Ep114: Chris Gibson on Industrializing Drug Discovery

Thanks for having me, Luke. Excited to be here.

You know, look, I think what we're building is going to take decades to put in place. And so it's my belief that creating a track record of trust and transparency with the folks who are going to go on that journey with us in the long run, so to speak, no pun intended, is really important. And I would rather have us be upfront and upset some folks in the short term, but create the kind of trust, especially as a new management team, with transparency.

And so this specifically was written because we actually shared an update on a program that was the right decision to make, but not, you know, a good one. We decided we needed to take more time before we started a clinical trial. And we got a little bit of pushback from some folks that we didn't sort of bury that in the 10-K. And that line was meant to share that that's not the kind of company we are. We're going to put things up front, good and bad, and we think over time that'll earn us respect of the folks that we want to have alongside us on this mission, because it's going to take a long time to build what we want to build.

I grew up in Portland, Oregon, actually.

So, my mom was an artist. She's still an artist today and does amazing tapestry and encaustic work, kind of abstract landscape work. And so I got a lot of creativity, I think, from her. And my dad was actually an engineer and business person who brought the two of those things together to build a computer company back in the '80s that he built out of his early formative days at Intel into a company called Sequent Computers that went public, I think, when I was six or seven years old. So there's some interesting parallels there, as my oldest son was born just about a year before I started Recursion.

Yeah, I have a younger brother, Riley, who's, you know, as young brothers or young siblings do, you know, we messed with each other a lot growing up, but now we're super, super close friends. And he lives actually out in Denver.

So, we ended up going to a school called Catlin Gable. I started there in second grade. It's a pretty, pretty incredible place. It's old school in some ways. So, you know, we had a history teacher who would grade us, as he said, his name was Dave Corkran, and he'd grade us on a college curve because he thought it was good for us, and I agree, to learn how to write. We had these five-paragraph essays like every few weeks in his class. And it was super rigorous and super hard. But it was an amazing place. There were only about 60 kids in my class. I'm close with many of them still to this day, frankly closer than a lot of my friends from college even. So it was a pretty special place, and I learned a lot about writing and communicating there, which, you know, science as well, but that was the thing that Catlin really did well is I think taught people how to think, think creatively, to think logically, and then to communicate well, which hopefully comes across in our podcast today, or it's going to be super awkward. But ultimately I think it's been a huge ingredient in being able to become a co-founder right out of grad school.

Yeah, all the way through. It was just a pretty rigorous place, but a place filled with a lot of empathy and love, and a really formative place, I think, for myself and many others who went there.

You know, Luke, I was always interested in science and biology. And where I remember there being a big decision point was in sixth grade when we had this big class project in science and we got to pick the level of difficulty of our final exam. And I picked the hardest level of difficulty and ultimately didn't do super well. It didn't meet my own expectations. And my sixth-grade science teacher, Carol Ponganis, could see the devastation. And she actually took me to lunch and she basically convinced me that I could be really good at science, that it mattered that I picked something really hard to try at, and that it was better to fail at trying something hard than to succeed at doing something easy. And it was that lunch that I look back on as a fork in the road where I decided that I was going to dedicate myself to this. And so thanks to Pongi, or Carol Ponganis, back in sixth grade, because I think that's what put me on the course, along with many other things, my parents and other influences in life, my own interests, but that was a pretty formative moment for me.

No, not at all. I've actually never been an A student. There's always been kind of one class every year, every semester that I just can't quite get there. And I think that, you know, that's okay, right? There's a lot of pressure around these days, the 4.0 A student. And that was not something that was ever what I did well at. What I could do was do reasonably well in a lot of different things. And so that was, you know, for me learning kind of to be a translator across many different areas was pretty important.

I think the complexity, right? And biology is just so massively complex, and yet there is fundamental truth in it. That's the thing that I thought was really interesting intellectually. And then it felt important to me in my life to have some kind of impact. And biology felt like a great place to have impact. Whether you end up going into medicine or biotech or any number of fields, you know, changing somebody's life through their health, through biology, is really meaningful.

I went to Rice University in Houston.

The honest answer, it was the best school in my top 10 or 15 schools that I got into. And if I'm totally honest, it was also the school that was closest to my high school girlfriend at the time.

So I went to Rice 2001 to 2005.

Yeah, absolutely. I was doing bioengineering and business, and so we were hearing about that and many other interesting developments at the time. You know, my early path was actually really deeply focused in tissue engineering. And so I was a little bit less focused on kind of the molecular biology, therapeutics, genetics side. But certainly, I mean, the Texas Medical Center, I don't think many people appreciate the behemoth that that place is and how much incredible stuff is happening right there.

So, I was in Texas, started an MD/PhD program down in Texas at one of the UT schools, and my wife actually was wanting to do her medical residency in neurology, and she chose to come home to Utah, which is where she's from. And so, after a year of flying back and forth as a couple, it was just not a great business model for our marriage. And so, you know, we had a conversation and I decided to fly out to the university and interview with the head of the MD/PhD program, who's a guy by the name of Dean Lee at the time. And Dean was an exceptional person. We ended up chatting for a couple of hours. And essentially he said, "If you want to transfer here, I'd love to have you in my lab." And that's how I ended up in a molecular biology and genetics lab in Utah, which is something I never thought would happen in my life.

That's right. President of Merck Research Labs, my co-founder. Incredible teacher and mentor to me as well.

So, I think I was a little bit naive to the level of genetics that was happening here in Utah. I think like many stories in the world, it was a little bit untold, at least outside of Utah. And so I was pretty amazed right away. You know, I sat next to Kirk Thomas in the lab, who was the first author on the paper that won Mario Capecchi the Nobel Prize. Mario was right upstairs. And so I kind of got vaulted into this pretty extraordinary kind of genetics-focused curriculum and in a lab that was really, really strong. And I started learning through Dean and through others in his lab again that power of translation across different backgrounds and different perspectives. So, you know, you had geneticists next to molecular biologists, cell biologists, physicians, all in this one lab. And then I came in as the bioengineer, which was yet another kind of language and perspective. And Dean and that lab made it work. It was a really, really exciting place. It was an intense place. And I loved it. It was five of the best years of my life.

That's right. Actually, I've heard Kirk talk a lot about their time together. They used to run for like a couple hours every day. The two of them, I don't know, they did like 100 miles a week of running and then talk about their work together, and other people in the lab would join. And there's still a deep tradition of climbing mountains behind the university during the lunch break. And one of the things that was extraordinary is that Kirk said that the lack of being in an echo chamber that you might have had on the coasts was actually a really important ingredient in some of the success that they ended up having because they didn't feel as constrained by what at the time was a perspective that what they were trying to do was probably undoable.

Yeah. So, we were working on a specific rare genetic disease called cerebral cavernous malformation, and we shared just a few weeks ago, to foreshadow a bit, that Recursion just launched the first-ever Phase 2 trial on that disease, with actually a molecule that was discovered back during that PhD. And the original discovery came out of failure. We had used a traditional approach to try and understand the pathophysiology of this particular rare genetic disease. That method ended up failing us after many years of work, not just my own, but other people in the lab for years before I arrived. And it led us to what it felt like at the time was a dead end. And in that failure, we noticed something, as you mentioned, about images of human cells. We noticed that when you modeled this disease in human cells, they look really, really different. And as an engineer, we thought about doing a phenotypic screen. And as an engineer, one of the things that I brought to that discussion was the desire to kind of move some of the bias out of the screen and to quantify a lot of things. So rather than just say it's our hypothesis that activation of RhoA is the driver of this disease, can we measure actin stress fibers and then find a phenotypic marker for the reduction of actin stress fibers? I said, why don't we measure lots of things? Why don't we use maybe even computer vision or other kinds of software techniques to measure things in a really unbiased way? And so I got to coding, and three days later I gave up because it turns out I'm not a great coder. But I found the work of a woman named Anne Carpenter at the Broad who has built or had built something called CellProfiler, which is essentially software that allows biologists to do phenotypic screens and, in fact, to use machine learning and now neural networks to actually do that kind of work. And so we used that system. We did a phenotypic screen of a few thousand bioactive compounds and identified a few totally unexpected compounds that looked like they were having a really interesting effect in the context of a cellular model of this disease. And one of those is now the drug that we put in, you know, well, I don't know, they're randomized, but theoretically one of the early patients in our clinical trial may be on that drug now that we kicked off just a few weeks ago. And it was that moment when, after failure using a traditional approach, we could move to an unbiased kind of machine-learning-enabled vision approach looking at cell morphology as a new kind of omics that the gears started turning, Luke, to maybe build something around that.

Yeah. What most people said was, "That's a really cool idea, but if that idea held any water, somebody else would have already done it because it seems really intuitive." And that's an argument that doesn't really resonate with me. I mean, it's an interesting argument, but you know, you dive in and you explore the data, and it turns out phenotypic screens are really good for first-in-class molecules. Many first-in-class molecules are discovered with phenotypic screens. The challenge is that building a strong phenotypic screen takes a lot of work. And so I think maybe one of the fundamental improvements that we made, and a lot of credit to Anne Carpenter, who I think is really the thought leader of this now, most large pharma companies have whole divisions or groups doing the work that we're doing, and she really paved the way for all of that work, is thinking about images of human cells like you would think about gene expression, proteomics, or metabolomics, where there's a set of things you can measure across any disease or across any biological context. And what that allows you to do is to aggregate data over time and to create one large data set. And so instead of building a different phenotypic screen for every disease one wants to explore, we wanted to essentially profile how cell morphology changes across knockouts of every gene in the genome and across, we hoped one day, millions of molecules and thousands of cytotoxins. And now today, eight and a half years later, Recursion has actually done that. We've knocked out every gene in the genome and measured using computer vision how the cell morphology changes. And it turns out that it's extraordinarily specific. And I think the field of pathology was kind of a hint to this, right? For 100 years, we've been looking through a microscope diagnosing disease. But it turns out if you use computer vision, you can identify with extraordinary robustness really, really sensitive and specific fingerprints of biology. And it's sort of this idea that in biology, structure suits function. And if you look at diverse structure, that it will accommodate the function of the cell. And so images of cells become for us basically this holistic readout of cellular state. And so that was the moment we decided we at least had to go try and see if there was a hypothesis that we could test that images could be powerful broadly across drug discovery.

That's right. Yeah. We started the company in the fall of 2013.

Yep. Yeah, that's right. All of these areas were relatively sort of nascent at that time. I mean, we used siRNA for the first several years at Recursion, and that's a messy tool. You know, we did hundreds of thousands of siRNA experiments, and we knocked down every gene with multiple different siRNAs, and we would use computational methods to figure out if we could extract the gene-specific signal as opposed to the, you know, off-target seed sequence effects that cause so much of the effect of siRNAs. And so there were a lot of challenges in those early days, and our path was a bit different. We didn't have the network or the experience or the pedigree, right, to go out and raise a huge brand. It wasn't like I'd ever built a company.

So, the three founders, myself, Dean, and a colleague of mine from undergrad named Blake Borgeson, who's a computational biologist, we all put in a little bit of money. Mine was a... I was a grad student. Mine was a cash advance on my credit card. You know, Blake had already built one company, and, you know, Dean was an established professor. So between the three of us, we put $130,000 in to start the company. And the very first thing we did was sat down and wrote a direct-to-Phase-2 Small Business Innovative Research grant to the NIH. It went to NCATS. Chris Austin was the leader of NCATS at the time. And we ended up on the very first try coming back with an almost perfect score on that grant submission. And we essentially said we're going to test the hypothesis that you can use very basic machine learning to look at images of human cells in culture to discover medicines across not one disease at a time, but many diseases at a time. And we set the really audacious goal at the time of looking at 100 different loss-of-function, if I recall, 100 or 200 different loss-of-function genetic diseases over the three or four years of that grant. And ultimately, when we got the letter back and the really good score, that enabled us to go raise a little bit of angel funding. But for three years, we raised maybe a total of three and a half, $4 million, $4.5 million from a set of kind of, you know, angels and a couple of other investors. But we didn't really raise substantial funds until the fall of 2016 when we raised our Series A. Those first three years, that was about testing the hypothesis. Like, this was a crazy idea at the time. People like Anne Carpenter believed, who'd done this work, but even her work was like not nearly as popular as it is now. And we believed it was possible that this could be a really interesting way to discover medicines, but we wanted to ask a lot of questions and prove to ourselves that the power was really there. And through those first few years when we were super lean, seven or eight people, I was doing a lot of the pipetting myself, you know, we answered a lot of questions and we became pretty convinced that there was the potential for this to be a disruptive technology.

Absolutely. I think, I mean, as we now, as I'm learning about clinical development and taking drugs through the clinic and the responsibility that comes with doing experiments in humans, working with agencies like the FDA and the EMA, and, you know, the deep responsibility that comes with that and accountability, everything, I don't think if a founder doesn't have doubt, then that's a problem.

Yeah, I mean, there was a lot of... Yes, I was naive. I still am naive. I think all of us are around different things in life. Maybe my bucket of naive was bigger. But that was useful because there were people who came back who still said, like, if this was something that anyone could do, it would have been done already, and a couple grad students from Utah aren't going to be the ones to do it. That is a lot less common today, but we got a lot of that in the early days. And, you know, I think one of the great assets that my teachers, my parents, mentors like Dean and others gave me and us, the founding team, was listen to people's doubts, explore them, but then, you know, explore first principles around why they might be wrong and why you might be right. And if ultimately you logically and rationally believe there's something still worth pursuing, you know, don't let it hold you back. But yeah, there's been a lot of challenging times.

So I would say the biggest delta for us was we tried CRISPR, I can't remember, 2015, 2016, something like that, kind of really early days, and it didn't work better than siRNA in our hands at the time. There was still a lot of noisiness, and I think looking back, it was probably that we used kind of a constitutively expressed Cas9 that was probably doing a lot of cuts. A couple years later, some people on the team said, we're going to try it again, and we're going to do it differently. And when we looked at that data from the first, whatever it was, 10, 100 genes that they looked at with a kind of a new take on CRISPR, that was game-changing because we went from being able to use our approach to understand the relationships for about 10% of genes in a cell type to being able to understand the relationships for around 50 or 60% of genes in a cell type. And those we couldn't explore tended to be genes that weren't expressed. And so it became like this, all of a sudden, this moment, this vision of building maps of biology across the whole genome went from, you know, banging our heads against the wall with siRNA to this clear vision and potential with CRISPR. And we actually changed the whole company to essentially say, no more siRNA, only CRISPR, within about two weeks after that data came out. We said, moving forward, it's only CRISPR. It's going to be hard. There'll probably be setbacks. We'll probably make mistakes, but this is so much better than siRNA now that we've gotten over a few of these humps. We have to make the change now. And that was probably the biggest moment, I would say, kind of looking back scientifically at Recursion.