speaker-1: Hey Momo. Listen, ⁓ I've always wondered something about the physics and the chemistry of human gas. Meaning, you know, what happens when people fart? Okay. And I was surprised to learn that the average human being produces about 1 to 1.5 liters of intestinal gas every single day. Yeah. And you know, if you scale that up,

speaker-2: Hey guru. Who hasn't?

speaker-1: Over an 80-year lifespan, your microbiome is basically manufacturing about 30,000 liters of gas inside your body. Okay. And that is enough volume to inflate a hard air balloon. It's sort of like we're all walking around like zeppelins.

speaker-2: Wow, that is a big microbiome party producing a lot of gas. But I am more interested that's a great number, thirty thousand liters, big number, but I'm more interested in the chemistry of that and how it affects human health than the the sheer volume.

speaker-1: Yeah, yeah, I know I realize that not all gas is created equal, right? ⁓ you know, there's there's hydrogen, there's actually some oxygen, some carbon dioxide, some methane, there's ammonia, there's sulfide. And not all these gases are active and playing a role in your health necessarily, but some of them are playing a pretty big role in your health. And today we're gonna go deep into their function and their relationship to your health. With Two really good molecular scientists that we've invited onto the show.

speaker-2: And I absolutely love this. Not all gases are created equal. I think people can have that as a dinner starter at their dinner table, whether it's at home with the family or out at a restaurant with friends. Hey, did you know that not all gases are created equal? And people are like, What are you talking about? Well, let me tell you.

speaker-1: Some of them are are smelly and some of them are not smelly, right?

speaker-2: That's that's at one aspect, but we want people to take the we want people to take the conversation much, much deeper. And we're gonna enable you in this episode to know so much about gases that you're gonna be able to have a really meaningful, funny, and educational conversation with your friends and family about gases produced by your best friends who are living in your gut. So thank you for that. That was a great intro. And ⁓ I am Omo, a biochemist.

speaker-1: Health aspect. Exactly. I'm Guru, an AI expert. Were two PhDs on a pause?

speaker-2: All right, let's start rolling, guru. We're going to announce the bio monthly challenge winner. So can we please roll the slides? And we have our monthly challenge winner. If you remember, the monthly challenge was to make a social media post on something to do with personalized nutrition. And once you do that, you just need to tag it with ⁓ hashtag two PhDs on a pod and ⁓ let us know about it. And so here we have one of our viewers, ⁓ Blake Barry, who posted about how important personalized nutrition was to him and even showed or linked a paper by Viome and boom, he was the winner. So he gets our full body intelligence kit and we're going to ship it to him and we're gonna provide hopefully some pictures if he is willing to show that he received the kit and he's enjoying it.

speaker-1: Awesome. Well, congratulations, Blake. A beautiful, great host you have here. And I think the paper reference is amazing. So thank you very much for that and congrats again. And speaking of the VIO Monthly Challenge, the new monthly challenge is going to be very simple for any one of you to participate in and potentially win the flagship VIOM Molecular Test, full body intelligence. And the only thing you need to do. is to come up with a question and post it in any one of our episodes in June. If you did that, and Momo and I will evaluate every single one of the questions that we received. There's some really good questions that we received in the last few episodes. We're going to talk about one of them towards the end of this episode. But when we look at all the questions that we receive, we will evaluate them and announce a winner, a lucky winner, who's going to get this full body intelligence from why.

speaker-2: Man, that is one of the easiest challenges I've ever heard.

speaker-1: Yep, Mama, this one is gonna be given to some lucky person here. So let's get into our episode. What's on your radar, Mo?

speaker-2: Woo! So many things, but I had to pick two, so here we go. So if you can share the next slide. So I'm really, really excited that I'm going to attend the 20th conference called SFAF. This is a top, top-tier genomics conference held for 19 years and now 20th year in Santa Fe, New Mexico. It spun out of basically the core Joint Genome Institute and Los Alamos National Laboratory Genomics Institute. At that time in like 2007, eight, nine, Los Alamos National Laboratory had sequenced more bacterial genomes than the rest of the world combined. That's how big it was. It was really a powerhouse. And ⁓ and I was a part of that core team back in the day. And so I have obviously a very strong emotional connection. So this picture is from my last year's presentation on how microbiome is related to aging. This year I'm going to focus on the clinical applications because we just launched a whole bunch of healthcare provider products. And I'm going to focus on those products, including the scores that we're going to discuss today and also disease risk stratification and just clinical applications, because I think that's really the next step for genomics is to really make an impact in daily lives and not just in the academic community. So super excited about that. And then if you go to the next slide, the second thing that's on my radar is. I want to introduce something that I've never heard people say, but Mexican stir-fry. This is like the simplest thing to make. It's super nutritious, it's super delicious. It's very, very flexible, meaning you can do whatever you want. But let me just introduce people to this concept. So, what you want to do is make that spice mix on the right side, chorizo seasoning. And there are many versions of chorizo seasoning. And here I just picked one that has some core components. And the ratios of these is up to you. Like for example, there's smoked paprika, which is very sweet and flavorful, and there's cayenne pepper. And you can adjust the ratio. So, like a starting ratio of 10 to 1 would be nice, where it's very sweet, very rich flavor with a little bit of a kick. But if you're a kind of person that requires a kick, you can add as much cayenne pepper as you'd like. Cinnamon is the same thing. If you add too much, it'll dominate, but if you add a little bit, it'll be really enrich the flavor. So Make that spice mix in some smaller batches until you tune the ratios. And once you do that, like we have done, we now make a kilogram of this spice at a time. We literally make one kilogram and put it in a jar and we use it all the time. And once you have that spice mix, cooking becomes super easy. You basically reach in your fridge and pull out all the veggies you have and/or mushrooms. You fry them in olive oil with as much of this spice mix as you want. And that's gonna be your veggies to feed yourself and your microbes. And then you're gonna pick your protein. And in this case, we chose beans. You can choose any protein you want, fish, chickpeas, ⁓ chicken, beef, doesn't matter. You cook it with the same chorizo seasoning spice, and then you pick your carbs if you'd like to have carbs, and that can be a flour tortilla or a corn tortilla or rice or anything you'd like. And boom, the meal like by the time the rice cooks. Everything else is ready. Of course, the beans take longer to cook, so you should have those pre-cooked, you know, just make a batch of beans in a pressure cooker and store them in the freezer and then they're ready to go. So this meal is less than 30 minutes and it's super nutritious, super delicious, and you can make so many different varieties that you're not gonna really get tired of it. So yeah.

speaker-1: Amazing Momo. I actually have variations of this with other types of seasonings as well. With for example, you can do the same exact thing with Indian seasoning, which is masala. I mean you do everything exactly like Momo just described, you just put masala instead of chorizo seasoning and it will still work, right? Or other kinds of seasoning too.

speaker-2: We actually do exactly that guru with Ethiopian seasoning and masalas. And what I found very interesting is Tandour spice is obviously famous in the whole world as being Indian. But guess what the number one ingredient in Tandour spice is?

speaker-1: ⁓ you tell me is it ⁓

speaker-2: Chile. Chile. And chile comes from Central and South America, not India. And berbere spice in Ethiopia, which is like in every meal in Ethiopia, the number one spice in in Berber is chile. And so we have these multi ethnic cuisines where chile is the core, like chorizo seasoning, like taco seasoning, like tandour. The core seasoning is chile, gives us that that that major flavor.

speaker-1: She

speaker-2: And then you make all these different varieties that add quite different quite different, you know, ⁓ extra flavors. Yep.

speaker-1: Sounds great. Well, listen, speaking of ⁓ food, I wanna I wanna tell you ⁓ something that was on my radar. I went to this ⁓ really amazing wedding of a friend of ours, a very close friend of ours, and ⁓ we enjoyed the whole thing thoroughly because it was done at the Shakespeare Garden in ⁓ Center Park in New York City, which is a beautiful place. If you haven't been there, it's like a little niche with lots of flowers and everything and there's an old clock type thing and you can just Take pictures there, and it's just such a beautiful, you don't forget that you're in the middle of ⁓ a big city. It's it's so beautiful. So anyway, after that wedding, we went to this Italian restaurant where they gave us some incredible eggplant Parmigiana. And I was thinking to myself, Okay, this is so amazingly tasty. How do I make this amazingly healthy at the same time? And I came up with maybe two or three ideas. Okay, so I could, for example. Instead of taking the eggplants exactly like you would make it in a in the traditional dish, you can literally salt the eggplant so that you can compress the cellular structure and prevent it from absorbing the excess oil. So that makes it healthier. You can let's say you can protect the polyphenols, right? So you can heat the tomatoes, you can unlock the bioavailable ⁓ lycopene. And stir in some extra virgin olive oil off the heat to prevent oxidation and so on, right? And then you can upgrade the proteins. Instead of using high fat mozzarella, you can take a casein-rich cottage cheese base and you can use some parmigiano for high impact flavor with minimal calories, right? So you can do all these variations on the eggplant parmigiana and you would make it like both super tasty and super healthy. So that was my little contribution to the food side.

speaker-2: I would love to eat that for lunch today. So yeah, that sounds delicious and nutritious.

speaker-1: I'm gonna try that out. You know, I still haven't tried it out because we went to ⁓ wedding just a few days ago. And ⁓ the other topic I wanted to mention here was a topic that I posted on LinkedIn, Momo. It's a very important topic because a couple of months ago there were a bunch of experts, academics from the field of microbiome research and microbiome science who got together, I think it's something like sixty, sixty plus of them, got together and published a what they called a consensus statement. on microbiome testing. Okay. And it's published in the Lancet Gastroenterology and Hepatology. So it's a very ⁓ visible journal and ⁓ a number of people are taking this publication very seriously. Now, what I want to point out is that they actually go through and list out the characteristics of good microbiome tests. And they say what are the clinical applications, what's the best way to use it, what's the best way to interpret it, what's the best way to explain it and so forth. And what they have concluded basically says that your traditional sixteen S based microbiome tests and metagenomics based microbiome tests are just not going to cut it because they are just looking at the taxa and they're looking at abundances of those organisms and they're just trying to put those together as if they mean something direct for a health based interpretation, which is really not true. Because the the organisms themselves are not saying anything about what's happening in the health. Instead, what makes a difference to health is what exactly they're doing, not who they are. And what they're doing depends entirely upon what they're expressing from their genome, which is not the DNA that they they have, but what is the RNA that they are expressing from their DNA. And from that, you should be able to infer a lot of functions that then you can. draw a direct line into health and sickness, right? I mean, that's more or less what they what they say. And then you up, you know, they also say things like you have to you have to explain this well to patients. You have to not overinterpret the data and so on and so forth, which are all correct. But what is going on, Momo, in summary, in my mind, is that they've described all the characteristics of the biome metatranscriptomics test without actually naming it. So it's it's amazing to me that somebody has written like a position paper in a very visible journal essentially explaining the characteristics of the Vyome test, which is what we do today and how we explain it to the world today. And if you want to see exactly why we think that they have explained the characteristics of the Vyome test, go check out my LinkedIn profile and you'll see this post in there. So it's you know, just go look at my name, Gurudad Banavar, in in LinkedIn, and you'll see. this post maybe put up a three page summary or a position on why we think the biome test satisfies all the criteria that they put out there in the consensus statement.

speaker-2: And and ⁓ one of our clinical advisors who's a gastroenterologist helped us write it as well.

speaker-1: Yes, yes. And and and y this is not just coming from a molecular scientist and a an AI scientist, but also from a clinician who practices and uses this kind of a test every single day and explains to other doctors how to use it and so on. So this is actually hitting all of the different perspectives in one shot. Yeah. So that was on my radar. And with that, I think we can get into the main body of the episode and go back to the gas production topic.

speaker-2: Super important. Let's go.

speaker-1: Let's go. Okay. In order to dive into the whole biology and the math of gas production, we've invited two experts in this field, Eric Patridge, Dr. Eric Patridge, who is a molecular scientist, and Doctor Matthew Moluski, who's also a molecular scientist. Welcome to the episode.

speaker-0: Thank you, Brian.

speaker-3: Thank you for having us.

speaker-1: Awesome. So to get started, I wanted to show something that was a background that we started off on in the previous episode. So here is the picture that I'd shown in the previous episode. Remember that when a user collects their biological samples and sends it to the lab, it goes through a series of steps that we described in the last episode in Vyome's Clear Lab. So that includes the sample prep and the sequencing, where the sequencer converts the atoms and molecules into bits and bytes, the data that is uploaded into the cloud, which is then pulled by our bioinformatics algorithms, which Dr. Lan Hu explained in the last episode, to understand all of the little reads, which are 150 nucleotide long sequences of data. Into a collection of genes and genomes. Now that is the output of the bioinformatics process. That output is still not fully interpretable because they're all little bits and pieces of the biochemistry. So in order to truly understand what's going in the biology, we need to understand the biological function. So, Momo, do you want to tell us and remind us about what biological function is?

speaker-2: Yeah, let's do that. So this is this was explained in one of our previous episodes in in in high level of detail, but let's just summarize here that basically what we're talking about here is that micronutrients and foods represent basically fuel for gut microbes to produce metabolites. And so what I'm listing here is simply two fuels. One is sulferofane, ⁓ and the microbes c can convert it to hydrogen sulfide gas. And then the other one is pectin from fruits that is converted to butyrate by certain microbes. And so this part in the middle is where we're going to spend the most time today is how do we actually quantify these conversions because the output of these biochemical functions is what affects our physiology in a very deep sense. So now I'll pass it on to Matthew to dive really deep into these functions and scores.

speaker-3: Thanks, Momo. ⁓ so as you saw as an example, so furophane is just one that can be produced into hydrogen sulfide. But if we take a look at the next slide, ⁓ we should be able to see there are multiple ways, multiple components and multiple metabolites that can be converted into sulfide. And this happens in your gut every day. And this is a pathway. These arrows The highlighted ⁓ circles are metabolites. And these metabolites get converted into intermediaries that might eventually get converted into sulfide. And so highlighted in red or orange is the actual metabolites, the entry points where your microbiome might produce sulfide. And these are shown on the periphery. But in the middle, and highlighted in green, this is sulfide, this is hydrogen sulfide, this is what we're trying to measure ⁓ in a functional way. And this is a pathway that we look at when we start designing scores. We start designing them based on what are the inputs, where are the micronutrients and other metabolites coming into this pathway. And so it's a little complex, but this is how we start building our scores.

speaker-2: So let's go over these ⁓ micronutrients a little bit, please, ⁓ Matthew. ⁓ can you explain the micronutrients and w what sources they come from? And there is taurine, which is not actually a micronutrient, we actually secrete it.

speaker-3: Right. I mean they can come from your diet. They can come from your excretions as a in in the host. There are ⁓ several amino acids that have sulfur compounds or sulfur on them that can be converted into sulfide. Taurine is another example where it can be produced by the hosts and converted into sulfide. And sulfates and sulfites, which might be derived from the diet, ⁓ can also be reduced to hydrogen sulfide. So those are the ones that we look at ⁓ quite a bit. There are other ⁓ sulfonated and other sulfur compounds ⁓ that can, you know, be in your diet ⁓ that might contribute to this as well.

speaker-2: Okay, yeah. So what we're showing here is basically on the upper right is glucosinolates from brassica family, so Brussels sprouts, broccoli, and so on. And then we're showing two amino acids, cysteine and methionine. And I want to just point out that amino acids are sort of bunched into protein, but what's actually particular here is that the ratio of cysteine to methionine is dramatically different in plant-based protein and animal-based protein. And depending on where hydrogen sulfide is produced from. in a particular person that that plant versus plot protein intake can actually make a big difference in hydrogen sulfide production. So I think we nailed that down pretty well. Yeah.

speaker-1: Actually, I have a question. I'm the non-biologist here, so I can I can ask the basic questions, right? So, first of all, every node in this graph should I think of it as an enzyme or an expressed gene that comes from some kind of a microbial process.

speaker-2: Both. It's both an expressed gene and an enzyme. So it's the same thing, yes.

speaker-0: Technically the nodes are the metabolites and the the arrows and the lines, the vectors are the enzymes.

speaker-1: Uh-huh, ⁓ huh. Thank you. That's very helpful. My next question, dumb question, is should I expect that in order for sulfide to be present or generated, that all of these molecules are present or a subset of the molecules are gonna be able to produce sulfide?

speaker-2: There we go.

speaker-3: Not every ⁓ enzyme is going to be present in each microbe. They could ⁓ be handing them off, ⁓ intermediaries, but not all of them have to be present to have sulfide production. It could just be from one ⁓ particular metabolite, depending on the microbes that ⁓ you have in your gut.

speaker-2: Yeah, I think this is a part of personalization in that so I would I would actually break down this complex picture into basically four pathways. And the four pathways lead from four micronutrients. So that would be glucosinolates on the upper right, cysteine and methionine, and then taurine. ⁓ those those are sort of the four major pathways. And so those are three micronutrients, and taurine is actually a part of our bile acid, so it's not strictly a micronutrient. And so in every person ⁓ that we would profile with our test, we would score these four separate pathways separately and understand where is this person's microbiome making hydrogen sulfide gas from. And then if that person needs less of that s hydrogen sulfide gas, we would suppress those pathways by withdrawing the micronutrients or taurine from that person. If that person needs more hydrogen sulfide, then we would look at where their microbes are already producing hydrogen sulfide and boost those pathways by providing those micronutrients. I don't know if I explained that well, but basically, if if we identify a person that whose microbiome is really good at producing hydrogen sulfide from glucosinolates, then if that person is producing too much hydrogen sulfide gas, we would our our biome recommendation engine would tell basically that person to either minimize or avoid brassica family ⁓ plants. But if that person's microbiome is producing too little hydrogen sulfide, we would actually put brassica plants into either enjoy or super food category depending on how deficient they are.

speaker-1: Uh-huh, I see. So so what you guys are saying is that by looking at the output of the bioinformatics, which gives us just a set of genes, you can then kind of organize them and you can look at their coexpression, so to speak. So when you put it all together, you can then estimate the amount of hydrogen sulfide being produced with the relationships among the various enzymes as you see them in this picture. I think I think that's the picture that you're you're you're talking about. Exactly. How how is that how is that calculated as a molecular score?

speaker-3: So this is how we ⁓ simply score things on a bioinformatic basis. We take these enzymes and these features that are wrapped in those enzymes and we develop a score from zero to one hundred. And we try to keep it as simple as possible, where at the lower end it's an at and it's an attention ⁓ to and maintain is in the middle. And then we have an optimal score on the right hand side. So it's literally trying to map how well your sulfide production pathways are behaving. ⁓ if it's in attention, you might have too much sulfide production from any given ⁓ micronutrient or pathway that feeds into that. Maintain, you want to still have this type of ⁓ functionality, but maybe a little bit improved. And then optimal. We're we've we're we're measuring things that make sense for you know balanced hydrogen sulfide production. And so we try to keep keep it simple from zero to a hundred. And that's that's sort of how we develop a ⁓ molecular score to easily transmit that knowledge to a a customer or a patient.

speaker-1: This is what a Viome customer would see in their app, correct? The the three words attention, maintain, or optimal for a single score. And of course there's many such molecular scores in the Viome app.

speaker-3: Yes, this is the the the the how we built our our scores out from the bottom up.

speaker-1: Excellent.

speaker-2: So this is basically the front end of the app or the the dashboard for the user. And I think that many people will want to know now what's under the hood.

speaker-3: Right. So this is the established what w the customer would see, but what exactly is ⁓ this score ⁓ representing? So if we go to the next slide, this is where the biochemistry ⁓ meets the math, ⁓ so to speak. This is an example of a definition or a score model ⁓ that we develop. And you can see on the right hand side of this table, these are actually the features that go into this. This is the K orthology that we use to define feature sets and in this case ⁓ a certain pathway function. And on the left hand side is where the math is. And this is how we weight ⁓ the features and develop a score out of that. So this is how we're measuring the biochemical activity of any given score.

speaker-1: Yeah, Matthew, ⁓ I wanted to ask you here that, you know, in the in the left hand side you see some that are positive and some that are negative. So how do you understand the the difference between those two those two sides?

speaker-3: So we have an idea in biology where there's a flux, right? It's not one way or or the other. It it goes in multiple directions. And so we're trying to measure a positive direction that might feed into production, but there's also a negative side that might take away from production. And so we're trying to develop a score that ⁓ represents that balance ⁓ in biology. And so positive features would be contributory and maybe negative loadings would be taking away from the flux into production of in this case hydrogen sulfide.

speaker-1: Got it. Thank you.

speaker-2: Okay, I have some questions to tie it back to the episode with Dr. Lan Hu. ⁓ we talked about KOs in that episode, but we're not mentioning really KOs here. So let's talk about KOs. ⁓ Matthew, can you pick a couple of KOs from here and explain to the audience what does that mean? Just e pick one line and tell us what that means and how do we how do we measure it in the lab and all that. Just give us your your your version of the of the explanation.

speaker-3: Right. So we can show, let's say for example, let's pick one that's ⁓ an enzyme. ⁓ a lot of these sometimes don't have to be enzymes. They can be transporters, they can be other features that are involved in the pathway that don't necessarily have an enzymatic function. But for example, K zero one nine four one, it's called urea urea carboxylase. This is a feature that has a function and it has an enzymatic function. And There are multiple organisms in your gut, your microbes, that can actually perform this enzymatic task, not just one microbe. So these KOs represent a bundle or a bin of transcripts within your gut microbiome that have the same function. And so we're really representing multitudes of microbes, not just one, that is capable of producing this enzyme and then using this enzyme for purposes like urea carboxylase.

speaker-2: Right. So we're scoring basically the the entire microbiome, not individual microbes. We're looking at the collection, the aggregate contribution.

speaker-3: And in our from our bioinformatics pipeline, we map the transcripts to these KOs ⁓ from the different microorganisms in your gut. And that's how we build a score. We want to go f ⁓ enzyme ⁓ functionality or any functionality, like transporters, for example, is another case that's not an enzyme, ⁓ that you know, contribute to a score model, production of hydrogen sulfide in this case or any ⁓ metabolite or substrate.

speaker-2: Yeah, I I mean I I think that it would also be cool when I show when I when I explain to people what a score is, I also include this polynomial equation that says it's basically what a s the the way that you calculate a score is simply it's a sum of weighted features which are listed here multiplied by their loadings or or weights. And so it's a it's a fairly straightforward model. We use machine learning. and and heavy math to come up with these loadings that are on the left side. But once we have those loadings, the ⁓ conversion from bioinformatic data to the score is very deterministic. It's not like we're using AI to do any sort of evaluation and then include any sort of AI glitches. It's really deterministic. Right, Guru?

speaker-1: It it is it is it is a linear function. I think that's the simplest way to think about it. So it is it like like you said, Momo, it is the loading times the quantified expression of each individual feature. So the quantification comes from the fact that we have detected X number of transcripts of a particular function, like this KO has is a cluster of different types of genes that come from different organisms. That Cluster together has mapped to a certain set of reads. So let's say in some cases it could be 10 reads, in some cases it could be a hundred reads or a thousand reads or whatever it is, that's the quantification for each one of these functions multiplied by the loading. I think that's the linear function that we are calculating. And there's nothing non-deterministic about a linear function. It's a linear function forever. Once it's learned, it's learned.

speaker-2: Yeah. Exactly. So let's let's just let's just sort of conclude this topic with so guru, if if you were given one data set, so we analyze a stool sample and we give you one data set and we and I ask you to compute the score for that one data file a thousand times, what would be the variance of those thousand computations?

speaker-1: Using it would be it would be zero, right? I mean it would be zero data set. For the same data set, it would be identical. So so I mean, y you know, we we actually do one we go one step further, Mama. You know, we even do some testing of our scores by looking at technical replicates of the same sample, right? And looking at the variance, and we also limit that variance across technical replica, the same biological sample to be very tight. So that we can be sure that if somebody sends the same sample multiple times, that they get the same score.

speaker-2: Yep, exactly. Perfect. All right. That was great. So now let's bring in Eric. He he's gonna he's gonna tell us about the breadth of scores and we're gonna dive deeper into some specific ones because this is all about gases today.

speaker-1: So I guess ⁓ we're gonna talk about ⁓ how gas production happens as a part of the gut health overall, right? ⁓ Eric. So please explain to us what how we conceived of this idea of gut health ⁓ broadly and then and then the the focus on on gas.

speaker-0: Sounds good. within the universe of gut health at VIOM, we think about multiple lenses at various levels. in looking into gut health through those lenses. So as you can see here, there are a few of our L3 scores which lead into gut health. ⁓ each of these L three scores are a mid-level score that sort of represents physiological concepts. That we then use subscores or L4 scores ⁓ to compute a more molecular lens, which then rolls up into the L3. So here we can see metabolic fitness, gas production, digestive efficiency, inflammatory activity, gut lighting health, and so on. And each of these scores represents a component of gut health, which our customers can relate to and which represents the physiological nature of what's going on in in the gut. And what we're going to do today is open up gas production a bit more and take a look at some of those functions and the molecular aspects that we're trying to capture in the world and the universe as biome sees gut health.

speaker-1: Yeah, Eric, ⁓ let me ask you let me ask you a question, right? So this ⁓ classification that we've come up with for gut health, is this based on sort of the general understanding of gut health that's out in the literature and in the in the field at this moment? Or are some of these things actually discovered from the data? Or is it some combination of the two?

speaker-0: I would say it's a combination of the two. ⁓ some of the L threes that we're taking a look at here are intended to represent a lot of what's available in the literature. The customers can go to the store and see some ⁓ supplements or whatnot that they wanna help to serve this area. They also represent giant aspects or swaths of domain knowledge or clinical knowledge that's in the literature. Gas production is a great example because of bloating. Bloating is a very common symptom. ⁓ a lot of people can relate to. And then within each of the scores, the ⁓ scores that are selected to roll into, for example, gas production are equally things that it's a combination of things that you'll see when you are walking around in the world or when you're reading very specific domain literature. So it's a combination across the board.

speaker-1: That's ⁓ that's great. That's kind of ⁓ I think it's kind of the best of both worlds, so to speak, right? We we take the knowledge that's out there, we also validate it, we also see what's available in the data, and we discover things sometimes that can augment what's in the literature, and we put all of that stuff together in these scores. The the other thing I wanted to just ⁓ highlight here also is that when microbiome tests in general talk about gut health. Many times they simply look at a list of taxa, like organisms, and simply say presence or absence, or maybe even some quantified levels of the abundance of those taxa are representative of gut health. But what we in biome think is that that is exactly the wrong way to go, because you cannot tell the health of an individual by just looking at the taxa, but you can by looking at the functions which Many taxa, by the way, share, right? You cannot, you know, some the same taxon can be sometimes beneficial and sometimes harmful. So you want to see what exactly is going on in the gut at at any given point in time, not just who is there. And this picture kind of represents the what is exactly going on in the gut, which is the functional lens for it. So with that, ⁓ please explain to us what are the components of the gas production ⁓ score, the functional score.

speaker-0: Perfect. ⁓ I'll suggest we move into the next slide for this. So here we have a number of areas that are captured in the subscores of our gas production L3. Each of these represents a score in themselves, and these are the molecular scores at the L4 level. The three primary components that are going into the gas production score at the moment include methanogenesis, sulfide, and ammonia. ⁓ we also do capture the microbial diversity of the sample, and it turns out this is extremely relevant for the gas production score as it is closely linked as well to the Bristol stool score and the consistency of the stool. So, in this particular ⁓ set of scores that are rolled into gas production, so methanogenesis is dedicated to the production of methane. And when the production of methane is high, there is an overall volumetric conversion. from other gases into carbon ⁓ into methane, and this conversion helps to slow down gut transit and contributes to constipation. In the case of sulfide production, we're measuring or setting out to assess the accumulation of hydrogen sulfide gas, which can disrupt metabolic processes and also damage tissue and contribute to diarrhea. It can be an indication of the metabolism of cyclic acids and decorated mucins within the gut lining, which can be a source of the hydrogen sulfide and be an indicator of damage. When it comes to ammonia production, we are measuring the release of ammonia during the metabolism of proteins and metabolites. And so this is a direct measure of or this is an approximation of the microbial activity related to the release of ammonia. And this is a complex interaction between gut motility and ammonia. And as the ammonia levels rise during constipation, for example, this can also result in protein fermentation while gut motility is decreased. So this particular score is quite complex. And then active microbial diversity does affect the ability for microbes to grow and diversify. And it is impacted by both bi diarrhea and constipation. ⁓ and in this particular case. we look at it and it is extremely significant contributory vac factor to the gas production scores that we have at the L three level.

speaker-2: I have some questions, Eric. So, ⁓ or comments, I guess. So is it fair to say that this ⁓ production of methane is sort of like a break on the gut motility, meaning the more methane there is, the slower it is. And then hydrogen sulfide is like a gas betal for the motility. The more there is, the faster the poop goes through the intestines.

speaker-0: So I think we've designed the scores with those in mind. ⁓ whether or not that's always true is a question, but what you said is accurate for how we've intended to design the scores. Of course, you could imagine scenarios where you just have methanogenesis up the wazoo and if you have enough gas, you're gonna, you know, ex have some explosive things going on. But in general, we have designed the scores to ⁓ align with how you couched it quite well. Increased methanogenesis will correlate with slower gut transit, and increased sulfide production will correlate with higher gut transit.

speaker-2: Awesome. And ⁓ in general, the source, so that fuel that we talked about for ammonia production, is it in general the amino acids from the ⁓ from the protein that we consume?

speaker-0: Yes, that's the common thought across the domain is that the ammonia production within the gut is largely coming from fermentation of proteins.

speaker-2: Okay, so I guess every person would have a certain capacity to digest proteins and in healthy states that's a lot more than in some state where their digestion is out of balance. ⁓ and so if if sufficient protein reaches the colon, then it'll get fermented and produce ammonia. Great.

speaker-0: That's right.

speaker-1: All right. One more question from me ⁓ is of in terms of the microbial diversity, ⁓ there's a lot of ⁓ literature about microbial diversity in the microbiome field, right? People have a lot of ⁓ I would call it preconceptions of what diversity is supposed to be good for or not good for, right? In the case of gas production and gut motility, what is the relationship between microbial diversity and motility?

speaker-0: So, this is a really complex and challenging aspect to answer, and also to transfer this to the audience. So, what I would urge in this particular case is to think about the stool in a chemical sense, but also in a sense that is reminiscent of cement or water flowing. And what you can imagine is that when you have A flow that is active or faster that is also growing things in it, the faster that material is moving is going to equate to a slower amount of time for microbes to grow. So the faster you have a flow, there's going to be less diversity present simply because you have less time for the microbes to grow as they transit. The nine meters or so of intestinal space that we have through us, right? So that for me is the most reachable ⁓ knowledge in this area that helps to explain the active microbial diversity and why it's so related to methanogenesis and sulfide production. Another aspect of that could be considered in terms of methanogenesis, is going to be coming from a lot of anaerobes. And you need to have a certain kind of environment to support. a very low oxygen environment where those methanogens are going to be present. And so you might have a little more complex chemistry happening when you have a slower flow as well. So it could be a combination of factors here.

speaker-1: Got it. Got it. No, but but your explanation makes sense intuitively. So yes. Thank you.

speaker-2: All right, what's next?

speaker-0: I guess we can think we

speaker-1: to ⁓ dive into methanogenesis and talk about the functions.

speaker-0: So I think ⁓ methanogenesis pathways is near and dear to my heart, having done my graduate work in a meth meth ⁓ methanogen lab. What we have done here is represented the key aspects of methanogenesis pathways in a slide set. as you can see, we are latching on to a couple of comorbidities that we have illustrated on the left-hand side of this slide. Diarrhea and IBSD, as well as Crohn's disease. We also have ⁓ on the right hand side summarized a number of the key components and key functions that we have considered in designing this score. In the center, we've also highlighted in the center yellow circle a number of the very key metabolite and enzymatic aspects of methanogenesis in general. And I I guess I'll start there because these are really the key components that people are going to latch on to and say, hey, we're trying to figure out what is this score about. Well, it's certainly about methane, and that's represented in the center by the carbon surrounded by four hydrogens. We also have ⁓ coenzyme ⁓ Flavin F4 twenty, acetate, formate, trimethylamine, and methyl transferase. So those experts who are working in the field of methanogenesis will be able to look at this and understand that this is the wheelhouse of methanogenesis and methane production. Now, when we think about how to measure this within the gut and how to assess The level of methanogenesis that is happening, we can't just look at these key components and functions that are highlighted in the center. We have to take a much larger swath to try and ex examine this production or the production of methane with respect to a lot of other things that are happening in the gut. And so, yes, when it comes to cofactors that are specific to methanogenesis, we are looking at coenzyme ⁓ coenzyme B, F420. F430, methanoterin, methanopurins, and so on. These are all the central aspects of methanogenesis, but you can't get methane without having carbon coming from somewhere. So we're also looking at small carbon molecules like short chain fatty acids. We're looking at the source of carbon coming from methylamines, from formic acid. From carbon monoxide, glycerol, beta, serine, fructose. Now, these sugars like fructose also have to have sources of sugars. So we are also looking at larger polysaccharides like leaven, cellulose, starch, menon, larger fructans. And then when you're talking about methylamines and related compounds that might have nitrogen in them. We're also looking at nucleotides like purine biosynthesis. ⁓ other molecules that we are taking a look at include serotonin, because this is related to some aspects of gut transit and the stimulation of metabolism that's happening in the gut. In taking a closer look at the key functions, a few things that I haven't mentioned that we are also taking into account for the methanogenesis score. include alternative energy pathways. So the energy is going to be coming from other places, including hydrogen. ⁓ we are looking at the management of pH and anaerobic waste at the same time. a couple of things related to antibiotic metabolism, chaperones, some chemotaxis going on. And finally, perhaps one of the most important aspects of this is electron transfer. Because You can have all of the compounds you want, but while without electrons flowing from one place to another, either through the carbon atoms or through the membranes and the electron ⁓ acceptors, you're not going to get the reduction of small molecules such and the production of bethate. So I think that pretty well covers a lot of the aspects that we're considering in methanogenesis pathways. And we will incorporate other components that are not necessarily mentioned here, but by and large, this is a good representation of the things we're looking at for the score.

speaker-1: So all of these ⁓ Eric are are represented in some KO or another, right? Or sometimes maybe multiple KOs. That's sort of the universe of ⁓ functions that we start with and then we or components that we start with, and then we you know, we find the models that capture the balance between the ⁓ phenotypes that we are trying to also ⁓ reflect through this score.

speaker-0: Yes, absolutely. So the long list of KOs that Matthew presented earlier is a good representation of the number of KOs that would be in the score for methanogenesis pathways as well. And we are specifically selecting features that represent all the key components and key functions that I mentioned. And sometimes they represent one or more of these, and sometimes multiple KOs will represent some of these, right? So one KO can represent several of these. And as we design the scores with these concepts in mind, we are also, I think, as you pointed out, keeping in mind the comorbidities, diarrhea, IBSD, Crohn's disease. And we use case control studies to to make sure we're selecting not only those KOs that are representing the key components, key functions. But that are also making sure to give us a large differential across the comorbidities so it's representative of reality and and our general cohort population.

speaker-1: Excellent. And that's captured I think in some of the publications that we put out there. I think we'll talk about one of them later on in this show. So do you want to talk about ⁓ sulfide next?

speaker-3: So on the opposite end of things we have hydrogen sulfide, ⁓ where we actually would get some similar comorbidities, but in the opposite direction. So diarrhea, IBSD would be ⁓ some of them, as well as Crohn's disease.

speaker-2: Also, Matthew, opposite on the scale of ⁓ smelliness from methane.

speaker-3: ⁓ I don't know. I mean I guess, you know, choose your poison, right? ⁓ whichever one ⁓

speaker-2: I think methane is odorless. Correct. Whereas whereas sulfide is definitely not odorless.

speaker-3: Yeah, I guess ⁓ I guess ⁓ cooking so much with gas has just allowed me to smell as methane, but I know they add s

speaker-2: They add so I think they add sulfur compounds actually to methane to make it smell something.

speaker-3: Yeah. So I guess sulfide done, right? Instead of methane. So, okay, right. ⁓ when we think about the components, as Eric talked about before, the main components, ⁓ we can think of hydrogen sulfide obviously in the middle here. We can think of ⁓ phylosulfate dehydrogenases, other enzymes that might be important for liberating sulfide or sulfur from other compounds. ⁓ for example, ⁓ cysteine, which is amino acid like we spoke spoke about before, ⁓ in ferrodoxins and taurine as well. So these are the main components that we would want to include, but extending out farther to what the score key components and key functions might be. So for example, sulfur metabolites, sulfate, sulfite, which are oxidized forms of sulfide that might be reduced by the gut microbiome, sulfocarbons, ⁓ thiocarbons, ⁓ thiosulfates, ⁓ and several others, ⁓ sulfur amino acids, including taurine and cysteine. Reductive enzymes, including ferrodoxin, cytochrome, these are like ⁓ that would ⁓ reduce sulfate and sulfite back to hydrogen sulfide. So that's that's what we're thinking about here. Others include ⁓ nucleotide backbones, adenine ⁓ phosphosulfate, ⁓ things around nucleotide metabolism. And so key functions then would be around sulfur reduction, sulfur cleavage, amino acid metabolism, ⁓ catabolism, and anabolism. We would want to know if they're actually making more cysteine or methionine. versus catabolizing it, which would be breaking it down. ⁓ transfer transferases and transporters, as I sort of spoke about earlier. How do ⁓ these compounds get into the microbe itself? So these would be transporters that would be specific for ⁓ powering transport, for example, and other transferases that would be moving sulfur around, like sulfide adenyl transferase, phosphate acyl transferase included in that. And again, like Eric said, electron transfer, because we are playing with carbon molecules here and how electron transport would affect certain components of sulfur reduction is quite important, especially since ⁓ you need electron trans transfer to ⁓ reduce oxidated compounds like sulfate and sulfite. So overall that's like the general ⁓ concepts that were incorporated into this for and it's again still based on comorbidities that we find interesting, including I B S D, Crohn's disease, and diarrhea in general.

speaker-0: I I would also add to this because if I was a methanogenesis expert taking a look at the score, there are similar enzymes that are going to overlap between scores. And I think it's relevant to say that we try to make sure to minimize the overlap between any two scores. So if people are looking at this and they see ferroxins and they're wondering, well, that's also relevant for methanogenesis, you know, are you just measuring the same things for both scores? The answer is no. We actually take very special care to pick as few overlapping enzymes as possible. While feradoxin, for example, is relevant to multiple scores, because of how we choose each feature, the set of features that we've chosen are really the best ones to represent sulfide production pathways, for example, in the context that we've designed it.

speaker-2: But I want to also balance that point, Eric, with the fact that some of these scores do have overlapping pathways that compete against each other. And so it is important to include some of those features. So for example, hydrogen is a source that is necessary for both hydrogen sulfide production and for methane production. And so you cannot just exclude that because the production of hydrogen is important. It's like a fuel basically for both, right?

speaker-0: Right. But we can also but there are also tens, if not hundreds, of features that are going to be related to hydrogen and sulfide, and we don't have to choose every one for every score. So we make sure that to minimize overlaps as much as possible.

speaker-2: Yeah, no, that's fantastic. I wanted to comment that I just want to zoom out of this whole discussion that we're having here and just convey to the audience the the number of scientific discoveries that have been made over the last few decades that include next generation sequencing and bioinformatics and microbial physiology that can actually interpret the function of a microbial gene and then the connection of these metabolites that that we're computing. The connection of those metabolites to the human physiology and then putting it all together using super smart people like Matthew and Eric and many others at VIOM and elsewhere enables us to actually bring stool to a laboratory that just it it doesn't tell you anything about like when you see it with your eyes, it doesn't tell you anything. But what's what's going on under the hood is that we're doing these phenomenally deep measurements and then interpreting those data in a very very quantitative way. To tell people something about their biology that they could never otherwise know, it's it's simply an incredible amount of of human knowledge that I I want people to appreciate.

speaker-1: Yeah, Momo and and along the same lines I also marvel at the fact that when someone goes to a laboratory and gets a typical blood panel or something, you know, they measure a hundred or, you know, twenty markers or something, and they think that they're they understand a significant portion of biology, which I mean they do understand some levels of it, but the point that our audience should take away from this is that

speaker-2: Fifty or twenty markers.

speaker-1: Each one of these scores represents many, many genes. Sometimes when you look at all the clusters of KOs and everything else that we're talking about, it represents hundreds of genes, but they are aggregated into these KOs, which could be dozens of KOs in a given score. And then there are hundreds of these scores. So when you when you look at all of these total amounts of biology that we are capturing by Providing these scores and insights in the Biome app, we are literally talking about thousands of biological, well-understood features that are interpretable and that they have been put together in a form that computationally makes sense, algorithmically makes sense. That is a very comprehensive view of biology that you don't get anywhere else.

speaker-2: Yeah, we marvel at that every single day. I would say it's not thousands. I would say it's definitely more than ten thousand molecules or biomarkers that we measure. And if you if you consider sort of the the the best number of biomarkers you can get anywhere from any doctor today is about a hundred, we are we've elevated the game by at least a hundredfold. And that's really where Viom shines is that we've made this quantum leap in measuring human biology and how it relates to health. But also tying it back to nutrition, because if you get high APOB today on your blood test, they'll tell you to eat healthy. That's that's like l almost meaningless. You don't know what to actually eat and what to avoid. Whereas when we tell you that some part of your physiology is out of balance, it ties back strictly to very specific nutritional recommendations as to how to adjust that part of your physiology. And that's really another quantum leap. So I'm obviously hyper excited about what we're doing.

speaker-1: Yeah, I mean i if you want to just talk numbers, right? I mean the numbers we we ⁓ analyze, we essentially look for a hundred million possible genes in every sample. And then across the entire population of Vyome customers, we have so far detected about ten million microbial genes.

speaker-2: The expression of ten million microbial genes.

speaker-1: Of 10 million microbial genes. And if you look at any one particular kit, meaning three samples, right, we are talking about hundreds of thousands of expressed genes that we detect. And then we aggregate that into these functions and clusters of functions that are annotated. That brings us down to the ten thousand level that we were talking about earlier. So it's really a huge funnel that we try to make it more interpretable and make it more I would call it tractable is the technical term that I would use, but digestible is a is a beautiful way to say it. Yes. So that everyone can understand what's going on. So going back to the functional description here, Eric, did you have any other functions that you want to talk about or Matt?

speaker-3: I think we're gonna move on to ammonia as our last one. All right. ⁓ again, these are all gas production scores, so we would want to think about transit in the gut of material, including ⁓ constipation and diarrhea. ⁓ ammonia is a backbone molecule in amino acid ⁓ linkages and so digestion of ⁓ protein ⁓ or other ⁓ compounds that ha and potentially can't be ⁓ contain ammonia, including urea. These are all important and ⁓ central to the design of this score. ⁓ and so when we think about key components, we wanna think about nitrogen in general as well because a lot of microbes will use nitrogen for different ⁓ purposes. ⁓ urea and purines are an example of nit nitrogen containing compounds. Amino acids, I mean ⁓ arginine, have side chains that have ⁓ extra amine am ammonia groups, ⁓ histidine, glutamine, asperrogine, lysine. We wanna think about transcriptional regulators that Utilize polysaccharides, other things that might be important for growth ⁓ bacteria ⁓ in the gut. Carbon electron sources, including ⁓ beta glucosidases, formate amino acids, like we said before, but other things ⁓ that ⁓ the microbes use, including osmoprotectants, so against osmolarity or different changes in concentrations of liquid or water in the gut. And then key functions again, nitrogen metabolism, amino acid metabolism. carbon metabolism because carbons are a source of ⁓ the amino acids themselves. ⁓ and again, we always want to think about ⁓ which metabolites are feeding into this and what they're contributing to.

speaker-1: What's

speaker-2: Awesome. So I also read that ⁓ urea, which is actually produced by our liver, actually can physically leak from the blood into the gut and it it's it's a substrate for ammonia production, is that right?

speaker-3: It's it can d urea can definitely be ⁓ producing ammonia. That's for that's

speaker-2: Right. So just want to tie it to the protein consumption because protein is just such a big hype today that ⁓ if you overeat protein, then some of it may directly go into the colon and produce ammonia, which too much ammonia is bad. The second is even if you're able to absorb all protein, if you do not use that protein, then your liver will basically ⁓ degrade proteins, remaining amino acids into urea. And that urea will end up in the colon where the microbes are going to make ammonia from it. So one way or the other, from the dietary protein, if you do not end up consuming it, then you can make pro-inflammatory compounds. And I I just want people to understand that when they say you need to be consuming a lot of protein, what they're really referring to is bodybuilders, people who are trying to build a lot of muscle in a very short time. And that means they're also in the gym multiple times per week. If you're one of those people who is not in the gym all the time and you're c over consuming protein, you may actually be causing inflammation without knowing it. And this is one of the ways that that happens.

speaker-3: Yeah, and that's one of the benefits of biome. We we know ⁓ can personalize what you're doing and how much you should be eating.

speaker-2: That's right. That's right. So this ammonia score is very important and the protein fermentation score, which we are not talking about here, but we have those two scores that can tell people are they actually consuming too much protein?

speaker-1: Yeah, okay. So I have a really important question at this point is that yes, we've built all these scores. How do we know that they are actually representing the disease biology that we were hoping to capture? Let's go. Who can ⁓ who can tell us that?

speaker-2: Let's go.

speaker-0: So I think w what you're asking is what do we do after we design the scores to examine how they are holding up? And this is a really important question because we do design the scores on a reference cohort. Currently we're designing scores on a twenty two thousand person reference cohort. And as part of the process, we do use clinical data to drive some of the truth in these, but we're not training on that data. Now, after we finish designing a score, we examine their functionality on separate cohort. So we're going to be looking at, for example, a validation cohort. And our current validation cohort has around eight to ten thousand individuals. Now, what we're looking at here are some of the odds ratios that we would be generating on a validation cohort, for example. And what we're able to do, if you think about the thresholds that Matthew described earlier, where you have one side for a good score in another side for a needs improvement or a not optimal score. I think the alternate verbiage was maintain and attention. If you take those individuals and break it down for people who have, for example, IBS D and IBS C on this particular slide set, you're able to say those who have a not optimal score, for example, ammonia production pathways, have an increased risk for not optimal Hey Eric. Now yeah.

speaker-2: Do you mind explaining odds ratio to those people who are not familiar?

speaker-0: We can do that. I feel like we should have the math there.

speaker-2: Okay, but just conceptually. Yeah, conceptually. Go ahead, Guru.

speaker-1: Yeah. So the idea of odds ratios is that when you have an exposure, like a, you know, in the general case, you can say, you know, someone's who's smoking, right? If someone's smoking, right, that's an exposure. You ask the question, what are the odds that you would have a particular disease compared to people who do not have that exposure? Simple.

speaker-2: Love it. So simple.

speaker-1: way of thinking about it, right? That's the ratio that you're trying to measure. Yeah. So here what we are trying to see is if somebody has an exposure of ammonia production pathways being in the attention zone, which is not optimal, what are the odds that they would have either IBS constipation or IBS diarrhea relative to people who do not have that exposure of having the ammonia pathways being not optimal. That's The simple way of thinking about it.

speaker-2: And what are what are bad odds ratios, Guru? What are what are like fine odds ratios and what are like the ones that people be sh should be concerned about?

speaker-1: I think in general epidemia epidemiological literature, anything that is one and a half or more is taken seriously in terms of odds ratios. When it goes beyond two or three, it's considered a very, very significant effect. And you want to pay a lot of attention to it. And and again, remember, in this scenario, we are looking at so many different functions. So they would be a contribution from multiple functions for a given disease. So

speaker-2: It's all additive. ⁓

speaker-1: diseases, then it becomes additive that you have multiple functions contributing a little bit to the same disease phenotype, and that aggregation might actually take the odds to be, you know, three, four, five. In fact, our in our product, we only talk about diseases that have an odds rare cumulative odds ratio, C O R, of more than five, which is considered very significant.

speaker-2: Very high, yes. Yeah, people can o people also hear, for example, like popular press that some some lifestyle ⁓ has like a th increases the odds of some disease by thirty percent. That would only be odds ratio of one point three. So one point five doesn't sound like a lot, but that's already a fifty percent increase in relative risk to that disease.

speaker-1: Exactly right. And and when you get to two, right, odds ratio of two, that's a hundred percent increase. Increase, yeah. So so that's why it becomes super significant in the epidemiological literature. So back to you, Eric. So ⁓ please explain the specific ⁓ pathways here and their odds ratios.

speaker-0: Sure. And the context for odds ratios, as you've explained it, is quite helpful. And back to the original question in terms of validating these scores on an external cohort. ⁓ so we are we do design on the reference cohort and what we're taking a look at here, for example, might be odds ratios on our validation cohort that demonstrate the signals retained beyond our initial training set. Now, in the two comorbidities that are presented on this slide, IBS C for IBS constipation type and IBS D for IBS diarrhea type, we're able to use these odds ratios to also help to differentiate the two. So we take a look at which scores are have a higher odds ratios for each of the four scores, we're able to see that there are big differences between these two IBS subtypes. So the two of the primary scores that are relevant. To IBS C and IBSD are methanogenesis and sulfide production pathways, which Matthew and I covered. You see on IBS constipation type, the odds ratios for a not optimal or attention methanogenesis pathway score as compared to good or maintain is significantly increased by about 1.6 in the case of IBS C. But when you look at the same score for IBSD, there's no significant significant difference from good. With respect to sulfide production pathways, we actually see a lower odds ratios, which means an improvement for IBS C. So IBS constipation type has a very low sulfide production pathway score relative to good. And when you look at IBS D for diarrhea, the sulfide production pathway score has an odds ratio closer to two. Now, of course, we can extend this to ammonia production pathways and microbial diversity, but we're gonna see the same trends that we're talking about now, right? So the IBS C has an increased odds ratio for ammonia. IBS D has non-significant difference from good. And for both of these scores, there is an increase in microbial diversity where ⁓ the IBS diarrhea has almost an odds ratios of of three, where IBS constipation type is between one and one point five.

speaker-1: Sounds good. Sounds good. Thank you.

speaker-2: ⁓ I would like to expand a little bit on what on what Guru said earlier that these are additives. So I just want people to understand that IBS and all chronic diseases and cancers, they're not like infectious diseases where one virus equals the disease. It's not that simple. It's many, many different factors. And in some people they're dominated by a certain subset of factors, and in other people by a different set of s factors. So there's different root causes of these conditions. And what we're showing here are individual pieces of physiology. that are predictive of these conditions, but there are many of them. And then when we get to an episode where we talk about disease risk stratification, we're gonna put them all together and show you many that make up then a much bigger impact on the disease.

speaker-1: Exactly. One way I think about it is also that we are showing ⁓ root cause diagnostics in some way, right? Yep. In other words, we're not telling you that you have a specific disease because that disease may have may not yet have manifested itself as an external symptom in your case, but the functions that are all driving that disease.

speaker-2: That's a that was really important.

speaker-1: potentially are already active in you. And that's generally the case for most chronic diseases. Like we've discussed in the past episodes, diabetes starts 10 years before somebody gets diagnosed as a diabetic because they have 6.5 or more HBA1C. There's a lot of other factors that have led up to that point. So these pathways that we are talking about here are the early markers of disease function that if you don't pay attention to that they could all combine and they could get worse and could lead to one of these diseases. So that's the core value proposition of biome is that we want to give you early detection of these kinds of functional biomarkers that can help you change your lifestyle so you prevent disease. And our main focus in biome so this has been published in a paper and here is the paper. You see all of us as co-authors on the paper, we have all contributed. Eric Has led this paper to ⁓ explain all of the eight pathways that ⁓ you see on the right hand side in figure four. Those eight pathways have been explained in great detail. You'll also see how the disease cohorts were extracted, validated, and also are called substratified to make sure that things are the way we expected them to be.

speaker-2: Yeah, so this paper basically opens the hood and shows to people how we generate these scores, how we validate the scores, and it discloses one of the scores which we shared with you earlier. It discloses literally the structure of that score at the biochemical level and mathematical level to the tiniest possible detail. There's no additional information that could possibly be revealed for that score. And that's really nice that we've opened up our books for at least some examples. So that p people can see basically what's going on at VIOM we of course scaled this to hundreds of scores, but we opened the books on one of them for gut microbiome and then in another paper for oral microbiome. And that's really important because we've had lots of critics from the academic community where they say what biome is claiming to be able to do that doesn't exist. Well, that doesn't exist because they are not using the right technology. They're using DNA sequencing. And DNA sequencing cannot do anything that we just showed you today. And so they are not able to do it. But just because they are not able to do it because they don't have access to the data, that doesn't mean that it cannot be done. And we are here proving that it can be done. That it is done, that this is what we're doing, and look at the numbers here. I mean, seventy-one thousand people were used in an independent validation. That means blinded validation. I mean, to me, that's an extraordinary thing that we really need to highlight.

speaker-1: Absolutely. And we're gonna we're gonna have other episodes in which we're gonna get get into the disease biology with the disease risk scores and how they relate back into these pathways and so forth. So that's for another time. But for today, I think ⁓ we've had ⁓ tremendous discussion around the biochemistry and the ⁓ functional analysis of ⁓ the scores that underlie gas production. So Thank you very much, Eric and Matthew, for joining us. And ⁓ we'll let you go now and ⁓ more on I will pick up on the rest of the topics for this episode.

speaker-2: Yeah. Thank you. Thank you, Eric and Matthew, for doing so many amazing things every day. You guys are so brilliant and ⁓ thank you for joining our podcast.

speaker-3: Yeah.

speaker-0: Thank you for having us.

speaker-1: Thank you. Bye.

speaker-2: Bye.

speaker-1: Okay, Mama, that was a great discussion with ⁓ Eric and Matt today. So let's ⁓ get to the closing of our episode today. And ⁓ we are going to discuss a question from an audience member. And here is ⁓ Julie Herman, 2841 user, who asked a question on episode four, which was the RNA versus DNA episode that we made. And she says, you talked about E. coli being good and bad. I understand that there are different strains of E. coli. And I wondered if you could talk a little bit about if the bad strains are also good. I lost my mother to stomach cancer. She battled an ulcer the year before she died, and I came to understand that there are specific strains that go on to cancer. I would love your thoughts. Thank you. That was a question, Momo. So how would you respond?

speaker-2: Yeah, let's dive into this. So E. coli is kind of a special case in that the microbiologists actually assigned so many different kinds of bacteria to be strains of E. coli, where very commonly two strains of E. coli, typically two strains of E. coli would be more different from each other than two other species. That's how different the strains are. So that's one little nuance that I want people to understand. But let's now make it generic and and it applies to E. coli strains. So I talked in a previous episode about five functions that E. coli can perform. Two of them were harmful to us, and three of them were beneficial. That was just a summary. There are many additional functions that E. coli performs. We talked about an average microbe having 3,000 functions encoded in their genome. So, how does this relate to strains? So, if we were to take a hundred strains out there of E. coli, we would find that most of them have. Most of these functions. Some of them have all of these functions encoded in their genome. And some of them are missing quite a few functions. And so, for example, there is a strain of E. coli that is actually sold as a probiotic. So this is a strain called E. coli nissle 1917. And it's actually sold as a probiotic because it has been shown in clinical trials to have benefits. Now, I did not look into the genome of that specific strain to look into. Does it actually produce the toxin coliobactin? Well, what I did find is that the lipopolysaccharide that nissle strain produces, it actually has is one of the beneficial LPSs. So we didn't dive this deep deep into this, but basically lipopolysaccharide is normally known as a bad player, pro-inflammatory molecule. And in general, that is, the vast majority of them are pro-inflammatory, but there are actually Types of lipopolysaccharides that are still lipopolysaccharides but they actually are anti inflammatory.

speaker-1: ⁓ We're always an exception in biology too.

speaker-2: There's there are so many exceptions exactly. And so it turns out that the type of lipopolysaccharide that nissle strain produces is actually mildly inflamm anti-inflammatory, at least protective. So that ⁓ yeah. So okay, so how do beneficial strains of E. coli protect us against harmful strains? And so that's one question. And so beneficial strains can protect us from harmful strains by first ⁓ producing these blocking lipopolysaccharides. Second, they can be consuming oxygen in the colon, preventing other pathogens. And third, they can be consuming iron that prevents pathogens from setting in. And specifically when we're talking about E. coli pathogenic strains, beneficial E. coli strains are going to occupy the same niche and consume the same metabolites and the same micronutrients. And so they're going to crowd out the pathogenic strains. But the second part is that I believe that The vast majority of E. coli strains can be beneficial or harmful at the exact same time. And this is what we discussed last time. And the only way you can tune that, meaning the only way you can say, I want this strain to work for me and not against me, is with nutrition. So you have to first understand everything that we discussed today with scores. How do we understand the scores and quantify scores? And then how do we relate that to micronutrients and then tell someone exactly what they need to eat? To stimulate those beneficial functions and what foods they need to avoid in order to slow down or stop those detrimental functions. That's really what it is. So there's kind of two different sides of the story. And this is applicable to bacteroides and clostridium and many other types of bacteria where you have the same genome encodes for harmful genes and beneficial genes. And the only way you can tune them is with nutrition. You don't want to just kill them all because. Most of them are both harmful and beneficial at the same time. Hopefully that answers Julie's question. And if it doesn't, or if she has follow-up questions, just please post them and we will answer them. And that is the end of this episode. That's all we have for today, folks. I am Momo, a biochemist.

speaker-1: Guru, an AI scientist were two PhDs on a pod.

speaker-3: Two PhDs on a pod is a Viome Health Science Podcast production, written and hosted by Viome CTO Guru Banavar and Viome CSO Momo Vuyisic. Produced and edited by Lisa Shomo. Podcast art by Kevin Wu. If you enjoyed today's episode, like and subscribe for new episodes every other week. We'll see you next time.