 Okay, it's not on. And it told me to just plug it in right there. Okay, and how do I locate it, Dennis? Where's the old Christy mouse? Thanks. Let's see, where do we go? I see it now. Okay, good morning. Let's see, here's the forward. Good morning, Lita Proctor with NHGRI. I have about 15 minutes, so I'm going to give you a very broad brush overview of some aspects of the human microbiome project. Let me just remind you, or let me recall for you, there are a lot of components to the human microbiome project. There's certainly the normal cohort study, which is what I want to go over with you today. There's also the demonstration projects. There's the data analysis coordination center that oversees the data analysis of all the data sets. There's also a technology development component. There's a computational tools component. There's a LC component to the HMP, as well as other elements. But this morning, I wanted to just focus, because I have a short period of time, to give you highlights from the normal cohort study and also products that are coming out of the HMP. I believe the last time I gave an update was roughly a year ago. And then I actually want to discuss, because I was told that the topic should be generally microbiome. So I thought I'd take the opportunity to talk a little bit about two brainstorming sessions that recently happened that are tied to the future of microbiome research. Let me just remind you of the scale and scope of the normal cohort study that's associated with the human microbiome project. I don't know if the pointer is working. Is it working? I can't tell. On the left panel, I'll just remind you that we sampled 300 people over five body sites. That was a non-trivial exercise, because everybody had to be screened, clinically screened for a healthy state in all five body sites. It turns out that many of us can be healthy in 305 and 405, but not always 505 body sites. So we kind of changed the terminology from healthy to normal. These are the five body sites, nasal, oral, skin, gastrointestinal, and urogenital. There are also 15, actually 18, if you include women, subsites of some major body sites were sampled in multiple regions. The middle panel will tell you about the multifaceted data that came out of this sampling. Last year in May and in July, we declared a data freeze on 16S data and whole genome shotgun data. So the data I'm referring to today include 12,000 samples from 200 people. So about two-thirds of the total cohort. Something like over 50 million 16S sequences were produced from those samples. Something like 4.6 terabases of unique microbial metagenome sequence were produced from those 200 people. The individual were sampled three times, but because of the data freezes, we're actually going to be talking about two visits of these individuals. And obviously the full clinical metadata associated with these individuals and separately, around 1,900 reference microbial genomes were produced with a project goal of 3,000. There are also multifaceted analyses that this group undertook. Studies of the human population, of the various microbial populations in the body sites, various novel organisms that emerged from these studies, studies about the metabolism of these microbiomes as well as the inclusion of viruses. I just want to remind you that the final study will also include other microbial forms, including bacteriophage and microcariods. And all this activity involved two clinical centers, four sequencing centers as well as a very large data processing and analysis activity and computational tools as well as ethics considerations. To look at it from the sequencing side of things, the data products that came out of this are something like eight terabases of total sequence coming out of all these different body sites. As you can see, the nary's, stool and the three oral sites gave the most amount of microbial sequence from those body sites. And once duplications and errors and other things were removed, we got it down to about 4.6 terabases of unique sequence. All this sequence was converted then into a number of products. So I don't expect you to read all these various forms, but just look at the blue squares. The blue squares are actually a data product of some form or another. The raw reads are the 16S reads. I actually can't tell if the pointer is working on screen, is it? No, it's not. Ah, thanks. Where's these? Thank you. The raw reads that get deposited in NCBI include the ones that are starting the 16S data, the whole GM shotgun data, and from the two different platforms. I'll address the other products in a minute. The reason I've even threw this diagram up here is for you to appreciate that going from this to this is actually was about a nine, roughly nine month activity of around 100 people to really make intellectual decisions about how to process the data, what are the quality control measures, all the various things that basically agree on a common data set that then could be analyzed. So the data processing step was around nine months of very hard work. So let me just very briefly go over some of the fundamental questions that are being addressed by the group and then periodically I'll use the term dog. It just means the data analysis working group which is kind of a catch-all term include all the various individuals around the country that got involved in the analysis of these data. So the three kind of basic questions that the group is dealing with and it was the mission for the HMP was is the microbiome a stable entity in humans? Is there just a thing as a human microbiome? Is there a core? And finally, these data also allows to get, try to get a sense of what the microbiome function is. So in fact, the question about whether the system is stable is really based on the body site. So what you should see right away, this is like a rarefaction curve an accumulation curve over a subject. So we're looking at 200 subjects and the accumulation of genera as you sample more and more of those subjects at those specific body sites. These are just total data. What you should see right away as the rarefaction curves start to flatten out at about 200 subjects for three of the body site microbiomes, the vaginal microbiome, the gut microbiome and the oral microbiome. But in fact, the gerry is still out for the nary's microbiome as well as the skin. You know, you can take this rarefaction curve and try to predict how many subjects you'd actually have to sample. Turns out the skin microbiome is a very diverse microbiome. Maybe low in abundance, but very diverse in composition. And along those same lines, these are data that were where two different sites were compared for the same subjects for different body sites. And I've tried to highlight them for you in these color-coded lettering. And this is a Spearman correlation coefficient comparing two visits. And what you should note right away that the oral and the stool microbiomes are fairly stable. If you go back time and again, you get approximately the same community. That's not the case with the nary's microbiome, the airway microbiome, in other words, the vaginal microbiome in the skin. Every time you visit, you're going to have enough differences in the population that it starts to look like noise. So you have to have more samplings of those microbiomes to get a sense of the community composition. But nonetheless, one can, in fact, discern distinct microbiomes. This is a kind of a cluster analysis. It's a principle of component analysis. And I know what I forgot to mention is in each slide, of course it'll be part of this record of this meeting, each slide I actually try to name the particular individuals that led to the analysis for each of these kinds of analyses. So there are distinct microbiomes. This is, like I say, a personal component analysis of all the sequence data for these five major body sections. You can see, you know, that the gut clusters together and the oral microbiome clusters together. And because of the sort of diversity and dynamics of the skin and nary's, you have less tight clustering. But in fact, they do cluster fairly well as does the vaginal microbiome. But in fact, I suppose this was kind of implied, but I didn't say it explicitly. If you try to compare the data, you don't see them clustering at all. By gender or by age, they cluster by body site. So basically microbes see different parts of our body as nutrient rich sources to exploit and you see communities colonizing and developing according to the particular characteristics of your body sites. So there are distinct signature microbiomes for each of our body sites. One thing that also has come out that wasn't necessarily one of the goals of the HMP when it was first conceived of is with whole genome shotgun data, let me just remind you what that is. It's actually sequencing all of the genes in a microbiome. So extracting all the microbial DNA and sequencing. The first set of data I showed you was just focusing on the 16S genes which is a kind of characteristic or signature or phylogenetic genes for groups of microbes. So now I want to talk about all the data. So these are whole genome shotgun data or metagenomic data from these various body sites. And actually you can actually take and we get about a million reads per sample. And some of them can be assembled around 40 or 45% of all those raw reads can be assembled into context. You can annotate them. You get something like using reference genomes you can annotate them and around 90 million proteins have been annotated from these various body sites. And then using various kinds of metabolic pathway databases and so on you can begin to assign to those 90 million proteins roughly create a melabolic reconstruction of the various microbiomes. So let me just show you a few of those data. Here's an example of a metabolic reconstruction of oral and gut microbiomes and just to let you know about the color coding the gut is represented by stool is in green and the three oral sites as a supergeneral plaque buccal mucosa which is the inside of the cheek and tongue dorsum which is obviously your tongue is in yellow, blue or red. So let me just walk through this very quickly with you. Sort of focus on the green first. You can see that in the green highlighted metabolic pathways that the gut microbiome is enriched in metabolic pathways for nucleotide, amino acid, carbohydrate and lipid metabolism. Those are tied to either digestion of the food products the enzymes that are expressed to digest food products or remember that the gut microbiome is the largest microbiome on the human body so it's also tied to reproduction of new cell biomass. The oral microbiome on the other hand seems to be enriched in metabolic pathways for environmental information processing genetic information processing as well as some energy metabolism. So this is the kind of analysis one can do using whole genome shotgun data. I'm going to take just a minute to walk through this. So I tried to show you that the microbial community composition is very diverse. But actually here what you should see here is these are different keg metabolic modules across various body sites only a few are represented here. It doesn't even matter how many are represented. What you should see right away is there's not a lot of diversity or variability in these metabolic modules. It might be counterintuitive but in fact the microbiomes are functioning as communities and so the functions they bring to the table if you will to allow that system to operate is not going to be that different. The diversity is rather seen at specific enzyme properties. So the bottom line when one compares the microbial composition of the microbiome versus the metabolic reconstruction of the microbiome is that across all these microbiomes the microbial species are very diverse but the metabolic pathways they appear to support are less so. That's actually going to be helpful probably when it comes to developing biomarkers and other clinical application outcomes from microbiome research. Let me just give you as a taste two unexpected observations that have emerged from these normal microbiomes. One of them has to do with antibiotic resistant microbes and the second has to do with toxin producing microbes. Let me remind you what I said earlier that all of these subjects all 300 people were clinically assessed for illnesses and also they were required not to take antibiotics for some period of time. I don't remember what it was six months or whatever it was prior to being sampled. So there was every effort to sample an unperturbed microbiome from these 300 individuals. However, what you should note right away is that the normal microbiome actually does contain antibiotic resistant members as well as toxin producers. Let me just highlight these two studies. Apparently in our oral microbiome we carry microbes that carry penicillin binding proteins or in other words antibiotic resistant genes. So even when we're not taking antibiotics we are still carrying around microbes that have possessed antibiotic resistance capabilities. In the case of MRSA, which is that resistant gene associated with Staph aureus apparently normal individuals carry Staph aureus with this MRSA gene in a variety of microbiomes and I'll highlight a few of them, particularly on the skin and particularly in nares. A significant fraction of those Staph aureus on our skin and in our nares contain MRSA genes. Can I just ask a quick question? So this thing about the antibiotic resistant. Yes. Since at a certain point we start ingesting food that actually has been treated with antibiotics. Earlier you said there was no difference in these colonies across age and things like that. There's been no evidence that over time that resistant community gets larger or represents a higher. That's certainly an area of active research. I do want to remind you this is a very, very young field of research. I'm not saying that it started two years ago but there's not that much data. I should say there are individuals like Marty Blazer at New York University, Eric Aum at MIT and others who are actually investigating this. Elise, oh shoot. What's the youngest? Silverberg at Johns Hopkins. What's the youngest? In this case the normal cohort study was all adults. So when I said through age it's really a limited period from 18 to 40. However, your point is well taken. It turns out that antibiotics are routinely given to pregnant mothers right when they're getting ready to give birth. So there's this question of how much antibiotic exposure is a newborn exposed to from the very beginning of their life. Those three names I mentioned, Silverberg, Aum and Blazer are three individuals who are particularly being attentive to the impact of antibiotics on actually microbiome composition and stability of microbiome and how much it's altering as a result of antibiotic use. Any other questions about that? Yes, sir. I should have listened more carefully at the beginning. The subjects are 200 individuals from the U.S.? Yes, and I probably didn't relay that very carefully. Actually the total study is 300 individuals from 18 to 40, half women, half men from two population centers, St. Louis, Missouri and Houston, Texas. Those happen to be the clinical sampling sites. And then also clinically shown to be at least normal if not healthy in those five major body sites. Any other questions? The bottom panel here, the bottom table, actually addresses the other issue that came up in the analysis of the normal microbiome. And that is apparently many of us carry around toxin producers in our microbiomes. The examples here of seven of them, I'll just highlight a few of them. Shigella enterotoxin producers are apparently found in our GI tract. They're found apparently at a fairly high prevalence, around 40%. Cytotoxin producers from an E. coli strain is also found in our GI tract. If you agree that stool is a good proxy for GI, around a third of us carry around cytotoxin. And let me skip down to another body site. Apparently many of us, and I already mentioned nary's and skin is a site for staff. Staff is also the toxic syndrome toxin producer. And apparently we carry around, about 6% of us carry around this particular toxin. But they're not dying from these toxin exposures. So there's some kind of balance that's occurring between these toxin producers and potentially the rest of the microbiome. Yes, Claire? Presumably the relative abundance of these organisms or these toxins is low. The prevalence, I'm assuming, is prevalence in the 200 subjects. Pardon me. Yes, prevalence is across the 200 subjects. And you're right. But sufficient signal that obviously we can pull it out. Let's see. Next one. Okay, so let's look at products coming out of the HMP. Examples of products are that there are, as of yesterday, and I think I had the wrong date when I was mentioning today's council meeting, there are something like 93 publications coming out of the HMP that cite support of HMP. Admittedly, many of these are much more technical papers, either computational papers or technology development papers, or some kind of benchmarking study or something smaller. But nonetheless, we have now a tremendous resource of publications that support all kinds of, that cite all kinds of HMP activities. This month, the group has decided that they actually want to release three additional value-added datasets. I think I told you the original goal was just to submit raw reads to NCBI, but in that nine-month activity, the group agreed that all that effort should pay off for others as well. So they want to release the processed reads and assemblies from those metagenomes. They want to release an annotated gene index for all those predicted proteins, and they want to release the processed 16S sequences. Also, HMP is going to be showcased at the next ASM meeting, and I'll address that in a second. And finally, there's now, in the works, two publications that the group wants to develop for nature. Nature's invited us to submit these manuscripts. One will be considered a resource paper, which is more of a description of all the HMP data types and datasets, and also then a separate paper on more of the analyses that are emerging from this very massive dataset. I showed this diagram before. The three starred boxes, those products are already in NCBI, either DVGAP or elsewhere, but the group decided also that they wanted to release the assemblies that they spent so much time developing from the metagenomes, the phylogenetic clustering, as well as the gene index, and that's going to come out this month right before ASM. For those of you who don't know, ASM, actually, the American Society for Microbiology is the largest and oldest life science organization in the world. It has something like, I sound like an advertisement, but I guess I sort of am, there's something like 43,000 members out of 27 different fields in microbiology that have been meeting since 1899. About a third of that membership is international, and it's a huge meeting. My graduate students and postdocs used to call the Disney World for Microbiologists, like 10,000 people come to this meeting annually. This year, in about a week and a half's time, it's going to be in New Orleans, hopefully, if nothing floods, and what has happened, it's not really funny, actually I'm really sad about that possibility, is the recent development in the last two or three years is ASM has decided to hold a session open on the last day for late-breaking news or late-breaking information, and actually they kind of violated their own terms because they approached us and asked us if we were ready to use this opportunity to showcase the HMP. So Dirk Jeavers at the Broad and Curtis Hutton at Harvard are going to be the 30-minute bookend talks that are representing the HMP Research Network Consortium. Dirk is going to discuss the composition that's coming out of the normal cohort data. Curtis is going to talk about the functional data, and I think it's going to be very well attended. And the other part, the last slide, which is HMP and beyond, this is the beyond, we're about roughly halfway through, not quite two-thirds through HMP, and so we've been looking into the possibility of, if there's a new initiative, a follow-on initiative warranted, and there's a couple of ways we handled that. About two, three weeks ago, we held a very informal microbiome brainstorming session. For about six hours, we had some real champions willing to sit through six hours of discussion with a very light meal at the end. Basically, the question on the table is what should be the focus of a future initiative in microbiome research? Very interesting conversation. We had four individuals, including Claire Frazier-Liggett, kind of give their vision of what they think any future initiative should be. It was very well attended. We had 22 external experts join us in the conversation, and something like 30 NIH representatives from different ICs sit in on the conversation and contribute where this felt appropriate. So there was no group consensus per se that there should be a future initiative. I think that's probably a frank and fair statement. However, of the individuals in the group who wanted to suggest a future initiative, we certainly have some excellent models to pursue, so we'll follow through with that. To commit, unbeknownst to me, until I got invited to moderate this microbiome two session, the Common Fund had their own, and Eric mentioned Innovation Brainstorm, their own kind of big think this year. It happened a week after I sat in on that conversation and helped moderate one session, and it seemed that that group, made up these young scientists, were very keen on a microbiome initiative follow-up with the Common Fund. And that is the end of my presentation. Can I answer any questions? Yeah, I just have one more quick question. Yes. So what do you consider, you know, in your graph that you had showing the stability with respect to Pearson correlation or something like that for the... That was a visit for individuals. Yeah, two visits. So what do you consider stable? What range of correlation do you consider stable? Because those correlations didn't look that strong to me, so I'm just... They're not. It looked like they were around 0.7. And that analysis might have been a little premature because there are only two visits and we only had three visits for the entire cohort. In other studies, they're not funded by HMP. Other studies, they're looking at multiple visits as a way to really get at a sense of stability over time. So I think three visits may not be sufficient to get a sense for that. And in some body sites, it's going to take more than just three visits. Yes, sir. Where do we stand on the... I think there was going to be a study done on monogenic twins versus diagenic twins. Is that in the works or is it not in the works? No, I don't think it's in the works. It's certainly not part of the normal cohort. It should be on. No, it is. Okay. That's work that Jeff Gordon's doing and it wasn't part of HMP. He has published on it and he has a variety of results that you would expect them on a zygotic to be different vice versa. So I guess that's a good point that Jane's making. The HMP's main focus was to develop a resource of data, for example, particularly with the normal cohort data for the community. That's not to say that there aren't, in fact, lots of other microbiome projects around the NIH. For this brainstorming meeting, I did a landscape analysis of all microbiome research to date and it looked like around $100 million in microbiome research across 15 different institutes. Any other questions? Thank you.