 All right, so I'm the last speaker and I think what you'll find is most of the things I have to say have been said by other speakers So I'll try to be quick But I think I have maybe a little different perspective from most of you here number one I'm not asking for any money. I'm in the intramural program. You can't even give me money if you wanted to That's number one number two I am I'll say I don't definitely am addicted to encode. I'm an abuser of encode for sure and haven't contributed in any way But maybe we'll get to that at the end as to whether People have small lives like mine, but end up and doing big data Sort of that's just sort of how the technology works now Whether that'll be important and the other part is that I guess one of the few people here has an MD after my name And I think I'm the only rheumatologist. So let me give you that sort of perspective So that some of the things that I wanted to focus on here were You know what NHGRI can do for me, thank you very much Eric. I really appreciate you, you know advancing the goals of my lab But what does the draft genome really mean for rank-and-file physicians who are trying to engage in and Precision medicine and and then sort of practically thinking about here at a place like the NIH where we have a big Research hospital when a patient comes in how can encode and roadmap help us? And conversely though, how can encode and roadmap in animal models help us better understand disease pathogenesis and humans and Think about new therapies So, you know, this is personalized medicine from my perspective We're sort of a mom-and-pop operation picture my lab in the early 90s and you know, this is encode And the question then for me is how does a little mom-and-pop operation? How do we benefit from encode and I think there's a lot of labs that are in my position? So, I mean this is in I can summarize my life in 20 years or so in one slide here And you know in the old days we cloned like probably many of you Started off cloning a single gene when that was hard and took a long time So again, I'm not really sure whether I'm a geneticist or a biochemist or an immunologist But in essence we cloned one gene we sort of helped figure out this pathway Which it turns out to be a very tractable pathway in terms of understanding a lot of the things people have talked about today How you have an environmental signal ultimately altered gene expression. We related this to a rare disease in this case Severe combined immunodeficiency and then use this as a basis to collaborate with industry to get a new class of drugs That are now FDA approved But obviously the goal of encode is is not to have this take 20 years But to do this a little quicker and and obviously I think that certainly is the case now So that to me this is this is the question that I'd like to sort of pose in my talk How does Costco relate to this mom-and-pop corner store that that I have? And so one thing we certainly have done with the advent of massive parallel sequencing and this sort of thing is We've certainly benefited this pathway as I as alluded to is very amenable to thinking about where do does the signal induce a Transcription factor like stats to deposit within the genome and then how does this relate? This is a paper that just is coming out in nature shortly is online And and we're very interested in taking You know assays we did on sort of small level and then integrating them to sort of a genomic scale So the one thing that I did want to emphasize thinking about you know How can encode be used by rank-and-file clinician scientists is is getting back to sort of the ease in which they can use the data So for for genes, you know, it's pretty easy You put in your favorite everybody's favorite gene or my favorite gene at least jack three and in PubMed And you get a whole bunch of hits and that's that's pretty accessible for people who have very limited Computational skills. It's very intuitive. It sets or etc. It's like sort of shopping on for itunes in itunes or Amazon You don't Need a master's degree in computational biology to get the data you need Similarly, you know On your iPhone you have your ImGen app and whenever we're sitting in any meetings And somebody says something about a gene that I don't know about I put it into my iPhone and right away I'll oh, that's that's sort of interesting Now if you put jack three into epigenomics no hits, you know, and I I mean I know why that is You obviously if you put jack three Into the encode roadmap site and you get the UC UCSE browser view and everything But I sort of wonder a long term I know you in in one level if you tried to make everything accessible to everybody It would collapse the system at the same time I think it gets to the point that not all the users are the same There's some users who really know that need to know the details and then you know for clinician scientists They probably don't need all the details and and Eric I think had I forget the these the Tool to use it sounded like crying lack black Ramos or something but But that sounds like a great idea, but I would certainly have NHGRI think about the the end users How this can be most effectively used for the greatest number of people So Let's see. I think I covered a lot of this Integration and visualization. Yeah, that was the other part is that how do you visualize I'm definitely with you Ross on I'd like to put on my my you know virtual You know and and sort of be just like you see and you know on movies and stuff and figure out my favorite enhancer from my favorite Gene, you know ideally on my iPhone now Cross-integration integration of cross species epigenomics. We talked a lot about that user-friendly Obviously more and more we're using epigenomic therapies And that's obviously going to be useful is that if you put a patient on a drug that targets the epigenome Well, what are we going to measure? And I think of diseases like lupus which really disorders of epigenetic homeostasis now again Whose problem is this? Maybe it's not your problem in NHGRI. Maybe it's NCBI, NLM, etc And some of that has been a little bit too before What technological breakthroughs we talked a lot about identification of rare variants in in regulatory elements and so That it really is a challenge to understand Mendelian disorders of the regulome and it will be a challenge when Place like the NIH now we have an interesting patient come in they gets the whole exome sequencing if we don't see anything they get whole genome sequencing and under font or understanding and defining Pathological regulome mutations has been alluded to is obviously a challenge and trying to understand You know which of these are actionable and you know Are you going to do a bone marrow transplant on a patient or something because they have a regular mutation? And I think we've talked about a high throughput assays for this So we talked about how many different cells are there in the body? 400 well This is a Sammy speaking Laplander, which if I truly if I get any of this wrong correct me And my understanding is is that Laplanders have two thousand different words for reindeer Now we have one word for reindeer, but for this guy here His livelihood really depends on understanding the subtleties of Reindeer and and probably is evolved over many millennia etc etc So when you're thinking well, how many cells do we have to do? I think it all depends if you're you know if you're in North America you say you only need one You need to do reindeer But if you ask this guy here who whose livelihood really depends on Understanding all the subtle differences between reindeer then he he would say you know What do you nuts you need to do all two thousand? So you know if he was bad at this if he didn't really distinguish all the subtle differences between all these reindeer He might you know end up having working in an epigenetics lab or something like that rather than you know hurting reindeer We wouldn't want that Now for an immunologist point of view if you asked an immunologist how many cells are there in the body? You know, I think we would say there's probably 400 immune cells, you know This is an experiment. We've done but many people have have done this as well This is an attack-seek experiment and we've looked at all these different cells and all these different states of cells I think there are 80 different regulums expressed here and and But that sort of gets to the point that I don't know that it's encodes problem or NHGRI's problem to look through all of the amazingly interesting subsets of CD4 T cells that Anjan and I and I are so you know in Thralled with just like the Sami speaking reindeer herder So, you know, how many different types of cells it depends upon who you're asking it depends upon the state of activation versus Differentiation depends upon how you activate the cells by diverse receptors and different types of signals depends upon what state of the cell cycle therein Immunologists now have discovered metabolism and and all its effects on immune cells in essence We just talked about that about Exhausted cells and how do all these in data sets get into integrated in the endcode roadmap and This brings up the issue Mike Paysen sent me a paper thinking about this about you know So-called long-tail data that how do you incorporate small data sets produced by individual? Scientists which really do represent the vast majority of scientific data with you know these big data initiatives And so my response in a way is you know, what am I chop liver, you know We do chip seek and we think it's pretty good. I mean, maybe it's not up to your standards But you know, I you know, it's I think it's pretty good So and the question is you have to ask yourself is I think, you know, you obviously don't want Garbage in garbage out that was alluded to but we've also talked a lot here about when is good enough good enough and You know our chip seek data is good enough for me but you know, but you know that that will be your choice and something to think about and So and one level I guess you'd you'd have to think about the lost opportunities by by not incorporating other data sets and the ability To access this Sami speaking reindeer herder when he has done epigenetics on on the two thousand different types of reindeer Versus the garbage in garbage out problem So but that's that's sort of think the challenge I think the other thing that if you're going to Deposit all this information. We do really need a better history of the of the cells that are acquired If they're from humans what diseases that people have allergies etc etc infectious disease geography whether they smoke and all these things again, I alluded to the Well, I'll come to it in a moment with with lupus We talked a lot about epigenomics of single cells We talked a lot about dynamic views and for me It's really exciting the idea of how signal transduction really meets Chromatin biology with these epigenetic changes and and so better understanding of how these changes occur Dynamically and how quickly they change and what changes quickly and what doesn't I think is really important going forward We talked about different ways of viewing these changes Other speakers and understanding this really in real time by imaging etc etc At the same time one can think about this from a different point of view which is longitudinal views Studying the epigenome in in humans over their lifetime, and I would point out again from our business With a disease like lupus that auto antibodies proceed the disease by many decades and so That we can study we can see how the disease evolves, but again a disease like lupus is really quite interesting because there's only 25% concordance in in twins, and so this is really a disease of the epigenome This has been brought up as well using model animals, but I'd remind you that species like The mice we have actually lots of different mouse strains and people like Chris glass This is taking a page from Chris glasses book of using variation in mice to understand the regulum Primates have been mentioned equally people have done large-scale Ian you genesis Projects and one wonders whether that can you can benefit from that as well Another thing that immunologists think about is that we can do a lot with cells But we need to do some of the experiments in vivo and there have been examples where people have transplanted essentially large segments of the human genome into into mice and and and and in many cases these regions really do behave themselves quite well not just in cells but in vivo large regions that are highly regulated in by Diverse in diverse cells in diverse ways like the IL-10 locusts and again this helps moving from association of causality that we talked about And again others have emphasized Evolutionary approaches and I'm done and I think I'm done two minutes quicker. I say two minutes Thanks We almost recovered back our lost time and I think we can hit off on the discussion Do we want to pull back the questions we had before I have no I simply don't know where they are This is just a small point the Is your chip data good enough? I think it's not quality as much as the interoperability questions So having some kind of standardization So I think that's probably has to be worked on first so the standard and so for instance duplicates, you know There's good argument for it not everybody wants to depends on the cells and so so it's it's I don't think it's as much quality for Many of these as it is Is it's just being able to mix the data? I don't know all the issues in that but these guys do I? Just before we do oh to respond to yeah, it's just gonna come in on that I mean we are doing some of this as part of regular omdb, so there is but it's true. It's not Probably a systematic as it should I think we're getting a fair amount of that data and so but I think most people were aware that to get their data deposited into endcode they had to do duplicates That would sort of be a reason to do we usually do duplicates But you know, but I know your standards are out there But but that may be something to think about is make sure everyone understands that that's That's the price of getting into this data set I will point out in this context reproducibility is an incredibly important piece But when you start thinking about working with specialty cell types coming from in vivo coming from cells that are difficult to grow and so on There is also a huge premium on the experiment being conducted by labs that actually does the cell work Well, so there is huge value if there were an incentive to the right labs with the right biology expertise to also do the Genomics part to the standards of a genomics community I think great things would happen which would be very difficult to just siphon off to somebody else doing the experiment before We delve in deeper into that I just wanted to because many of the speakers didn't have a Q&A due to time constraints I wanted to just give a very quick bird's eye of what I think are the key point that each one made so that people have Them in their minds as they start discussing