AI & Computational Biology With Recursion's Chris Gibson, Ph.D. Business Of Biotech ยท 26 Dec 2022 Source: https://www.buzzsprout.com/1011661/episodes/11910761-ai-computational-biology-with-recursion-s-chris-gibson-ph-d.mp3 Transcript: https://podcasts.opennash.com/shows/business-of-biotech/ai-computational-biology-with-recursions-chris-gibson-ph-d-3ef739/ (via OpenNash Podcasts, machine-generated) Hey Annie, guess what? What? We just launched a Business of Biotech newsletter. Yeah? Yeah. I know what you're thinking. What am I thinking? We don't need another f***ing newsletter. Yeah, I might have been thinking that. Annie, I swear on my grandpa's grave, you're going to like this newsletter. That's a pretty bold swear, Matt. Hear me out. It's monthly. Only once a month. It's ad-free. And it's modeled after the Business of Biotech podcast. It's got the same insight from the builders of biotech that you see in the podcast. So what's not to like? That actually sounds like it doesn't suck. Pretty high praise, Annie. Check it out. Bioprocessonline.com backslash B-O -B. Go there and sign up for this newsletter. You won't regret it. As early clinical stage biopharma companies go, Recursion Pharmaceuticals has been a bright, if not steady spot on an otherwise drab mid-cap biotech market scene. The tech-heavy drug developer has built a pipeline of oncology and rare disease candidates that's broader and deeper than most in its class, and it's done so more efficiently than most as well, in large part by applying advanced tech tools to big data by way of a proprietary platform it calls its recursion operating system. Machine learning, artificial intelligence, and even computational biology are just now beginning to make headway in bio and particularly among emerging biotechs. But recursion has been in that game since its inception back in 2013. Here to share stories about the pioneering effort is a guy I've been trying to get on this show for like six or eight months now, and to whom I'm incredibly grateful for the time now, Dr. Chris Gibson, co-founder and CEO of the company. Dr. Gibson, welcome to the show. Thanks for having me. I'm glad we were able to make the time work. Yeah. If you could just clarify for our audience, it's not like you were trying to avoid me or anything, right? No, not at all. It's been a pretty busy last six months for us. We've actually kicked off four clinical trials in the last three quarters with a relatively small clinical development team. And so I've just had my head down executing. Understandable. As I mentioned in that rambling intro. I mean, the pipeline there is broad and deep. And when you bite that much off, you're going to have to live up to the promise and run three clinical trials in as many quarters, I guess, right? Through four, as it were. Yeah. Yeah, it's been very busy, but it's what we're here to do. We're here to try and not only take drugs forward to patients that could hopefully have an impact on them, but also to try and change the way that drugs are discovered and developed with a long-term goal of trying to do it at a higher scale and for less cost, which I think would be a fantastic outcome for everybody. Yeah. You know, usually when I conduct these interviews for the pod, I start out with a bunch of getting to know you questions for my guest and then kind of segue into the Genesis story of the company that my guest represents. I want to kind of dovetail those two things. And correct me if I'm wrong here. I figured we would dovetail those two things because Because for as long as I've been following recursion and as far as what I know about you is concerned, I get the sense that in a sense, I mean, not that Chris Gibson is synonymous with recursion, but it's hard to tell where one starts and the other ends. You know what I mean? Like you guys are โ€“ it's core to your fabric. It's woven into your fabric as a professional. So I'm going to let you tell, just tell us the genesis story of recursion and kind of weave through that, like how you came to co-found it personally. Yeah, absolutely. So I came to the University of Utah in 2009 for an MD-PhD, and I was lucky enough to join the laboratory of a guy named Dean Lee. And some folks who listen to the show may have heard that name before. Dean is now the president of Merck Research Labs. He's one of my co-founders. He's on the board of Recursion still today, and an incredibly translational, thoughtful physician scientist who I had the pleasure of working with for five years. And it was an exceptional time. And what I loved most about my time in Dean's lab is that he filled that lab with people from incredibly diverse backgrounds. So you had cellular biologists, molecular biologists, chemists, physicians, myself as an engineer. There were, you know, there were folks from kind of every technical field that all descended in that lab. And it was a bit of chaos. But at the same time, it was also beautiful to operate at the intersection of all these different technical fields and to appreciate how much you can learn from each. And this was roughly when, Chris? This is 2009 when I joined the lab. I mean, that's pretty early days to be mashing, you know, kind of doing a synthesis of those titles in the same space to an end, right? It was, it was. And it was a ton of fun. And, you know, Dean, we had five R01 grants, I think, when I joined the lab, four or five. So it was a big lab and a ton of energy, ton of excitement working on a bunch of different areas. But the area I focused on most heavily was trying to use engineering approaches to understand questions in biology that had been hard for that lab to approach previously. And one of the most exciting projects I got to work on was a disease called cerebral cavernous malformation. And Dean's Lab had been involved in helping to identify the genes that caused this disease, that built some of the early cellular and molecular models, and actually had built, just as I joined the lab, three of the animal models, the conditional lockout models, which became sort of the gold standard or one of the gold standards for the disease. And we thought we figured it out. We thought we understood the pathophysiology of the disease, and we used a well-known, widely available FDA-approved drug in that animal model to test our hypothesis. And while there's lots of reasons why this could have happened, what we found was that that particular drug, this was simvastatin, actually trended towards making the animals worse. And it was this, you know, pit of the stomach for me. I'd had experiments fail all the time, but we'd been working on this for so long and we felt so sure of this mechanism of sort of row A activation that to not see this particular drug rescue in this animal model was, you know, really surprising. And it was in the weeks and months after that failure where I think recursion was really born. And it was in those moments as we thought about what had we missed? What was our bias that had led us astray here? And what could we do to actually understand how we could still approach this biology in a way that maybe gets us a better answer? And that's where we came upon this idea of essentially phenotypic screening, which has been around for a long time. We're certainly not the first to propose that. But phenotypic screening is actually a really useful tool when you don't know the mechanism or the target you want to go after. You just have a phenotype that you want to understand and exploit. And so we started exploring that path. And as an engineer, I really honed in on work by a woman named Ann Carpenter, who's at the Broad Institute still to this day, who really pioneered a field called phenomics, which is image-based omics, where rather than trying to build a phenotype, a phenotypic screen where you look at one thing or some increase in fluorescence or something like that. You actually used microscopy images of cells that you stained for lots of different cellular organelles, and you use computer vision and machine learning to extract much more sophisticated measurements of these kinds of images than a person can actually see with their own eyes. Because our brains just haven't evolved to see some of the intricate subtleties of those kinds of images. And so Ultimately, that's what we did. And about a year later, we put two compounds into the same animal model and both of those compounds worked. And that ended up becoming recursion. And I can share more about how the business came to be, but that's the idea that became recursion. We asked if we could scale that, if we could do this for lots of different areas of biology. And I'm happy to share that the phase two trial of one of those molecules for patients with cerebral cavernous malformation has now been enrolling patients for nine months. So it's great to see the origin story translate into, you know, first medicine that we're a potential medicine that we're advancing towards, we hope, some benefit for patients in the coming years. Yeah, yeah. Well, I mean, certainly it's a lot further along than it was when you shut it down, right? All right. So there's a host of questions in there. I want to stay with you, though, like, you know, you personally. Definitely. When you were sort of running through that experience and I guess the seed was planted, the seed of recursion was planted. At that time, you were a bioengineering guy at University of Utah. Is that correct? That's right. So at that time, were you thinking engineering solutions company, tech solutions company, or were you already thinking therapeutics? Uh, neither. I was actually training. I hoped at the time to be a cardiovascular surgeon. Uh, so I was doing sort of the MD PhD thing. I wanted to learn the critical thinking that comes with a PhD, um, because I wanted to do research in, in, uh, you know, heart disease. Uh, but yeah, this was a big, uh, a big change in career for me and a lot of discussions with my wife, after investing nine years at that point post-undergrad to decide to take a big diversion into business was, yeah, it was certainly a bit of a change. All right, level with us. Is your wife still disappointed that you're not an MD? Well, my wife's an MD and so we've got - So she's got that on you. Yeah, exactly. So I think she's fine with it. Awesome, that's terrific. So now getting to the company and sort of like the company's aha moment, when did it become clear to, like as the company kind of was in those super early days and forming up, when did it become clear to the company, you leaders of the company, that this technology, this idea of imaging and machine learning, which again, it was super early days for that back then, would be integral to what the company would become, right? It's sort of at the core. We knew from the beginning that if we were going to be successful, all of those tools would be integral. I think we knew that we were not a set of founders or the kind of company that was going to compete with, you know, a Bay Area or Boston-based company with great cellular molecular biology and a target hypothesis. And that was okay with us. We wanted to explore a different hypothesis. which was that you could use technology tools to accelerate discovery and development and to scale it. And so I would say that was the hypothesis that we tested initially. And we were pretty bootstrapped in the early days. So we started out in November of 2013. Very first thing we did was start writing a direct to phase two SBIR grant to the NIH. We got an almost perfect score on that grant in July of 2014. Still took another eight or nine months to get the money, but it was enough to get some angels to invest in the company. And for the first almost three years until September of 2016, we were funded by one institutional investor, some friends and family angels and ourselves and the NIH. And we ended up getting multiple NIH grants. And those grants were designed to help us ask the question, could you build a new kind of omics that could help you discover drugs at scale? And we never knew with certainty that that could happen. But by 2016, we had a lot of evidence that the probability was really high, that it was likely to work. And it was a question of how well it would work and how far we could extend that philosophy beyond not just target discovery and hit discovery. But could we even take the same philosophy of applying technology through lead optimization and through translation and maybe one day into development? And so we've now grown from a company that was really founded in a MLAI-enabled phenotypic screen into a company that's built technologies at many steps of the discovery process and soon, I think, in the development stage as well. Yeah, good. We're going to save those. I had some questions about that around development and the clinic and kind of what the endgame looks like for recursion. We're going to shelf that for a minute. I'll put that on the shelves. All right. Because I want to learn a bit more about the technology. And bear with me here because I'm not an engineer. I did cover IT for a long time, but I'm not too far in the weeds. So I want some clarity around this. And if I'm โ€“ transparently, if I'm phrasing these questions in the wrong way, if I'm framing them in the wrong way, I'm going to lean on you to tell me that. Will do. But this recursion operating system, I want to learn about that technology. So I guess, you know, at a high level, the recursion operating system, one, is that sort of the core engine that drives the discovery effort? It absolutely is, but it is many things. It's an umbrella term for lots of people, processes, and technologies we use. All right. So let's break that down a little bit. You know, what is it? Yeah, so the foundational piece of that is essentially a set of technology tools we've built, and in particular, things called convolutional neural networks, which are software programs that are trained to get really good at doing specific tasks. So people who are looking at the news today might be seeing something about this new AI that's allowing you to put questions into a web interface. And it's like writing essays for people that are pretty good. You can train AIs to do language like that AI has been done, or you can train AIs to be really good at other tasks. We trained our AI to look at microscopy images of human cells and to be better than any human in the world by multiple orders of magnitude and seeing small differences in those images. And so at recursion, we have a giant laboratory full of robots where we do about 2 million experiments a week, up to 2.2 million experiments a week. And these are different human cell types. We use CRISPR to knock genes out. We add potential drugs, small molecules. large molecules, combinations of those things. And at the end of all of those experiments, we take these microscopy pictures. And then this machine learning AI system turns those essentially into mathematical representations that data scientists and software engineers can help us learn to read. And those are essentially what we call maps of biology. So it's a lot like the way, it's a lot like some of the tech companies that folks may interact within their everyday life. So if you want to get from point A to point B in the world, in the old days, you used to pick up the yellow pages, identify the place you wanted to go, call, make sure they were open. And then you'd pick out your 120-page paper atlas. You'd figure out where like square F7 was on page 126 and draw your fluorescent marker line between your home and that place. And then you'd start navigating and you'd run into traffic on the freeway and have to just figure it out. Now everything's on our phone. And not only is it automatically directing us, if there's traffic, it gives us diversions. We're trying to build the same thing for disease, but to build that, you have to have a map. And we're generating these huge data sets. We've got, no, go ahead. Let me just interrupt you real quick, because I want to be clear on this, because I think it's fundamental. You know, when we talk about machine learning, we talk about big data and machine learning. Oftentimes, I'm having conversations with folks who talk about, you know, when the data the big data is coming from external sources. This is when a lot of problems occur, challenges, I should say, hurdles to overcome in terms of data structure and input. In Recursion's case, it sounds to me like, so square me up on this, it sounds like the data source is internal, or at least largely. Like you guys are creating this data as a matter of practice on a weekly basis and feeding the machine. That's exactly right. We've generated over a billion of these images. and those are our core data set. And you're exactly right. External data is not only sometimes hard to use because of the way that it's, you know, it's essentially not relatable very often. So some of that data may have been generated at a lab in Boston and some at a lab in California and little things like the barometric pressure or the humidity of the air can actually really throw off those experiments and often they're not recorded. And so munging all of that data from lots of different sources can create a lot of noise. And then often it's basically saved in different formats and it's really hard to kind of tie everything together. There's tools to do that, but it's tied together. And the other challenge, of course, is that there's hundreds of companies now who are building software tools to explore biological data. If everybody's moat is that they're all using the same data set and they're going to make a slightly better software tool than somebody else, I think it's challenging to stand out. It's challenging to differentiate. If your moat is that you've built one of the largest relatable biological data sets on earth, and on top of that, you have fantastic tools and talent and teams that are leveraging that data, that's much more defensible and much more differentiating, I think, as you're trying to build a company that can be successful as a business. Yeah. The creation of the data, these images that Recursion is creating on a weekly basis, what parameters go around what images you're building? Like what directs that? What guides that? Is it indications you want to pursue? Is it disease states in the global population? Like, I'm just curious because, you know, if you think about inputting cellular data, you know, you could โ€“ like, what does that mean? That's everything, right? You totally get lost there. So I'm just wondering, like, what sort of guide rails, I guess, go around the creation of that? So we think having maps of biology is really important that are really broad because it's really hard to know how everything is related. I think we probably only understand like two or 3 % of biology as a species today. And so think of us a little bit like when Google started making their maps, they sent out these cars that drove every street. And yeah, they started in the big cities, but eventually they drove like every street in the world. And today, if you're lucky, you'll see one of these cars driving by. They basically drive around. They had airplanes flying. They were licensing data from people. All of that data got put into one consistent source that became a map. That's what we're doing at Recursion. So every week we do these 2 million experiments. We've now knocked out every gene in the human genome in multiple human cell types. We've profiled nearly a million molecules at multiple doses in one cell type and some of that in other cell types. And at the end of all of those experiments, we take the same pictures with the same camera with the same stains, which means all of this data is relatable. And so essentially everything we do every week at that base level of the company is to build a map. And that map then becomes the thing that our scientists navigate to novel relationships. And we don't just take it the traditional way from there. We also are doing massive sequencing data sets. We've explored proteomics data sets. We have cameras in the cages of our animals for animal models, and we use machine learning to extract much more sensitive and specific readouts of animal physiology or potentially toxicity. So we're really building technology into every step of this process so that we can start making predictions about a molecule. The second we get a picture of a drug, a potential drug on a human cell, could we make a prediction about what that drug might do in the context of an animal model? I think if you get enough data, you'll be able to make solid predictions in that way across the entire stack. And that's really what we've been focused on at Recursion. So we have now close to 20 petabytes of data. And just to put that in perspective, if you took every movie in human history, in every language, like every feature length film in every language in 1080p high definition, you're talking about four petabytes. And we have about 20. So this is a massive data set. And all of that data, no human could ever even make a tiny, tiny dent in looking at all that data, let alone understanding how it all interacts. But it's what machine learning tools are really, really good at. And they don't make a drug, but they give us insights about the relationships across this huge, complex data set that our scientists can key off of to say, you know what? There's a gene in this pathway that the world doesn't know about yet. we're going to go drug that particular part of the pathway because it's way more druggable and the rest of the world just hasn't published that it happens to be involved. Like that's what we want to be spending our time doing is really going after either novel targets or targets in the context of a disease where they haven't been explored before. Mm-hmm. So a couple of questions on the data set itself. One, is it currently and will it remain proprietary to recursion? I mean, you're talking about a data set that, you know, regardless of the, I guess, type of data, that quantity is going to have some significant value ascribed to it, right? To an industry. Is it recursions and recursions alone, and will it remain that way? Well, a lot of it is ours and ours alone. Even if you wanted to download it, it would probably take you six or seven years just to download the data set without a special setup. So it's a lot of data. That said, we've released a bunch of large data sets. So in the first few weeks of the pandemic, we did a screen of about 1 ,700 FDA-approved drugs against live SARS-CoV-2 virus on human cells in a BSL-3 laboratory nearby. We took 300 ,000 pictures of those and made them available. And we also made predictions about what drugs would work and what drugs may not work. And we published this in a preprint form in April of 2020, it was before a lot of the other data, but we predicted that remdesivir was going to be positive. It was going to help in that context. We demonstrated that hydroxychloroquine did absolutely nothing in that context. And so far we're eight for nine on the predictions we made in that work in April. When one of those drugs has gone through a randomized controlled trial subsequent, those predictions bore out. The one we got wrong so far is dexamethasone for a variety of a variety of reasons. But I think we'll continue to make these datasets public. I think we have three or four large datasets we've made public. There's been a big competition run with 800 machine learning teams on one of the core datasets we released at Google, a company called Kaggle that runs these big crowdsourced projects. And who knows, maybe there will even be a really big dataset we'll release in the near term as well. We'll see. Yeah, that's awesome. That's a far better answer to the question that I anticipated getting, Chris. Thank you. But we're not going to release the whole thing, just for clarity. No, no, no, understood. But I mean, you know, it served some serious cause, right? That's had some serious good cause there at times. And I'm not going to put you on the spot and ask you what your analysis told you about ivermectin. We'll just pretend I didn't even say that. Now, it occurs to me, I have this hypothesis around the adoption of artificial intelligence intelligence and machine learning and interest in computational biology in this market that we're in right now and the hype among new and emerging biopharmas many of whom are like seriously cash-strapped right now you know i just i recorded a an episode that um just this morning that'll probably air i think the week before this one so for reference for our audience last week uh alan shaw and i talked about some stats around the fact that like 40% of the biopharma the emerging like preclinical biopharma market has less than 12 months cash runway, right? So, and that's just one snapshot of this bigger problem that I'm referring to. And I have a theory that this sudden and profound interest in automation, artificial intelligence, machine learning has to do with efficiency. You know, it has to do with like, you know, we all know that the discovery of biologics is a long and arduous process. That's before you even get into clinical and moving down the pipe. So efficiency is incredibly important to companies that have to preserve cash runways and be very efficient with borrowed money. But it's one hypothesis that I think is driving interest in these technologies. Now, the question, that's a lot of words, but the question, like, for recursion, building this data set, this giant, massive data set, you know, maybe as far as building giant data sets is concerned, it may have been an efficient process, but it's been a long road. Like, how have you rationalized or justified as you've built this thing? Are we a U.S. genomics? Are we a great big giant data generating company or are we a therapeutics developer? And we have to answer to our shareholders that question. So just give me some flavor on that. Absolutely. It's a great question and certainly one that's very timely. So from the very beginning, our belief has been that the way we differentiate ourselves is by investing in the platform and the process of discovering medicines in a more efficient way. And when I say efficient, I am talking about scale. I am talking about speed, but I think the biggest lever there is success. If you simply, if you go the same speed as everybody else at the same scale as everybody else, but you have a higher probability of success in phase two and phase three, you will dramatically lower the cost of medicines and dramatically increase the efficiency. Because most of the challenge for us as an industry is that biology and chemistry are really, really hard. and most of the drugs that we discover and develop fail. And that's not news to anybody here. So that's where we're focused. And that's why we think these maps that show us the terrain of biology and how chemistry is interacting with it could be so important. We think that opportunity is so big that it's worth investing really deeply in exploring it. And it's a bit like when, you know, railroads and automobiles were the main way to travel. The idea of building a flying machine was crazy. and it took a ton of effort and a ton of failure to make it happen. But when it finally did, there was an explosion in uptake. And now none of us think twice about jumping in a giant metal tube and shooting across the ocean. It's like pretty normal as a part of our lives. I think we see the same thing as possible in medicine. And we think it's really important to be the one that invents the plane in this kind of analogy. And so we're willing to make that investment. But But one thing we've done differently, I think, than many peers, we knew from the beginning that the currency of this industry, credibility and credibility not only with the industry, but with investors comes from assets in the clinic that are going to make a difference to patients. And we did not believe that we would have some endless amount of time to try and build this cool contraption. We believed that along the way, we had to show that the work we were doing was having some demonstrable effect on getting medicines towards patients. And so if you look at our pipeline, we now have five programs in the clinic. Four of those programs are molecules that our chemists did not design. They are molecules that we didn't have chemists when we found them. We used our platform to find a new place to take those molecules, a new disease context in which to apply them. And that was a way for us to show early progress of the platform, to show that this investment might make sense. And I think that's positioned us really well to then have these long-term visionary thought partner investors say, hey, great, this is good early proof of concept that you can do something that matters. now we're willing to continue supporting you to keep building this and make it better and better over time. And now we're investing much more in the chemistry side of things and applying technology to that. So I think, you know, you have to show progress as you go. And that's been core to what we've tried to do. Of course, plenty of setbacks along the way, but core to what we've tried to do since the early days. When you're striving to excel in a new arena, the best guides are the ones already doing it well. The business of biotech brings those voices forward to help new and emerging biopharmers turn their innovations like mRNA and cell and gene therapies into clinical realities. Tune in and subscribe for insights on hiring, regulatory, and other need-to-know topics for biopharma leaders. The podcast is brought to you in collaboration with Cytiva. Check out their resources at Cytiva.com backslash Emerging Biotech. That's C-Y -T-I -V-A .com backslash Emerging Biotech. Again, a great answer to my question. And one of the things I love about this job is I get to interview super smart people who actually can extrapolate the question within all the words that I'm using when I try to present the question. And you did exactly that. Thanks, Chris. you know the thing I've thought a lot about is what biopharma companies start with and how quickly or not they take what they're starting with and develop a pipeline which is beautiful segue you just offered me is this pipeline development conversation or recursion you know you know if I'm looking at it just at face value it occurs to me that the proof is sort of in the pudding to use a bad cliche because you've got this broad you quickly grew a deep and broad pipeline across indications, multiple modalities. Give us some, I guess, color on how the digitization of biology enabled that. And you already have. I mean, obviously, to a degree, you've explained that to us. But even if you can give us some specific examples within candidates in the pipeline, this is kind of how that came to fruition as a result of the platform. Yeah, absolutely. So one of my favorite examples is that we have a program right now, it's still early, but a program right now that I think has used a lot of our tools at many steps. And so it's a good one, I think, for this question. We call it Project Gamma. And it is focused in HRD-negative ovarian cancer. And there have been some papers that have demonstrated that a target called CDK12 might be really important in the context of ovarian cancer. And that's well-known, that's in the literature, but CDK12 is kind of hard to drug. Not only is it hard to get a molecule that binds to CDK12 and not some of the other proteins, CDK13 and others that are quite a bit like it in biology, but it also turns out that some of these CDK proteins do a lot of other important things that mean if you drug them, you can have some potentially pretty limiting toxicities. And so what we did was we looked at our map of biology and we noticed that there was something adjacent. So, you know, you're like zooming in on a neighborhood and you're looking at the house of CDK-12 and you notice that across the street, there's something that the world hasn't yet described as the world doesn't know this thing is right next to CDK-12. It's not a CDK. It is a gene that encodes a protein that is not known to be involved in that kind of biology, whatever, in any kind of way. And we thought that is really interesting. And so we designed some experiments to test whether this was a relevant association. And in fact, when we took our first drug targeting this novel protein into a PDX animal model of HRD negative ovarian cancer, we had not only were we able to essentially sensitize these tumors to PARP inhibitors, which was our original hypothesis, it actually turned out that we could get 100 % complete response just with monotherapy going after this novel target. This is what I'm excited about at recursion because we're now using our chemistry technologies to further improve that molecule. And our hope is that we could have a precision oncology drug going after an area of high unmet need in HRD negative ovarian cancer, going after a target that the rest of the world is unlikely today to be working on. And that's exactly what you want in this industry. I don't think we want 16 companies trying to go after inhibitors, the same target and inhibitors of that same target. And then having Bruce, we just talked to, uh, with Alan, I'll reference again that Alan Shaw episode. We talked about Bruce Booth's, uh, you know, uh, uh, his year in review, uh, that was just recently released and he identified more than I think 110 companies targeting CD19. That's a great example. And, and I, you know, I, Bruce is very thoughtful in the space. I agree with him there. To me, that's not the most efficient use of capital. One other way would maybe somebody can go after a target nobody else is working on. And we might fail, we might succeed. But I think if more folks were going after more different targets and we had the right tools to help us understand biology in that way, we might give ourselves more shots on goal. It's a touchy thing to talk about, but even if you look at the amyloid beta field, and obviously there's been some recent news that some consider success, some don't. But we've spent a lot of years and hundreds and hundreds of clinical programs going after the same couple of targets with maybe less upside than one could derive from if we'd had a much more diverse approach as an industry 10 or 20 years ago. Easy to say after the fact, for sure. But I do think in general, We need to go after more novel targets. Yeah. I'm interested in your ability or willingness to share some strategy from the leader's chair around that, right? Like I'm just going to kind of free wheel with a couple of questions on that. So, you know, you're doing that. Like you're practicing what you preach. Recursion is going after more novel targets. Now, you could apply all that technology, all that technology, all that compute power to a giant big swing, you know, a giant big swing to some giant global, you know, global indication that affects millions and billions of people. But, you know, you've spread that out. So how do you, I guess, rationalize that? Why is it, I don't know, I guess, I mean, I could presume a couple answers, but how do you rationalize that approach and why is that strategic for recursion? Yeah. So let me jump in with one like very tiny, tiny correction. No, you correct me all you want. We do have multiple programs in our pipeline in the clinic, which are going after well-known targets. And oftentimes we get this critique because I say things like I just said, and the people are like, wait a minute, your clinical program, you have a MEK inhibitor and an HDAC. We're deploying those in the context of diseases where they haven't previously been explored. And so it doesn't matter to me that the target itself is novel. It's can you try a novel target disease combination? That's, from a scientific perspective, important for me to just say because I can imagine the replies in the podcast. Now, to your question, we are going after giant disease indications. In December, almost exactly a year ago, December last year, we signed a huge collaboration with Rosh and Genentech to deploy our technology in partnership with them for more than a decade, exclusively focused on the whole of neuroscience and one, I can't specify which, but one significant oncology indication. And we chose to partner with Rosh and Genentech there because I don't think we have the resources to run a really good trial in one of the biggest areas of unmet needed neuroscience as a small company. And I would hate for our early development stage as a company to mitigate the potential upside if we found some new target of getting it to patients. And so when we go after big areas of biology, we often do it with a big, extraordinarily experienced, savvy partner like Roche, like Genentech. We're also partnered with Bayer and Fibrosis, another area where these are huge areas of unmet need. But we don't do it alone. Where we built our own pipeline to demonstrate to the world we can do it is in areas of, I would say, more niche unmet need in rare genetic diseases, precision oncology indications. where, you know, in the context of CCM, the program we started talking about at the very beginning, our program that's most advanced in our pipeline, we're unaware of any other company that has any program in preclinical or clinical stages for that disease. And there's six times as many symptomatic patients in the US and Europe as cystic fibrosis. But yet there's no company that we are aware of going after it. And so we're deploying in these areas of concentrated unmet need with our own pipeline and then partnering with large experienced companies to go after, you know, the really, really huge indications that are more of that big swing like you're talking about. Yeah. All right. Good. Why? Just curious follow-up question. Why the lack of interest in CCM across the industry? Is it difficult? Is it just not sexy? Is there no money in it? No, nobody knows how it works. I mean, this is what happened back in Dean's lab is we thought we understood the pathophysiology of the disease and we were wrong. The genes that cause the disease when they're mutated encode three scaffolding proteins, which are kind of hard to trace back. And there's some great academic groups that have made a lot of progress on the disease, but fundamentally there's not an obvious target for the industry. What our phenotypic approach did was give us a non-obvious target. One that probably if you just read the literature at the time and somebody said, what about this? It would be a little bit hard to get excited about it. But that particular molecule went on to be really, really impactful in several different animal models of the disease. And that surprising target is now in phase two. And look, we should succeed or fail based on our readouts in human patients. Like we shouldn't get a pass if those particular drugs don't end up working. But I would much prefer being the only company going after a big disease than being the 101st or however many Bruce said were going after whatever target that was. Yeah. Yeah. CD19. Yeah. Yeah. Okay. So we're running short on time and I want to, I want to be respectful of yours. So I'm going to, yeah, you're okay with a few more. Yeah. Yeah. We're good. Good. I want to, like I said, I shelved that, that conversation about what you know, beyond the platform, what, what recursion is and is, is to become right. You bill yourself currently as a tech-enabled drug discovery company. Now, everybody knows a company at recursion's age on this continuum with the technologies it has and the resources it has. You have more than 400 associates, I think, now. Some good resources there. You can shift. You can add. You can move, right? You can become a number of different things. You just mentioned partnerships, which kind of plays into the conversation. some really strong partnerships, and then also some independent candidates in the development pipeline. So what's the intention, I guess, to the degree that you can share with us? Knowing that things change all the time, what's the intention? Do you guys want to take candidates all the way to the market? Do you want to be a drug discovery company and maybe once we get to early stage clinicals, be looking for partners and dump things off? I'll leave it open. What does Recursion want to grow up to be? Well, what we look like when we grow up has changed several times, as we've learned. But what impact we want to have has not. And our mission since the very early days has been to decode biology to radically improve lives. And I think we will become whatever allows us to radically improve lives. And it's the lives of patients, the lives of their families, the lives of our employees, the lives of the communities and the people in the communities in which we work. We have a broad vision of what that looks like. That doesn't mean that we're just a drug discovery company necessarily. I think that if we see the opportunity to, in a differentiated way, commercialize drugs, which for us would not mean hiring a very large sales team. For us, it would be, can we deploy technology, for example, to make sure that physicians understand the data behind our medicines so that the right patients get those medicines when they need them? Could we perhaps find ways to partner or build our own technology in clinical development that helps us select our patients better or run more efficient trials? All of that is on the table. I think in the current market context, we're staying very focused on early stage clinical development, a narrower pipeline of oncology programs that we can move more quickly and delivering for our partners. But if you fast forward out the next 10 years, I think it's very likely that this market cycle will shift. And this is a company that has big ambitions to broadly impact healthcare around the world. And I would rather fail trying to do that than succeed at not moving the industry forward and maybe discovering a medicine that impacts a few people. Because I think the industry needs both people delivering medicines using the current technology. That's super important. I mean, my life has been dramatically affected by a medicine in that way. And I'm really grateful for the scientists who did that. But you also need some companies who are pushing the way we discover the medicines forward. And some of them are going to fail and some are going to succeed. And ultimately, that's how we create progress for the industry so we can do it better in the next decade. And I think we want to be one of those companies. And I think we've got a high probability of success, but we're certainly pushing really, really hard to use technology in lots of different ways. And that could come all the way up and down the pipeline, all the way to the commercial side of things, for sure. Yeah. Yeah. Pretty cool. Pretty exciting. You know, in a lot of what you talk about, like the big vision stuff, I can't help but think about, you know, Chris wanting to do things in a better way. I mean, you just referenced sales, pharmaceutical sales. Oh, God. Right. I don't even want to get in. I'm not opening that can of worms. Oh, come on. Let's do it. No, but I do like the beautiful and altruistic intent behind what you referenced, right? Like here's the data, here's the education. It's easily, you know, let's, let's make it accessible to physicians and kind of skip that, um, you know, that, that, that person carrying the bag, no offense to those who carry the bag. I, I, you know, that's, it's not them. It's the machine. I'm so tempted to talk about the ads on TV where at the end you have no idea what the drug does, but the ads anyways, we won't go there. The only thing you know, the only thing you know is what the drug could potentially do. And it's very scary, right? Like, yeah, it's, it's awful. It's awful. Yeah. Um, I hope we're not going to ever make those ads. I'm, I'm, I'm sure I'm going to eat those words one day, but for now, I really hope we'd ever make those ads. Yeah. Well, and that, and that gets to my next, uh, my next question or a couple, couple of questions. And that's that, uh, you know, the intent, uh, to create a better way of doing things. And one of the things I know about you, Chris, you may not know that I know this about you. Um, I I've only, but I've been covering this space for, I don't know, three years maybe now. Like I said, I come from an IT background. I'm not a scientist, not a biology guy. It's no joke that I did drop out of AP biology my senior year of high school. But I love this space and I love learning from people like you. And one of the first people that I had an opportunity to talk to when I took this role at Bioprocess Online, and then eventually, this was pre-Business of Biotech podcast, was one of your employees, Tina Larson. Oh, awesome. Tina Larson was one of my, yeah, she was one of the first people I interviewed when I joined the Life Sciences team here at Life Science Connect. And I'm very transparent about what I know and what I don't know. So when I sat down with Tina, I said, listen, I covered IT. I have I have an English, you know, my degree is in English. Like, you know, you're one of the first people in the science world that I'm talking to. And she was so beautifully gracious with her time. And another thing that we talked about that she was wonderfully gracious about was this. We talked about the culture at Recursion. We talked about, we spent a lot of time talking about women in biotech and diversity and inclusion. And I admitted to Tina in a very transparent way, I think. I think I was doing the right thing there. is very uncomfortable for me is, you know, I admitted to her that I'm like a, you know, middle-aged white guy from rural Pennsylvania, and I want to put together an article about diversity and inclusion at recursion. And it's kind of like, I don't even, I don't, because I'm a privileged middle-aged white guy, I don't even know what to ask. Like, I'm not sure how to frame the questions. She helped me so much. Like she said, that's okay. That's okay. We're, you're engaging the conversation. Like you're starting the conversation. That's where we start. And then she talked about you. I'm using a lot of words here, Chris, but then she talked about you and she talked about the culture that you're creating and have created at Recursion around diversity and inclusion. Anyone who follows you on LinkedIn or has connected you on LinkedIn or pays attention to what you're doing on social, they see that, right? Like it's your front and center. You're loud about it. So I guess first I want to get to the core of that. How does this guy who's, you know, he's a bioengineer, I'm not going to call you a propeller head, you're very well-rounded, got a great well-rounded personality, but you come out of, you know, bioengineering school, you're a whiz kid as far as engineering is concerned, you're building this tech-focused company, where does the drive, I guess, or passion for diversity, inclusion, equity, and doing the right thing, building a better thing and doing the right thing, where does that come from in Chris Gibson? Uh, if it, well, I appreciate all those words and I'm not surprised you had a really good, a really good conversation with Tina. I think of her more as a mentor to me than, uh, really as an employee. Um, she's, she's our chief operating officer and president and, uh, she makes sure that, you know, we don't get too far out of line, but that we're able to really push limits. She's incredible and super experienced. I've learned a lot from her. Um, and I think that's where this comes from in me is in my life already. I've had the opportunity to learn from incredible people who have a wide variety of backgrounds. And I've seen that value in places that maybe you wouldn't have guessed at the beginning. I mean, thinking back to all the ways recursion may not have happened. I think back to some of my teachers in middle school who encouraged me to stick with science when it was hard at the time, when I wanted to believe that I could be good at it, but I wasn't sure and I was questioning it. And I wouldn't be here if those teachers hadn't taken and invested deeply in me back in sixth grade, right? And so when these people step into your life and they invest in you in a big way, the very least you can do is when you start having a platform is invest back in people. And it doesn't matter if you've got smart people, it doesn't matter where they come from, let's invest in them. And I think it's important that all of us make sure that we pick the people to invest in based on who they are and what kind of value they can deliver and not anything else. And I think we have made that a very intentional piece of our culture. And it's made us very much stronger than I think if we were a bunch of anything monolithic. And it even carried through, like I said earlier, in Dean's lab. There were physicians sitting next to engineers, sitting next to molecular biologists, cell biologists, chemists. It was a super diverse place of thought. And so that's just always been something that I've valued. And yeah, I think that's where it comes from. Yeah. I'm sitting here trying to think about that. It's an abstract thought, but the jibe between, and I'll make a word out, between having a big vision, right? And creating culture. You could hang your hat. You could create a company and hang your hat on the product that you're producing, the technology that you developed, the widget that you make. You could just hang your hat on that. But it occurs to me that when your vision is as big as Recursion's vision is, beyond the development of a therapeutic pipeline based on some technology, what that technology at the end of the day can do, the culture becomes more important, right? Oh, absolutely. It's more important to develop that culture and nurture it when you're taking big shots than perhaps when you're just pounding. If you look at the world of companies that are iconic, companies that have changed the world in the biggest ways, both good and bad, right? I'm not just talking about Amazon or Google or Genentech or any of these companies. I'm also talking about some of the big failures. There have been some recent ones in the crypto world, right? Companies that create massive impact, both bad and good, tend to have created really, really clear cultures. And some of them did it intentionally to make sure that the impact they had was good. And some of them, I think, did it intentionally or unintentionally in a way that created a lot of bad outcomes. And so, yeah, culture is the most important thing if you want to deliver. And I think whether you want to deliver a single medicine or you want to deliver a pipeline or a whole new paradigm for how one discovers and develops medicine and healthcare generally, culture and people are going to be at the core of it. And you'll live or die by the kinds of people that you bring together and the culture you create. Because once you get past about seven or eight people, it gets really hard to be involved in all the decisions. And so you've got to have created a culture where you trust that people are going to be making decisions based on principles and values that are aligned. So, yeah, I think it's the most important thing, actually. Yeah. What haven't I asked you in this broad and rambling conversation, my question set, what haven't I asked you? You know, you might walk away from this interview thinking, man, I really wish he would ask me this. He's missing something. I think we've covered it. It's a nice, broad set of stuff. Hopefully I wasn't too rambling for folks, but I always tend to be a little bit. I get that feedback. No, you're great. You're great. Enjoyable. Thanks for the time. See, it was enjoyable. I thought it was enjoyable too. I don't know why you put me off for so long. Like I said, we were executing and I got to get back to it now. I'm going to let you go do that. Thank you. This has been great. Thank you a million for coming. I really appreciate it. Thank you. Take care. So that's Recursion Pharmaceuticals co-founder and CEO, Dr. Chris Gibson. I'm Matt Pillar, and this is The Business of Biotech. We're produced by Bioprocess Online.! In partnership with Cytiva, visit citiva.com backslash emerging biotech to see firsthand the company's commitment to emerging biotechs. And go to bioprocessonline.com backslash B-O -B to subscribe to the Business of Biotech newsletter. Finally, if you like listening in on conversations with the leaders in biotech innovation, leaders like Chris here, subscribe to The Business of Biotech. Leave us a review. Leave us some feedback. Share us with your colleagues. And as always, thanks for listening. Thank you.