options, we have actually lots of different kinds of treatment options. They are limited in their efficacy and their generalizability. And choosing a treatment is always, I was going to say nearly always, but really it's always a try the first thing that occurs to you. And if that doesn't work, try something else. So these are challenges that are probably not unique to NIMH and to psychiatrists treating their patients as compared to other branches of medicine but are much more significant than in other branches of medicine. To address those challenges, I'm trying to build what I consider to be a balanced research portfolio that includes all kinds of measures of diversity but it also includes diversity I think of time frames where we're trying to conduct research and sponsor research at NIMH that will affect the patients of today. And that will affect the patients of tomorrow in the medium and long term. An example of a short term goal that we have of high priority is suicide prevention. These are curves of the suicide rates per 100,000 people broken out by sex, but also totaled in the middle. And what you can see of course is that these graphs are almost linearly pointing upward with no change in sight. And I could put the 2015 and 20, actually we don't quite have the 2016 data yet. I could put the 2015 data on there and again they are increasing. So we need to do something about this, what I would consider to be an epidemic of suicide in the country, over 44,000 deaths in 2015 from suicide. And we have tools to address this. So this is a short term goal because it's a matter of figuring out how we implement those tools on a wider basis. On the medium term goal, there's tremendous knowledge emerging in preclinical models, most dominantly in rodents and in fact most dominantly in mice, to understand the relationships between neural circuit function and behavior. And to be able to manipulate those neural circuits to alter behavior. And one can easily imagine two things coming out of this knowledge base. One, two, we could imagine identifying molecular targets in specific circuits that would modulate those circuits in one way or another. And therefore apply traditional pharmacological techniques to try to develop novel therapeutics. And we can also imagine developing the technology to intervene on a circuit basis in humans in a way that we can only do so in rodents now. Both of those approaches require more focus than we have currently in the NIMH portfolio. And so I'm looking forward to working towards enhancing our ability to translate circuit knowledge into treatments on the medium term. In the long term, I have a goal of trying to enhance the role of computation and theory in psychiatric neuroscience. This includes methods which the NIH GRI community is well aware of, such as data mining, big data, and computational approaches to the same. But it also includes aspects of theory and computation that are more unique to the brain, I shouldn't say unique to the brain at all, but really more unique in terms of what they can do for us in the brain. Where biophysical modeling and modeling of circuits and how they function in a mathematical basis has the potential to really transform our understanding of brain circuits. But we're not going to be able to do a good job of that unless we recruit more people with computational skills, be it from the mathematics or physics areas or engineering areas, to study the problems of psychiatry. And so that's a major thing that I would like to do, which I hope will have long term benefits. And just to give you one example of how I think computation can enhance psychiatry, I'll turn to a slightly big data way. We have this new initiative or reasonably new initiative that was again started by my predecessor called Research Domain Criteria, which is an attempt to address this issue that I brought up earlier about the question of whether our diagnoses in psychiatry adhere to the natural boundaries of disease in the brain. And an attempt to ask that basically is represented in the Research Domain Criteria initiative where we try to break down behavior into its component parts, apply a set of uniform measures to assay that behavior in a trans-diagnostic fashion and ask whether that will teach us more about the underlying neurobiology, or perhaps be more clinically useful. Either one of those things would be nice. I actually am really interested in using a data driven bottom up approach on very, very large data sets with computationally informed behavioral tasks that adhere to the categories of our doc to ask the simple but really difficult, simple question to ask, difficult question to address, is our doc actually helpful? And I think we have the potential to do that if we can roll out a big data approach, a data driven approach to trying to understand how behavior is split up in the brain in terms of function. So next, I want to turn to genetics. Again, this is the part that I think will be most germane to NHGRI. And this is a slide that I've been showing for, I don't know, 15 years in one way, shape, or form, maybe 20, which I actually thought of longer ago than that in graduate school when I decided to study the neuroscience in mice because I knew mice would be genetically tractable organisms. And it's sort of the promise of genetics. You could put any fields data in this, but this is written up for the brain, right? So genes, as we all know, don't code for disease phenotypes directly. Genes code for molecules which are expressed in the cells and in cells that, in the brain anyway, work together in circuits. Those circuits are linked together in far-flung neural systems. And it's activity and function in those neural systems that produce behavior. So if we think of a genetic mutation that predisposes to a disease phenotype in order to really understand how that mutation leaves the disease phenotype in the brain, we need to understand the effects of that mutation at multiple levels in the nervous system. The trick, of course, in psychiatry as well as in other illnesses for that matter is we don't know which of these levels, the cell, the circuit, or the system is going to be most relevant for our treatments. We have treatments in psychiatry tried in true ones like ECT or more modern ones like repetitive TMS that address systems-level function. We have other treatments that target molecular processes in the cell. And as I told you before, we have emergent methods, at least in preclinical models, to manipulate specific circuits. So we don't know which level of understanding will lead to treatment fastest in psychiatry. And that's why this imperative of understanding mutation to disease phenotype across multiple levels is so important. However, and so I want to just to concretize that for a moment, turn to, I could have shown my own work, but now that my privilege is as an NIMH director is to be able to claim credit for everyone else's work, too. So I'll tell you about some really creative work done by a friend and colleague, Stanislaw Sakarenko at St. Jude's, where he's been studying a mouse model of the 22Q11 syndrome. It was mentioned earlier in Eric's talk. This, as most of you know, is a de novo copy number variant. And as many of you know, this variant predisposes to a psychotic disorder otherwise indistinguishable from schizophrenia. And Dr. Sakarenko, as well as others, including myself, have studied a mouse model of this that lacks the corresponding section of chromosome 16. And he's studied a particular circuit in that mouse, the circuit that leads from the auditory thalamus, labeled here as MGV, to the auditory cortex, here as ACX. And he does so by stimulating that pathway in a slice and recording from the auditory cortical responses. And what he's shown in the mutants here in red is that for every ounce, if you will, of stimulation in the pathway leading into the cortex from the thalamus, you get an increasing response as you increase the strength of stimulation to each of those stimulations. But you do so at a lower slope in the mutant. So in effect, what we say is that this thalamocortical input is decreased in its efficacy in the mutant. And what's intriguing is he can rescue that intensity deficit by overexpressing a particular microRNA, which is down-regulated for various reasons in this particular mutant. Now, that's the latest of actually what's a series of really wonderful studies that Dr. Zakarenko has carried out, which suggests at least part of this pathway for mutation disease in the 22Q11 mouse, which is that lack of one of the genes in that region, a regulator of microRNA synthesis, leads to decreased microRNA expression of a number of microRNAs. But one of them, that one that I just mentioned that he used, actually is important for down-regulating the dopamine receptor, specifically in thalamic neurons. And so when you don't have enough of that myRNA, you increase the expression of that dopamine 2 receptor specifically in these neurons. And therefore, they bind dopamine. And the function of the DRD2 receptor is to inhibit the efficacy of those axon terminals in the cortex. So what you get is weakened inputs to the auditory cortex. And there's various reasons to hypothesize, and at this point it's really just a hypothesis, sorry, that that might be related to psychosis. And the easiest to explain reason might be that without this corresponding input, the auditory cortex is left to generate its own activity. And we do know that individuals with schizophrenia, when they hallucinate, have inappropriately elevated levels of activity in the auditory cortex. So this pathway is one potential pathway by which that mutation, the 22Q11 microdeletion, might lead to a psychotic episode. And so that's really nice and powerful when you're talking about certain kinds of mutations, like the large effect size mutation, the 22Q11 microdeletion represents. And this was the promise of genetics, as I said, that I've been talking about for quite some time in my own work. But of course, this is the reality of schizophrenia genetics. So as you guys know better than I, what a Manhattan plot is, I won't belabor the explanation except to say that this paper back in 2014 suggests that there were 108 different schizophrenia associated genetic loci. Now the number is probably closer to 180. The predicted number based on the statistics, if you could, I don't know, genotype everybody in the entire planet would be about 2,000 loci, which is a substantial chunk, as you know, of the human genome. And each of these loci contributes a very, very small amount of risk. The biggest one here, the MHC, may be as much as a 20% increase in risk from odds ratio of 1.2. But these are still small amounts of risk. And it's really hard to move, not impossible, but hard to move from this kind of plot to an understanding of the neurobiology that leads to the illness. And Eric asked me to throw this piece in. We have in the brain, like probably most of the rest of the body, an important additional complicating factor. And that is that in the brain, it's been shown that there's tremendous somatic mosaicism. So in this inherited mutations like the 22Q, they're present actually, 22Q is not familial, but it's present in the, actually, we think it's probably coming mostly from the sperm of the father, but in the zygotes. And then, sorry, it's transmitted to the entire organism if this is your somatic cell division. You can also have early or later somatic mutations. And what's been shown, if you look at brain cells, is that brain cells have a lot of these late somatic mutations. I don't know if it's any more than any other cell type in the body, but the brain undergoes tremendous numbers of cell divisions to produce the complexity and actually just the size, sheer number of cells that make it up. And so there are lots of these somatic mutations that are found. And if you're interested, I suggest you go look at the review that was just published in Science that details the methods, the really, really careful methodology that's been done by NIMH researchers and researchers funded by other institutes, including, I'm sure, NHGRI, to demonstrate that these somatic mutations really are mutations and not mistakes of the process used to find them. So that leads to a likely complexity that is much more challenging and daunting than the simple, linear picture that I showed you before, where we have a whole bunch of genes. And for that matter, perhaps even genes different in different cells, I'm throwing in some environmental factors because we do know that the genetics don't explain everything. And then we really have a patchwork or a mesh or a network of potential pathways leading from these genes to also multiple phenotypes. Because another thing that I didn't emphasize with that Manhattan plot is that many of those loci also are linked to in the schizophrenia GWAS are also linked to bipolar. And so we have multiple potential phenotypes and a web that leads one way. And features of this web could be convergence, where multiple mutations or insults affect a singular process. You can have divergence, where a singular process at one level might lead to multiple phenotypes at another level. And what we hope for in the neurobiology of mental disorders is some form of critical convergence, maybe not all, but at least some of these processes might converge onto phenotypes that we can study and manipulate in both preclinical and clinical models. So this likely complexity, it's frustrating, but we hope to be able to manage it. And that's what we're faced with now, actually, as we have these numbers of different risk factors. And so we've convened at NIMH a work group on genomics, which is really asking the questions of how do we prioritize the neurobiologic studies of the genetic signals we currently have? What are the best experimental computational tools to move forward with that characterization? And how can we leverage different population-based cohorts to enable large-scale genomic discovery with a level of phenotypic understanding that we can't currently with the GWAS studies cobbled together as they are? And another question is, should we start using dimensional phenotypes, such as our doc and behavior, to further delineate the genetic architecture of mental disorders? We've put out a number of funding announcements aimed at trying to spur on what we call an integrative approach or a convergent approach to studying neuroscience of these genomic signals, ideally that span multiple levels and that look at multiple causes and multiple phenotypes at the other end. And if you're interested in the generic approach I'll point you to this Nature Neuroscience paper review by Stainsburg and Zogby, where they discussed how to try to apply multiple approaches to the study of genomic clues in psychiatrically relevant disorders. And I think I'll skip that for now. And so if I could just summarize the road ahead for me anyway is to prioritize excellent science and within that realm diversity, particularly of time scale, I look forward to learning working with all the constituencies of NIMH to address the challenges that are facing. And I think to try to build momentum towards treatments that can change the lives of individuals, families, and communities affected by mental disorders. So thank you for listening. I'm happy to entertain questions or challenges, whatever you have for me. Thank you. Thanks, Josh. I'm sure there's going to be questions. I will go ahead, Carol. I have several myself, but go ahead. So genetics and computational, two of my favorite things. So to integrate data across the different scales that you're going to need to do so, and because these phenotypes are so complex, I mean the challenge, the computational stuff is a challenge, but really the integration of different data types from different sources when the vocabulary is used to describe the phenotypes and even something like schizophrenia, I suppose there's like a zillion different ways to describe what that is. So bringing the data together so that you get actual signal instead of spurious signal is going to be a big challenge. So in the discussions and in your thinking about bringing in the computational component to this research program, what are your thoughts on what are the data and data standards that you're going to need to actually achieve that kind of vision? So there's several different answers to your question, and I'm going to stick to the data standards part and maybe you can force me to elaborate in other areas if you'd like. And even there, there's a couple of different answers. So one is, of course, working with the existing Fenex toolkit stuff to at least get information out there about what we think are unifiable approaches to that. Another is to continue doing what we're doing with our databases, which is try to encourage and enforce people to use some form of common data elements. So we have in our databases at NIH the requirement that when you submit, you work with our coders to, at the very least, if different people have done different sorts of tests for schizophrenia and have different symptom domains scaled in different ways, to tell us which one is about, say, severity of hallucinations and give us the numbers that you've used so that they get coded into an element, even if it's not the exact same test, at least you know where the scale for auditory hallucination severity is. And we have a way of translating between the different codes. So that's among some of the approaches. More promisingly, there are a number of different efforts, including one by the Psychiatric Genomics Consortium in the Stanley Center, to actually roll out wide-scale standardized phenotyping. Their effort is more around cognition. We are developing our own that will piggyback, hopefully, onto the Precision Medicine Initiative, where we'll actually try to cover all the bases of our doc to get databases that are actually acquired in a uniform way. So there's both trying to retrofit old studies and trying to create large databases with new data that's actually purpose-built. So one of the ideas then is the researchers that you fund going forward, there will be some sort of requirement to adhere to the standards so that the data they generate Yeah, what we haven't figured out is if it's going to be a requirement, or whether there's a couple of different approaches. One is to require, and another is to make it just so useful that they'd be stupid not to. And I kind of prefer that approach, which has worked more or less for us in the Human Connectome Project in terms of developing a standard for imaging and image processing that's so useful that now more and more of the imaging world is just adopting it. So I think there's both those approaches. And then the third approach that we're trying, as I said, is to develop our own databases that are huge and compelling to allow, yes, it means there's an investment up front. But on the other hand, the payout in the end is that anybody with a computer can access the data. And with a clever mind and computational abilities can study that database. So there's multiple approaches to the problem, only one of which would be enforcing. And this is a field that's really hard to enforce. We heard earlier from Eric that one of the major groups of disorders being sequenced as part of the common disease grants are neuropsychiatric. And so I was wondering, how are the two institutes interacting around that data generation and analysis? So Eric can probably answer that much better than I can. We had a meeting on this where I was just, you know, where I've been informed about the different collaborations. Let's just also say there are lots of ways that we're collaborating. And one of the more intriguing ones, I think, is going forward is trying to figure out how to maintain these databases that we're generating in a way that makes sense and that's integrated, that allows different people coming in with different studies to be collaborate. I don't know if you want to say anything more. I think Adam's going to come to Microsoft. I think he's in the best position to describe. In terms of concrete things, right now, there's some NIMH data in a pool that's going to be joint called with some CCDG data. So that's the concrete thing. Eric? Going back to your first or second slide about the mission, you know, why are we doing this? Have you engaged the AMP program and is there an AMP mental health initiative to turn genetic discoveries into therapies or even dreaming preventive measures? So that's a great question. We had had one proposal for the AMP around Schizophrenia biomarkers. I'm not sure if it was really genetically based that didn't quite cut the mustard. So we didn't get funding there. I think there are lots of opportunities now that we have specific genes. One can imagine, well, we don't have many specific genes. Really, what we have is loci, right? But as we move towards more and more identifying putative genes for those, we can imagine doing something on a large scale to screen for modulators and test them in preclinical models. One of the big problems is that I don't think we've got and this could be controversial in my field. So I apologize to anyone listening on the web because it completely disagrees with me. You don't get the chance to refute me. I don't think we have the preclinical models that we need for that kind of an effort. Because frankly, I wouldn't trust that a drug that works in a mouse or a rat model is other than a D2 antagonist where we know really well that they'll work is really gonna give us something novel in the area of say, any psychotics or autism where we have the best genetic clues. That said, I think that's one thing we really wanna try to develop is models that will be predictive in that way. And it's certainly worth an effort to try to start categorizing which of these genes are drugable first and then figuring out whether any of those that are drugable might be treatment targets. Yeah, thank you for your presentation. Certainly mental health disorders raise a whole host of ethical, legal, social issues both in clinical care as well as in research. So I'm wondering whether NIMH has had any initiatives in that domain or whether you're considering supporting initiatives in the ELSI domain. So through the brain initiative we support a number of different neuroethics efforts including some initiatives that we're just starting to put out to define the role, to identify the neuroethical dilemmas that novel technologies bring up. At NIMH we've tried to stick closely to our mission around mental health. And so we haven't gone very far into the domains of say forensic neuroscience or other things which I agree are important but which many of our constituents might see as a distraction. So I don't have any plans currently to expand our investments in those areas. Thanks for the presentation. Your last comment about trying to stick within the domains of mental health. As you know mental health has lots of comorbidities. Maybe the greatest one is drug abuse. So are you working with NIDA and what's your thoughts on crossing domains there? So we have a number of efforts then which we collaborate with NIDA. Those could be accelerated and increased. And I've gotten to know nor pretty well over the last six months and look forward to trying to figure out ways that we can enhance that collaboration. There are lots of very, very timely issues in which mental health and substance abuse overlap. One of those for example is the epidemic of opioid abuse which has at least one, at least has a considerable component of psychiatric self-medication I think. Although I think at this point we don't really know that for sure. I wanna come back to the previous question because I should add that one thing, one area in which we're very interested in the legal implications of mental illness and vice versa is the fact that in these United States the legal system, the correction system I should say is the largest mental health care provider. And so that's an area that we are deeply interested in and we have a number of efforts to try to try, at the very least to have quality care delivered by that system but also to try to figure out ways that we might be able to get people out of that system and into the mental health care system. I actually had one comment and then I had a question. So my comment, which may be more of an offer is it sounds like you created a working group of your council around genomics, if I understood correctly. My offer would be, it's easier for me to say it's, people around the state. I am quite sure if there was any desire, any interest in interacting either with this council or any of our working groups, one could imagine you'll bump up against things that you might have more input on and I'm quite certain the network of people around this table and the networks that they're associated with would be happy to help at any time. So I think that would be a great opportunity. Great, I will relay that on. It was actually set up even before I came on. I bet anything there are members of that work group that are your grantees and that are, and or that have served on your council but I'll relay that on to the folks who are heading up the work group. That's a great question. The question I was gonna ask you, is as much about, asked to a practicing psychiatrist but also as obviously an NIH director, it really reminds me of a topic that was very relevant for a workshop we had last week. We have a genomic medicine working group of our council and they put on a series of meetings over the handful of years since I became director and last week we had a two day meeting focused around pharmacogenomics and the current state of pharmacogenomics. And I guess I'd be curious to hear from, in many ways, there's so many questions in so many different areas of pharmacogenomics but obviously in the treatment of psychiatric disorders, that's an area where there seems to be great promise but maybe not as much proven. I guess I'm curious what your opinion of the state of the field for psychiatry, number one and then number two on the research side of it, what an NIH is doing in that arena. We're trying to define our interests and efforts in this arena going forward although many of these applications are very disease specific and are wondering how to sort of move that forward. Yeah, so I think a couple of thoughts along that way. First of all, there are companies out there right now that are selling, as you know, pharmacogenomic testing direct to patients or through their providers targeted at psychiatric disease and I've actually investigated just one or two of those companies on behalf of actually patients of mine or even since I've been here, former patients of mine who've asked me about them and I've seen very little in the way of actual rigorous data to back up claims that they will improve patient care and as you probably know, most of them are based around the enzymes that process drugs so even if they are efficacious really that what they're doing is simply putting numbers on to something that we're already aware of in terms of that some patients are better metabolizers than others and how accurately they're doing that, I don't know. That said, obviously in psychiatry we have such heterogeneity within a disorder and where we are applying the same set of psychotropic medications to multiple disorders it would make total sense to try pharmacogenetic approach and we have funded a number of studies in pharmacogenetics over the past several years. I haven't seen anything come out of it that's particularly groundbreaking so that colors my opinion a little bit. My concern as I'm sure is your concern is that if we need 50,000 patients to find a decent genetic structure of schizophrenia then we probably need that many to figure out which drugs work and we just don't have treatment trials that big. I guess in the back of my mind that's the second stage hope of programs like the Precision Medicine Initiative where you could have a few hundred thousand people and you could have 20 or 30,000 of them hopefully you have schizophrenia bipolar disorder and who've been exposed to various anti-psychotics or mood stabilizers and ask whether different types have different effects based on genetics. I don't know even if that would be powered enough but that's my main concern. And then again one can use these big data techniques to try to extract information but the problem with the big data techniques is that the more variables you put in the more subjects you need to put in as well in order to be able to get statistically robust output. So I guess I'm not particularly sanguine about the opportunities there. I'm happy to be convinced if someone comes along with a clever approach. Just to continue that theme for a second. Now I share your skepticism might not be the right word but you're sort of your recognition of the computational challenges. That said as a layman but a pharmacogeneticist I would put it to you that responses to antidepressant therapy and anti-psychotic therapies are notoriously variable and there is a downside that is I don't wanna lecture you about this because you know much more about it than I do that there's a sense in a general medicine community that well you know if one antidepressant doesn't work then the next one will work but there is a downside to sort of leaving people untreated for a long time. So while the computational challenges are considerable I think it is something that our institute and your institute probably ought to start to chew on. I'm a big believer in the idea that if you understand the underlying genetic architecture of the disease then the pharmacogenetics may more readily follow and there is this other piece the pharmacokinetic piece that is common to anti-psychotic drugs, anti-depressant drugs, anti-arhythmic drugs, anti-fever drugs, whatever. So that piece is a common pharmacogenetic piece but there's the disease specific stuff that I think ought to be a challenge that we ought to address because that's what people need. So I agree entirely. I think our priority now and when I say a priority I don't mean at all to imply that it's the only thing that we will fund. Our priority is on studies that try to break down disease into what we hope are biotypes, subcategories that may be more easily studied both from a treatment and neurobiology perspective that's both of them already then and also a genetic perspective and a pharmacogenetic perspective. Our hope is that we'll get bigger effect sizes and the like in that way that we can start asking these questions with more power with fewer groups. And I should say that at the outset underpowered studies will work if there's a real magic bullet. Was that what you meant to do with your finger there? Right? I think underpowered studies will confuse us. Underpowered studies will confuse us on the other hand like if there's something out there that really is a huge effect size then we won't see it if we don't do the study. So again, I'm open to clever designs and some exploration in that area. I leave it to the experts which would include you folks to tell me what's already been tried so that we don't duplicate things. Yeah, and if I could just add to that, Josh, in interestingly pharmacogenetic variation actually is pretty common and also is pretty strong. And so you don't need the huge sample sizes that you need for some other things. The challenge is getting an adequate treatment response that's measurable and reproducible and I think that's where we could really. And that's a huge challenge in psychiatry that we hope subtyping and or throwing out the DSMs from a biologic perspective, not from a clinical perspective, but starting with a different way of classifying might help us get to more robust treatment responses that we can then study with smaller groups and be adequately powered. We've had enough budget discussion here so I don't really want a budget item here, but what's your philosophy regarding investigator-initiated versus program-initiated projects? You know, so I think in psychiatry we're blessed with a really good investigator base, particularly on the neurobiology side, but I see a lot of wonderful young investigators on the clinical side as well who are creative and coming up with new ideas and it's really important that the majority of our funding goes to investigator-initiated funds. That said, I think there are ways we can tweak those investigator applications to adhere more to the idea that we want to try to make progress and not try to delve deeper just for the sake of delving deeper. Yes, we fund basic studies that are really all about basic neuroscience, but we also want to make progress and there are things that I keep asking my program staff. I ask a lot of so what's and what's next about our grants to make sure that at least we know what the investigator wants to do next, at least we know why the investigator thinks it's important and then we can ask ourselves whether we think it's important as well. To think though, there are places where we don't get enough applications, like for example, on our implementation, we don't get enough applications in implementation science where we're trying to apply the methods that we know work and so we have to put out applications, program announcements there. On the computation side, we're gonna try to be aggressive, as I said, to bring in people into the NIMH sphere that are not currently in the NIMH sphere, so they're targeted areas. And then I'm gonna take the liberty of answering a different question than the one you asked, but I thought you were going there as about small science versus big science, because I think, and the reason I'll take the liberties because big science is program initiated all the time, right? So what I like to do from the big science program initiated stuff is to try to create resources that will foster better small science. Akin to the mapping the human genome, there's now, I'm sure you guys are finding tons and tons of individual investigator groups that use data that come from the human genome project. So we want to build things like the one thing I was talking about before, a large database of phenotypes that's linked to potential genotypes into EMR, et cetera, that investigators can mine. And what I'd like rather than funding some big effort to do the science around it, I look at as funding the data gathering and then making the data public and then funding the investigators to mine that data. Let me ask one more question. We love big science stuff, but since we have you here, can you just give like a one to two minute summary of the current state of the brain initiative? Because we don't hear that much about it. Yeah, sure. It would be just fun to know, just sort of programmatically, budgetarily, where things. Yeah, yeah. So the good news is the brain initiative in the latest FY17 budget actually got an extra $100 million that helped us get closer to the original plan of about $400 million a year. I don't think we're there yet this year, but we're much closer than we would have been without it. We've been going now for about four years. There, all the grants are come in in response to program initiated announcements, but they're written in such a way that we get a lot of creative stuff. We've spent the last four years trying to encourage tool building. And those tool building operations are really a bunch of areas, but for the sake of time I'll mention two, and specifically one is tools that will enable us to monitor brain activity, different calls for preclinical versus human, but on a vast spatial temporal, sorry, vast spatial scale and very fine temporal scale. So massively multi-parallel electrical recordings, optical recordings and all kinds of other creative stuff. Another one which is sort of in parallel is tools that will enable us to categorize and characterize all the different cell types of the brain and their tremendous progress has been made using all kinds of single cell technology, a single cell transcriptomics, especially to characterize brain cell types. One of the interesting things you might be fascinating to know is that it looks like from the early returns in the Allen Brain Center where they're actually transcriptomally profiling the transcriptomes of single cells as well as patching them doing some basic neurophysiology is that you can have at least as you can tell in a single cell, which I know is not that deep, you can have two cells with pretty much identical transcriptomes with different physiologies in a dish. I don't know how that works or if that's true, but it's fascinating. So this gives us actually, I think from the NIMH's perspective, the chance to really do some deep neurobiologic dives at a single cell level and try to do what we're calling sort of genome-wide neurobiology or genomic neurobiology or transcriptomic neurobiology to understand these gene hits that we're getting. So anyway, those are two major technologies. We're now at a pivot point in the brain initiative where in the next year or two we're gonna be starting out, putting out more announcements that encourage investigators to use these tools, well first to disseminate these tools and second to use these tools to answer questions in basic neuroscience and then I think we'll really open up the gate to being able to use these tools for disease-specific processes that individual institutes might wanna fund. And just to say a couple sentences about the oversight of that project in terms of which institutes. Yeah, so there's I think 11 institutes and centers that fund different parts. The main institutes are the NIMH and the NINDS and Walter and I nominally chair the group of IC directors that gets together once a month to provide direct oversight. Grants are given out by individual institutes that with money that's divvied out to them by the brain initiative and so the individual councils actually provide a counselor oversight although there is a working group of the, it's called a multi-counsel working group that also serves to advise Walter and I and the rest of the IC directors. We've been searching for a brain director, the brain initiative who would sit in NINDS but would serve to replace Walter and I as the chairs of that IC group and or I shouldn't say replace but advise us in that capacity and that was put on hold because of the hiring freeze but we're hoping to be able to move forward with that soon. Carol? Hi, so this has been fascinating and good luck. Thanks. I wanna return to the question about your statement that you weren't interested in starting an LC program in your institute and I was fortunate enough to have a post talk a couple of years ago who got a K99RO award. He had, he was probably the most degreed post talk I ever had. He had a PhD in neuroscience. He had a law degree and a master's in bioethics so why he came to our program is really never nobody could figure out but he did get a K99 award to look at people, to study psychiatric genomics in a project that was ongoing in Pennsylvania in among individuals with really who are mostly in long term care facilities with really psychotic, mostly diagnosed as schizophrenia but not necessarily actually having schizophrenia patients and in my post-docs project he's now on the faculty of Baylor School of Medicine. His project is about informed consent and return of results to people either their surrogates or themselves if they had capacity, sort of what kinds of things one ought to do. How to get informed consent to participate in whole exam sequencing and then what to do about return of results. And as you've been talking, it struck me that you might be very interested in hearing about this. Other kinds of projects that are funded by the LC program, we've all been very caught up with return of results issues but there's really interesting LC projects and bio-banking and obviously informed consents always been of great interest. So while you might not want to start a program you might want to link up in a variety of ways with the LC program. So I would just kind of offer that. This guy in particular is Gabe Lacer-Almunoz. He would probably be delighted to talk about his project to some of your folks. So I just want to suggest that. Thanks, I appreciate that. And let me just say that we do have efforts in these kinds of areas, particularly intersection with mental illness and capacity. So it's not that we don't have anything in that area but I think it'd be great for us to work together with the existing resources so that we can learn about how to address these issues in our population. Okay, well, thank you so much for coming, Josh. I knew this would be a productive conversation. I think there's several areas we identified for continued interactions. And once again, I just want to stress as you even heard on this last issue even around LC work, this council is always very cooperative and interested in reaching out even beyond our institute and being helpful. So I'm sure if there's particular things you would like to get input from us about, we'd be delighted to be helpful. Thanks, I really appreciate it. Thanks again. Okay, we're going to break for lunch. Let's resume at one o'clock, please.