A Common Language for Therapeutic Synthetic Biology with Justin Farlow

Welcome to the Axial Podcast. Axial is an early-stage investment firm based in San Francisco. We partner with great founders and inventors investing in early-stage life science companies, often when they are no more than an idea. Axial is fanatical about helping the rare inventor who is compelled to build their own enduring business. Okay, I think we're recording. It's always really weird to press record, isn't it, Justin? So, uh, we're really excited to have you. Have you on this podcast. I think you're number 4 or number 8, actually. I did a bunch in the beginning, but yeah, really excited to have you. You're one of the geniuses of biotech, but you haven't had any social media presence ever. So this is kind of the beginning of a new era where people finally realize how smart and talented you are and, you know, get more visibility for Srirani. So hope you do well. How are you?

I'm doing good. Be fun to chat. It's always a little strange to make things public, but I enjoy it and I enjoy chatting with people. I enjoy the community. So, see, we can—

Justin. Exactly. Justin Barlow. Justin and Colin Farlow. You guys like the Wright Brothers, but for proteins. So you're gonna make— exactly, you're gonna make protein slides. Everybody loves brothers in America. It's American. It's such American— two brothers building companies. It's just part of Americana. So, um, you know, you got to tell your story. You can— you and Colin, you're going to start taking like Patrick John Carlson pictures one day. Yeah, it's going to be dope.

It'll be good. I hope so. We tend not to take the, the fancy pictures, just like the social media, but we have a good time.

So you're gonna do it like— yeah, you're gonna do it. This is the beginning of a new era. Uh, seriously, this is it. This is the— this is the, uh, round whatever. This is You're coming. This is like you're coming out. This Diana Ross song, you're coming out. I'm coming out.

Oh man.

You know, this is you. But let's talk about, okay, we're gonna talk about your background first and we're gonna go into Serotini and then we're going to kind of, a lot of your great ideas on SynBio and we could probably talk for a few hours, but we're gonna keep it short. So we have a, we're pretty busy. So yeah, to start with the conversation, yeah, we'd love to like kind of, I think for just for everyone's sake, especially the up and coming scientists and founder, kind of your background. And how did you get into science, biology? How did you learn? You know, one thing about Justin, not only an incredible immunologist, biologist, but a world-class programmer. How did you do both of those things? So whatever you wanna talk about, I would love to hear your kind of origin stories. Yeah.

Yeah, as far as science itself, more just a curiosity of how things worked, philosophical curiosity. So I actually wasn't— it wasn't obvious to me that I was going to be a scientist early on, but I really enjoyed trying to understand how things worked. So a lot of the early work I did was— early studies I did were in physics and in chemistry, trying to really understand that. And biology only came later. I actually studied a lot of the philosophy of science, and I found that to be almost as interesting as the science itself. In different tools and ways to get at what was important or what was foundational. By the time I got to grad school at UCSF, the biology had, I think, reached a point where you could start to program it, you could start to think about it in a technical way as an engineering discipline as opposed to more of just a cataloging discipline, and that really intrigued me. So I I had a little bit of background, had a bunch of friends who were really good at programming, even in middle school and high school, and they taught me how to program. And with that, built all sorts of little utilities and fun little things, but never professionally. But that was enough to be able to be part of the community. And I found these communities helpful. And being in San Francisco is part of that community of programmers as well as scientists. Then things like Hacker News and Y Combinator, I think, have done a really good job of allowing people to think about the technology and software and software engineering and programming in a way that really inspires people.

Cool, yeah, I mean, Justin's a really active participant of Hacker News. So if you check him out.

More than chat, but that's okay.

Check his comments out, I think he has pretty good comments. He has good comments, he has very thoughtful comments. And I remember talking to Colin and he was like, so, Justin just picked up like Objective-C real quick or things like that, or he knew how to just code in Golang. I think, I think Serati's coded in Go still, I think.

Yeah, we do quite a bit in Go. It's, it's a fun language to use, but best tool for the best job.

Exactly. Okay. And so you guys grew up in Indiana, you know, I love people from the Midwest. I grew up in Orange County, but something about the Midwest is romantic actually to me. I don't know why, it's just growing up in a boring area with open fields, right? So it's a great place to grow up 'cause it's safe and you can do anything you want. And so what was the kind of going from Indiana to San Francisco, was there a logic there to pick UCSF? Was Pacific Lab, was it part of a grand plan to start a company? Or kind of what was the logic to go from Indiana to SF? This is what, like 10 years ago, I think.

So maybe a little more than that.

More than that, actually.

I found Indiana, what was nice about it is it was full of life, literally. Again, I spent a lot of time outside in the woods surrounded by plants and animals. I didn't find that boring at all, actually. I really enjoyed that. At UCSF, it offered a, for grad school, so I guess to step back and to be clear, Colin's my brother. He helped me co-found Serotini, and we both grew up in Indiana, both went to Indiana University at Bloomington. And while I was there, Bruce Alberts, who was one of a few prominent professors at UCSF, had helped create a program called the TED Trad program. And he was actually giving a talk at UCSF. It was not actually for the students, it was actually for the faculty. And it was about this idea that many grad students didn't necessarily want to continue in academia, and that was hard for the professors themselves to acknowledge, but it was an important acknowledgement. And when you could acknowledge that, you could actually help the students even better, but you could also help society. And so the Tetrad program at UCSF had deliberately carved out space to, make it useful for the grad students to teach or to think about public policy as well as business or law. And so they recognized that not every grad student would become an academic, and I appreciated that. At the time, I actually thought I would be a more academic scientist, but I really appreciated that idea that scientists should be part of the greater community. And I actually didn't know about UCSF until he was there. I'm like, but that's the program I wanted to, to think about. So cool, we're He's a legend too.

He's like the biochemistry textbook king or something.

Absolutely. Truly on the textbooks, literally. He's a really good guy too.

Yeah, he's really good. I never met him, but I saw him on like— I see his lectures. He seems like a pretty chill person. So you UCSF, and I might get some shit for the Indiana thing. I love the Midwest. It is kind of boring, but it's beautiful. It's beautiful though. I might get some crap for that, but so well. And so UCSF. You're working in Zev Gardner's lab, you know, was like the first grad student from David Liu's lab and Bertozzi's lab. What was the kind of logic to join Zev's lab at the time and kind of maybe touch upon your research there and how that— yeah.

Yeah, I rotated through a number of really wonderful labs. Zev was a young professor and I was one of his first grad students. And it offered a freedom of projects and in retrospect, a diversity of projects because I helped to work with a number of my colleagues. And that was a wonderful opportunity to not have just a single project that I called my own, but to be able to have inputs and to learn from my friends and peers. So the projects were interesting. I didn't realize at the time quite how much of a synthetic biology approach that we were going to use in that work. But I really enjoyed that.

Oh, yes, I mean, it's always good to join a young lab and, you know, you're the— maybe talk a little bit timing then. Maybe it's one thing about your scientific career and that setup, Sriratni, is you joined UCSF at a really interesting time. This is when CAR-T was coming online. This is where, you know, Wendell Lim's work, a colleague of, you know, Zev Gardner's work, kind of started Cell Design Labs, got acquired by Gilead, and you've seen a lot of success with CAR-T. What kind of, can you talk about timing?

Yeah, so that actually happened later. So at first, it was really fun to see the engineering aspects and to see what were kind of considered toys, I guess, actually working. So I remember one of the fellow grad students who was ahead of me, like Anselm Lefskaya, had helped to create some of these optogenetic sensors. And it was just incredible that you could take a protein from a plant and put it in E. coli or in yeast, and now the yeast or plant would respond to light in ways that you could program. You could tell the new organism or this organism how to respond genetically in different ways. And that was really interesting to me because that started to look like the programming that I was used to. And it kind of spurred that engineering mindset. And it turned out this was happening not just at that point, but there were other features being built, new toys that were starting to be programmed. And this was the beginnings of synthetic biology. It had been going on for a while, but I was watching it, you know, in my neighbors. And about halfway through, you know, Cas9 was discovered, and we saw lectures about why that could be interesting and could be the future before it was. And same with some of these other synthetic biology tools that actually became useful and actually became what could be therapeutically useful. So it's fun to watch that happen.

Yeah, I think this is around the kind of 2012, 2014 timeframe, um, where kind of, um, a lot of these new modalities are having some success or getting discovered. And, and so you're kind of seeing this, you're, you're kind of developing tools that are related. Um, now we've established, you know, Justin's a genius, multidisciplinary, curious, studied philosophy and biology and everything, you know. One thing about Justin, he can talk about anything. I can talk to Justin about anything, actually. We talk about Chinese history and the Warring States, maybe, and we also talk about, you know, physics, something else. So that's one thing about Justin. He's really fun to talk to. Now we established that, let's talk about serenity. Let's spend the most of the time talking about serenity. Kind of, you're at UCSF, you're building these sets of tools, engineering various proteins, you're seeing colleagues have success in either CAR-T cell therapies or gene editing, you've seen some early signs there. What was the motivation to start Serati? How did you get the guts to start a company? And then, yeah.

I think one thing was watching others do it. And I know this is becoming a theme, but when you see that it can be done, you're like, oh, I could try that. Internally at UCSF, it's the same crew of people who had that insight, but Reg Kelly, who I actually met right as I was interviewing at UCSF, had gone on to create what became QB3 and the ability to incubate companies out of the UC system. And Reg Kelly did part of it and Doug Crawford did the other. And so they had a program to create a startup, startup in a box. They'd help you incorporate, help you get set up, and they'd really help you start to get those first grants from SBIR grants. And at the exact same time that I was graduating, my brother Colin was graduating from law school in Madison, Wisconsin. And not only were there companies coming out of the bio side of things, but in San Francisco, you're also in kind of the newer version of Silicon Valley. You know, you're a little further north from the true Silicon Valley, but you're also surrounded by companies.. And I could look out my lab window and I could see, you know, Reddit's headquarters and Adobe's over there and, you know, Uber's over there and Lyft's there. It's like, okay, well, just try it. So Colin and I initially created Seratni as a software company to make the design of those synthetic tools that I really enjoyed, those synthetic biology proteins the tools to make that efficient, the design of those objects efficient, the design of those objects predictable, and to couple the design with what you actually wanted to accomplish. A lot of the design tools at that time were how do you manufacture it, how do you, how do you build the DNA and make sure the DNA works. But a lot of the end product that you were working with was the protein that was encoded by the DNA. So you want to know the properties of what you were at the end of the product rather than the properties of the manufacturing. So that's really, we built that software to start off with.

Cool, yeah, and you definitely were living the full San Francisco experience. I think you guys, I remember you were living in Tenderloin for some time.

Oh yeah, on a couch and rent, definitely the full San Francisco experience.

Full, Justin and Collar have the full SF experience. Obnoxious rent. Paying really high rent in the Tenderloin. Now, you know, Josh, I think you live in a really good neighborhood now. I'm really happy for you. You're safe now. I remember you guys living in Tenderloin. I was a little nervous for you. I was like, "Oh, yeah, yeah." But cool, let's talk about Pinecone. I think you were calling the software Pinecone at the time, and maybe you still are. And so how did you think about defining the problem set? So you kind of have this inspiration to start a company. You're seeing kind of opportunities in protein engineering in general. For Pinecone, how did you find a problem set and how did you iterate across like, you know, building what people want? Because at the time, you know, you had other types of tools out there. How did you end up finding like applications for kind of this generalizable software you were building out?

It was initially motivated by my work as a grad student. I happened to be working on a protein called Notch. Notch is a ridiculous protein. It is a transmembrane protein. It is enormous. It is floppy. It is cleaved. It has every problem that any protein engineer would not want to have in the, in the protein that they were studying. I didn't quite realize that until later, but it forced me to have to be a little creative and made a lot of tools that other kind of protein engineering efforts might have, those tools weren't accessible to me. But what was also really interesting about Notch at the same time was Notch was a philosophically interesting protein because it was made of components, and the components are found in single-celled organisms. But when you put those components together into one protein, Notch is likely the protein which allowed for multicellularity to come about. So philosophically, it has all the components that all other organisms have, but if you rearrange them into the right way, you're actually able to get this enormous explosion of function of life itself. And so taking a little bit of inspiration from that and saying, what if we could decompose or, you know, deconstruct the proteins both of like Notch but these other proteins, as well as from the rest of the proteome, could we rearrange them and create new proteins that gave human cells new functions? And again, I was watching that happen over and over again, watching, you know, simple things like GFPs and tags and HaloTag and ClipTag be used to these effects. And it was very hard to say what's the best one, where should I put it, how should I do these things. And these kinds of protein design had nothing to do with a structure of a protein. It was very much trying to figure out how do I build and make the cell have the function that I want. So we built the software to catalog the components, to make the components easy to understand. You know, GFP, rather than having this long enormous sequence that is not readable by a human, we gave it a little blob that was green because GFP is green fluorescent protein. It should be green. And we didn't know, or we don't know as humans what many of these proteins look like because these are all the proteins that don't have crystal structures. So we would just give them icons and we'd give them unique icons and that gave it a game-like creativity where you could start to— when you're a grad student, you build your little presentations and your posters, you empathize with the creatures that you've made, even if they're little proteins, and you spend 6 years playing with these little creatures. And if you can make them empathetic and you can enjoy them and you get attached to them. And I felt that, you know, one big circle representing your protein you spent 6 years on didn't give you the kind of empathy you needed for something you spent so much effort thinking about. So we tried to give all the little components some kind of feature that we could work with, and that helped us build and design and catalog them.

Okay, cool. This context, Notch is a very important developmental pathway. You like learn, and Bruce Alberts teach you that or something, I don't know. And Notch is a receptor and a delta is the ligand, and it leads to a lot of signaling. And then you can create synthetic versions that lead to different orthogonal signaling pathways in cells. And they called synNotch, that was really big at UCSF at Wendell Lim's lab, and that, you know, kind of was a useful, uh, for various applications. And just move into then, like, kind of this idea of designing multi-domain proteins, alluding to, right, these big proteins have multiple domains, really large, hard to crystallize, and you want to create synthetic versions. Oh yeah, you create a whole different catalog of different proteins that then are almost impossible to crystallize at scale. And so you kind of need a new language to analyze these kind of large, complicated proteins that have a lot of high value. And you kind of talking about this too in terms of labeling and creating creatures. Okay, that's cool. That's a cool kind of cool analogy. How did you, when you were starting Pinecone off and talking to customers and thinking about the problem to solve for, especially this big problem of designing a multi-domain protein, that's what Srotini is doing, right? You guys design multi-domain proteins using machine, what's it called? Machine-guided evolution. That's your term. It's a good term. Right? I love it. And so how did you think about initially that common language for designing these type of proteins?

Well, what was happening was Colin and I, we'd sit on the couch and we'd be working on the software, but the data that goes into the software are those proteins, they are those machines. And every time a new paper would come out, we would go and try to find the sequence, which was stupidly hard sometimes, but once you found the sequences, you realize this fancy new paper that gave a fancy new name to something is actually just recombining 3 objects, putting those 3 together. Okay, cool, put that into the software. And these just kept coming out. Every week you'd have new tools, new, new, new objects, new proteins made in this way. It became clear that there were a few that were financially or therapeutically useful. So there were a lot of proteins that were being made for basic discovery, and we actually did a lot of work there, and we had some really great collaborations. But as a business, it became clear that there were some of these where if you could create new ones, you could, you could actually help out therapeutically. Specifically, the two real big categories were the gene editing tools like the base editors and the prime editors, all those things attached to Cas9, and then the chimeric antigen receptors, the CARs, and for going after cancer by changing the way a cell behaves. And in both cases, you could simplify the fancy new tool that undergirds billion-dollar companies to be a set of protein domains that have different functions. Cas9, the piece of CRISPR that allows us to edit a human genome, has two main functions. It can seek and find a specific DNA sequence, and then it can cut there. Turns out the seek and find is actually a little more difficult and a little more interesting, and you can change it real quickly just by making new RNA. The cutting actually is not so desirable. You don't really want to cut DNA. So then it became logically— the next step is to figure out, well, what do you want to do when you get to a particular sequence that you like? You don't want to cut, maybe you want to change, or maybe you want to recombine, or maybe you want to integrate. And so each of those ideas now spawn an effort to go find the protein domain that will do that job. And now you create a new tool. Uh, and same thing's happening on the, on the CAR side. Something that binds to the target that you care about on the extracellular side of the cell, pass that signal through the cell, and on the intracellular side, uh, tell the cell to do something. Yeah, in the common case, it's to kill the CAR or kill the cancer cell.

Yeah, one thing we always thought about, like, for small molecules, we can screen maybe a billion different variants. Antibodies, maybe millions. But for like something like a chimeric antigen receptor, hundreds if we're lucky. Time, or CRISPR, maybe a little bit more than that. And, you know, so serenity is kind of helping bring multigenome proteins to parity to other modalities, because if you can do that, you create different variants, you can test them more easily, and you can then, like, you know, have— you can potentially discover new medicines at a faster rate. And that's kind of the key, kind of, at least for me, what got me excited about serenity and what you were doing, kind of, kind of accelerating development of these kind of new drug modalities by building a kind of a more generalizable platform and language. And so you have this SaaS product, you can design these proteins, can annotate them and create new variants, and then you make a pivot towards drug development because all in all, building SaaS in biotech is too hard sometimes, most of the time, right? You have to be the exception. And so what was the logic to then say, okay, screw just building software, we need to make our own drugs, we need to actually generate our own data internally. What was the kind of the moment that made you, or maybe set of moments that made you make that pivot? And then that can, then we can talk about then kind of how that set up your kind of current trajectory and current kind of updates on the underlying platform and business.

Yeah. You learn a lot by creating a company and jumping into the fire and trying to keep it alive and to make it survive. So you end up learning truths after a while, where after a few months or a few years, you can distill it down to a sentence, but it takes years to learn that sentence. The therapeutic industry or the biotech industry is a strange industry because you have very few people who— there are very few purchases or transactions unless you're literally selling the drug in a hospital. You know, many biotechs sell a single time to a bigger company or to it as an IPO, and so the economics in the market looks different than it does for somebody like Adobe or even Uber or these other companies. So selling software that is especially very specific software, the total number of customers that you have access to is small. And then you have to think about what those customers are selling. It turns out a lot of times if they're selling a single therapeutic asset once or twice, then if your software doesn't, doesn't push that binary sale to be more successful or less successful. It, it, it doesn't round very well. So it became clear that the valuable thing, if you can move up the value chain towards the actual IP, towards the actual composition of matter, the actual protein itself, you can collect a lot more value. So we, we, in order to, to get get there, where we would start to be building our own designs, we needed to get more data. We needed a wet lab. We needed to demonstrate that what we were doing actually got there. And that in itself is a challenge too, where you— it's what's so clear to you may not be clear to others, and you have to go and collect the capital in order to collect the data in order to prove what is— what you think is pretty straightforward. And we had a really good team of— we raised venture capital to do that, a really good team of investors led by Nano Dimension and a really good team of employees to really get that first data to show that we could not only design these proteins in high throughput in, in a software sense, but that we could actually build them, manufacture them, actually get them into a laboratory and start to test these combinations of tens of thousands of these different designs simultaneously.

Cool. And I think at the time when you kind of made that pivot, you know, I think two of the CAR-T, you know, programs got approved by FDA.

That helps a lot. It's a, it's an enormous financial demonstration that what you, what, what could be built is valuable. So yes, it's not a coincidence that, that the approval of the first CAR-T therapies actually enabled that kind of forward-looking success.

Yeah, I think a lot of Serenity stories kind of tied into kind of, you know, trust between you two and like kind of really expertise, but also timing, being a little lucky. Some other events happen and that validates your technology and your approach and kind of, I think it's a lot of startups, just kind of lucky events and they kind of taking advantage of those events and riding the wave. And so you pivot to drug development. What were kind of the— how did you think about kind of building out a wet lab? How did you think about key metrics? How did you think about kind of what types of datasets you need to generate to build better models? That's always a tough problem when you start kind of these— I don't know what to call them— comp bio drug companies, AI drug companies, what do you call them? But you have to ultimately generate your own data. What kind of frameworks, how did you think about for Serotini, what types of data did you need, what type of datasets you need to generate, and how do you scale?

Yeah, I mean, it's, it's, you have multiple pressures and multiple problems you're trying to solve for simultaneously with never enough capital to solve for all of them. So in that regard, We tried to build datasets where every single time we did an experiment, we tried to make sure that we were pushing our platform forward, we were bringing data into it, that we would learn something that we could use later, as well as something that actually could be a shot on goal, something that actually— you wanted to make sure that there was always a chance. And sometimes it's lower and you're more on one side than the other, but there was always a chance that what you were building or what was in your library was a new useful molecule or new useful protein. So the kinds of things we would do, we knew it was kind of a blue ocean to go looking for these designs since much of synthetic biology at that point was kind of done one at a time, that the serendipity that you're both harvesting for the company itself You're trying to find those rare events where you get the timing right, but you also can do that scientifically. So we would try to be unbiased about what we'd look for. And we would go mine inside the human proteome for new components that we thought would be interesting, even when there was very little known about them. There was little literature. And then you try to quickly take them forward. You try to quickly take your hits, and you try to quickly make sure that what you're doing is therapeutically relevant. And so we spent a lot of time thinking about the assays that we were doing. So it was not just the cool design and engineering, but how do you show that the engineering that you're doing actually can help somebody? And so we worked with a lot of people, a lot of mentors to make sure that the tests that we were running were something people would care about.

Cool. I think shout out to Brian Natin. We did a podcast with him a long time ago. Shout out to Brian. I think Brian was pretty important. Brian was really useful for that.

Brian's a good guy.

But here's, you're talking about two things. You said some of the most insightful things you said. You said two things insightful to me. On the first part too, about making that pivot to drug development, I remember you told me, you know, drug development's like an economic black hole for biologists, right? Where it's like, if you're doing something cool in biology, you're kind of getting sucked into making a drug somehow. It just is what it is. And I think on the platform side, kind of talking about how each experiment enables the next one, it's something a couple years ago, Emre told me kind of this idea of versioning proteins. Yeah. And that's such a powerful framing, just as much as you version a pair of shoes or iPhones, you version proteins. And for Serotiny, how do you think about going from a version 1 of, say, a CAR construct to get into version 10? How do you think about getting there quickly? And how do you think about like, like, um, you know, going from like generating a first construct and then like thinking about, okay, um, how do we generate even better one? What, how do you, how do you, how do you even get there? How do you even, what kind of logic do you use? Um, is it, is it more kind of interpretation of data or do you implement some sort of systematic, systematic strategy to version things quickly?

Yeah. Yeah. So, uh, at Stratoni we do, Both. We go from 0 to 1 and do the invention, and then we go from 1 to— and then increment the versions. It was a fun thing to watch. So again, David Liu is the professor who's invented the base editor and the prime editor, and those, by the time he published his papers, already had their versions— version base editor 3 and 4, etc. And we watched that and noticed that it was obvious, it was clear that he was taking code and he was improving it. He would improve the enzymatic efficiency, he'd improve the way the proteins were fused together, the homolog or ortholog of a particular domain that was better suited for what he was trying to do. But the idea of the core had remained the same. When it changed, it got a new name. So we wanted to do that. So in our context, you actually change the assays a little bit, If you're going from 0 to 1, you know, you're looking for does it do it. And so that's much more of a screen. Does it turn on at all? So you can throw an enormous variation into a pool. And if you've created your assay correctly, all you're looking for is any signal at all. Just does it turn on? That's pretty fun. When you get a hit on that, it's pretty exciting. But always, you then need to improve it. And that's where part of our platform that's nice is it's not just a screen. We actually make sure that we know how each design behaves. So we can say, let's go, let's make it better. So we can put in controls, clinical controls, or our own controls. We can make an enormous amount of variation around the island that you've already found. And you can start to, if you find these islands of stability in protein sequence space, then you need to Once you get that hit, you then need to circle around that island to find the spot that is most efficient.

On the invention, it's just measuring, does it work as we want it or not? Does it lead to some sort of signaling or killing, or does it cut DNA or edit DNA in a certain way? Then go from 0 to infinity or something, I don't know. Um, 0 to— go from 1 to infinity, go from version 1, like, you know, the best CAR-T, CAR-T can invent. Um, You know, how do you do kind of machine-guided evolution, as you know, Serrani does? Is that more of an ability to do more experiments cheaply, or is that a function of the number of metrics you capture, or maybe a little bit of both? Like, how do you— how do you actually— like, what are the enabling factors to go from version 1 to, you know, version whatever?

Yeah, so to play on the term not too strongly, but the idea is you have variation. You try to create that variation, and then you have some selection pressure. You have some assay where you, you read something out. In general, the proteins we're working with are big, and the cycle time can be long. These are mammalian cells, and you can't shortcut a lot of that. So we would always love to improve our cycle time, and we're doing that right now. But our variation is enormous relative to other ways of building these kinds of proteins. So in many cases, the multi-domain protein itself is a very deliberate effort by usually one or a very small number of really smart people, really trying to do their best to predict what the best set of combinations of these components will be. But there are an enormous number of combinations of these components. And so what we've tried to do is create a— you can throw in any idea that you might have a chance of working, but we actually let the system choose which of those many, many, many medium to low percent working hypotheses will actually surface. Absolutely.

I think the major breakthrough for Rodney's platform versus maybe the state of the art is like, you know, state of the art card testing is maybe hundreds of constructs, maybe thousands. And in serotiny, you can go from millions of designs in silico and then test tens of thousands in vitro. And then that will then lead to, you know, I don't know how many designs, hundreds or tens of designs. But then ultimately you have data where you can show, you can go from millions of designs in silico to a functional molecule with, you know, better properties of maybe killing or efficacy, whatever. Engagement, target engagement. And so when you think about creating that variation, what enables that? Is it just like, is it the software? Is it cheaper? Is it just kind of the way you design libraries? How do you, how do you like think about maximizing variation? That seems to be the most important metric for that second step of versioning your protein.

That's right, so variation is important. One place we seek variation is in existing protein designs from existing proteomes. If we're going to make a drug though, we often try to stick with a human proteome so you don't cause problems. But the human proteome is pretty big, and most of synthetic biology has not explored that. And then that's coupled— looking for variation amongst the human proteome is coupled with the ability to just synthesize the DNA for these things. And when I was in grad school, synthesizing DNA was extremely expensive. You could do it though. And you really could, if you had a design you thought would work, you could spend a couple thousand dollars to make the design. But with new cloning techniques and with the reduced cost of DNA synthesis, you can start to be clever to actually produce thousands of designs or tens of thousands of designs by recombining some of the sequences of DNA, even if they're very long, not just mutations but including complete orthologs to each other. And that allows you to build variation very quickly. So, you know, there are just over 1,000 Type I transmembrane domain proteins, or Type I transmembrane receptors in the human proteome. And these are just your regular old protein that has something on the outside of the cell, crosses through the membrane, and it's got something on the inside of the cell. So it allows some signal to pass into the cell. And there are 1,000 of them. You can start to synthesize most of those either in components and reuse those components. Can't do it one at a time, but if you can do it in high throughput, you now have access to that whole proteome's worth of tools from a couple billion years of evolution.

Cool. Yeah, I think shout out to Twist for a lot of that work and a shout out to— is Gibson assembly still used? I think so, yeah.

Use the best tool for the best job. So there's lots of good cloning techniques that help us. Twist has been a useful partner in helping us build some of these libraries.

Absolutely. And so I think it seems like the software enables you to create all these invariants, and then these new synthesis tools, cloning tools, allow you to test them out. And kind of merging software with new tools allows you to sort of create kind of a whole new platform that maybe wasn't possible 5 years ago or 10 years ago. And so can you talk a little bit more about then about like kind of, kind of you had this workflow to go from maybe an idea, a prompt set to then IP or a product. Maybe we can talk about the, the business model now. Maybe one thing you could touch upon like, like, you know, let's talk about business model and we'll talk about data afterwards. But how do you, how do you think about Serenity's business model? Because you have this, you have this kind of, you start a software company, go to drug development, and, and you, you build out this platform that is really good at versioning proteins, right? Inventing and versioning, right? Uh, do variation of, of, of, of, of, of, of CAR or CRISPR. How did you call and think about business model? Because you definitely are doing something a little different. And we kind of talk about some of the deal, recent deals you've announced.

Yeah, so there's one last step there. So the last step is you have to use the platform in a therapeutic or in a, in a financially useful way. And so initially we built out some proof of concepts where we went around and tried to figure out what was the best way, the best assays, the best goal for therapeutic cell therapy. But it was also useful to us to partner with strategic partnerships, people who could provide us synergies where they already have some component that they have characterized and they really understand, or they have some technology that allows them to already understand that, that goal, the therapeutic goal. So to be clear, Srotinium was started as more of a synthetic biology company, that's been our expertise at building the platform. And we've actually brought on some, some, some great people who have that therapeutic expertise at this point. But in the beginning, that was where we thought we could partner with those who already have that therapeutic clarity. And the two places where this has been obviously useful are, again, the gene therapy context or the cell therapy context. With the gene therapies, one of the big goals is to be able to insert sequences into DNA with an arbitrary payload, make them how you want. And so we've been working with Testra Therapeutics, a really powerful flagship company who has found some really interesting ways to get these gene therapies up and running. And so we're going to work with them to, to, on their multi-domain proteins for gene insertion. And we've also, on a different side of things, worked with, just have a relationship with Janssen Therapeutics, Johnson Johnson's R&D side of things, where we're going to help them build cell therapies with improved and novel receptors for, for these cells. So, and in both cases, we're able to talk about that process that you talked about, where you start with kind of a conception, idea of what you want to achieve. You build out an enormous library of variations to test as many different hypotheses as you can all at once. And at the end, you're left with something that is a therapeutic therapeutically useful multi-domain protein.

Cool. Yeah, well, congrats on the partnerships. I'll put the links into the description of the deal announcements, and I'm sure there's a lot more. I know there's a lot more in the pipeline left. And maybe you have these partnerships, that's one part of the business model. You use your platform and you go to a partner and say, okay, they want to do something in a specific problem set, whether it's in cancer or maybe a specific modality. Or how do you— and how do you then think about sequencing those deals and stacking those deals. That's kind of the art of biopharma dealmaking. Maybe Colin spends more of his time on that, but kind of the art of the deal. And then how do you think about external versus internal versus something we'll do partnering with versus something we keep internal to Serotiny?

Because, yeah, so there are a few features to that. So one is In our case, what we have is we're starting with a platform, and we'll acknowledge that straight up. This is a platform technology. And I think there's always this push towards, you know, reversion to the mean where platform companies want to make therapeutics and therapeutic companies want to have the ability to make more than, than one by showing that they have some kind of platform. And so when does the therapeutic company build out its own pipeline? Versus when does the therapeutic company branch backwards. So that's one feature. The second is the state of the industry. So if you build out a platform company, you need clients, you need partners, you need customers. And so it's been fun to watch the industry see the value in the potential of these cell and gene therapies. And we've watched that happen before our very eyes, and that has enabled these valuable partnerships that, that are more than something like a software as a service from the early days. It's clear that, that sometimes the, the product itself can, in the therapeutic context, can be valuable, and I think that our platform is close enough to the therapeutic material that we can start to push in that direction. And we've brought on the expertise to help us do that, and we're looking to, to, to build these synthetic biology tools that control cell behavior in a way that improves these therapies beyond what currently often were built as one-offs and they're sufficient, but we think we can really improve upon them. And the space is big. There's a lot of things that a programmed cell or a programmed protein can do, a lot of different targets, a lot of indications, and whole new sets of modalities that are enabled by this kind of programmed biology.

Absolutely. I think, yeah, I think partnerships help scale platforms in a less dilutive way. You know, dilution, and then it's really important, uh, my worst enemy, um, and, uh, you know, my worst enemy. I, I have nightmares about dilution and nightmares. I have a cold— I wake up in cold sweats about recaps and, you know, stuff like that. Uh, but then, you know, on the opposite side too, once you get scale, you can then do things, do experiments that no one else can do. Yeah. Uh, and it's kind of the data for serotony. It's exciting. You kind of shown an ability to invent a new construct, version it, and then generate some really exciting data internally. And so I think, I think the world will begin to see all the stuff you've been doing the last few years and begin to see, appreciate, well, once you get scale, you can start doing things that maybe you could screen a larger number of variants, and that can lead then to kind of constructs that other people just couldn't even discover at all. And so enable new types of cell therapies and a lot more.

Yeah, there's a— back to the kind of timing issue, you— it's nice to work with technologies or to ground our own work on technologies that are themselves scaling. It's a little scary because you don't want to be caught where you need that technology but it hasn't scaled yet, but it's a really nice feeling where you can ride the wave of of the technology itself. So software helps us. That's already scaled. Everybody knows how to scale that. But it is scalable. So we can scale our software pretty well. The synthetic biology side, again, is scalable. And it's taken a while for it to become as cheap as it is to synthesize the DNA. And the sequencing itself also is scalable. You can crank that up. You can parallelize it. You can do many projects all at once. And so that is another feature of kind kind of the platform that allows you to take on lots of different projects and not just have one, all your eggs in one basket.

I think it's a really nice place to be, to be a market leader in an emerging market, emerging category. I think Serotiny is the market leader for the design of multivalent proteins, right? Just the data shows that, and you have the technology tools to do that. And so yeah, I think as cell therapy is more successful, serenity, you have a lot more success. And so, okay, you talk about serenity, talk about your kind of the platform, the business model, kind of any outstanding problems, you know, as you kind of built this company up, but that just more generalizable to other companies. You know, one thing I see in all these companies, this idea of like how to actually generate the right data for models, buddy Jacob Oppenheim, EQRX, you had this concept of doing the unphysical experiment to set the boundaries for models. Do you do the experiment that you wouldn't really do that doesn't create value, but it's valuable for a model to kind of know the boundaries? Like any problems like that, maybe that's a problem, but any more generalizable problems that you're kind of experiencing at Serenity that like maybe other companies are facing and kind of any ideas to solve them or, you know, because like maybe startup ideas and stuff like that.

Yeah, I mean, I think that it is important to be able to communicate your results to others as much as it is to be able to generate really cool science. There's always a little bit of a tension between doing an experiment that's obvious, that's easy, that will be less incremental versus doing kind of the right experiment the big way that's going to be expensive and require a lot of resources, and I don't actually know that there's a right answer to when you have to choose. Do I go all in on the one experiment that's going to be the big one, or do I do a few smaller ones with the budget that I have right now? But it's a choice that you constantly have to weigh. It's— and you have to make those choices not just based on the good science, but on considerations of communication. You have to be able to communicate to your employees and to your investors and to your partners. And so there are really good examples of creating data that helps you get a deal or helps you get fundraising or helps you hire even that, that allow people to see inside of what you're trying to do. Interesting. And so I think it's one of the lessons.

That's pretty insightful, actually.

Actually, thank Colin for that one. Colin and I spent a lot of time chatting about a lot of these ideas. So, even me, two brains here.

You have this idealized thing of platform companies that you're supposed to generate these huge datasets and, you know, and be comprehensive. But then when you're in the weeds of building a company, sometimes you need to generate the data that is valuable at the time. You kind of balance the act between doing something that's systematic versus kind of what's kind of needed. Any type of anecdotes or stories from Toronto you can like discuss that like maybe are like that? Nothing confidential though. If it's all confidential, we can just go beyond it, talk about something else.

I think it's been fun to— a little related but not entirely related— when you get to be unbiased and when you get to to kind of let the platform that you have shine and you trust it and you have to trust it because you don't know what— at the end of the day, the biology wins, the science wins, reality wins. You don't get to tell it what to do. But when you say, look, I think there should be something here and you run the experiment and you pull something out, it's often pretty exciting that you get a hit or you get a result that matches your theory. And what's really fun is when you go back and you try to figure out what's going on here, you realize that there was ambiguity in the literature or the literature would never have predicted this. This would have been way down on the list of what would have been expected. But that by trusting the process, by doing the right experiment, by letting it, the science work itself out, you found something in a way that other people couldn't and you found something that's impressive. So it's a good feeling when you do get to run that big experiment that really proves your point. Cool.

Okay, great answer. One thing too, like you and Colin, you guys progressing as leaders where you guys, you know, I can see you guys as public CEOs one day. You guys have good answers. Like, yeah, you're getting the public CEO persona.

I love it.

All right.

Can I ask you a question? I don't know if I'm working on that, but— Example?

And you just gave me some cool kind of overview. I love it. Maybe talk about more about your transition As leaders, last thing to talk about, uh, you talk about, uh, you know, one thing I love about you two, you guys are such empathetic leaders, incredible managers. I've seen you kind of really be thoughtful at team building and who to hire, how to construct culture. Um, what have you learned as going from a, you know, grad student to build Serenity as a SaaS company to all the way to leading this huge team, recruiting executives? Any kind of lessons you learned from becoming a better leader? Because I've definitely seen it, right? You and Colin have become definitely better, better, better leaders, and then you guys are going to be getting better. Any kind of key lessons you learned on this journey so far?

Yeah, I mean, it's a different job. So as a scientist, you're trained to do science, and that's one thing. When you jump into the deep end, you have to learn these other things. You have to want to do these other things. And you don't even know how valuable they are until you get into them. The business side of things is really important. BD is a whole different language. Leadership is a whole different skill set. And I think it's, at least from my perspective, it's been helpful to be humble. It's been helpful to acknowledge that I'm new at this and I'm learning too and actually have people We've had great employees and friends and mentors who have coached me and said, hey, hey, hey, you can't be doing that. We got to be going this way. It's like, I see, and I appreciate that. And that helps a lot. As a— it's a weird transition, a very strange transition to go from being the person writing the code to helping other people who are much better at that than you, give them the autonomy and the authority to run with that. And it's really fun when it— when you get to see a whole team working and you get to see so much productivity that you could never do yourself in because you've been able to bring such interesting people together. So I think being humble is really important. I think learning from your mistakes and learning from your lessons and acknowledging that you'll make those mistakes is really important. Um, and making sure everybody is on the same page and has the right incentives is very helpful. Um, and we'll keep learning. Uh, I don't think I'm done with my lessons yet.

Uh, but anyone— people just like being around you. I think it's probably something that's underrated too. I think people just like being around Justin. That's what I'll say. I think a lot of people just like hanging out with you, working with you, just pleasant person. And I think one big moment I observed, a big definitive milestone in terms of that leadership transition, was when you and Colin got new headshots. Okay, that's new headshots. Looking good, guys. I got new—

the publicity part, we're also—

I got new headshots, right? And it was a big, it's a big, big transition in my career when I got new headshots. I think it's a big moment for anybody's career. Is when you finally get a professional headshot done, it's like, okay, now you're in a whole different trajectory, whole different era.

You don't wanna do it too early, but you also don't wanna do it too late.

So, well, yeah, one recommendation, everyone, get professional headshots as early as possible. Yeah, it's something I didn't even know. I just took a picture, it didn't mean— go spend the $500, $1,000 to get a good headshot done. It's invaluable. So one leadership lesson too. And then could you talk about maybe like, last thing in terms of like hiring, the part of this leadership part, you know, one thing I'm observing is the biggest bottleneck for startups is no longer money, it's just team building, hiring, kind of what were kind of some tactics you've used early on to hire, recruit, you recruited a, I forgot the CSO's name from Kite, Maverick.

Maverick.

Yeah. So you've recruited some really top-notch executives. What were kind of some tricks and learnings and lessons you've learned from hiring, which is just so hard for every biotech company, no matter how much you've raised, whether you're seed or you're a big mega company, hiring is so hard.

Yeah. There's a lot of interest in talent right now. And there's a lot of people, especially on the science side, who may not realize quite how valuable they are to the industry right now. There's just such a need for people with skills to do things like mammalian synthetic biology that traditionally weren't a job that would pay very well. I think it's important to know what you can offer as a company and be clear about it. And there are some things that we cannot offer as a company, as a small little startup, and be clear about that. And if you're clear about it and other people know what they want, you can arrive at a place where everybody, I think, can get what they want. But if you're not clear about it or if you try to compete against the big companies for— with resources that you don't actually have, I think you can run yourself into trouble. So we try to be very clear. I think tapping into social networks as much as possible, even though that can be hard, is really helpful. Your friends and recruitment and friends of friends is often very helpful. Before COVID giving talks, and I guess that's coming back now, which is great. Seeing people in person, meetups and things like that can actually be a really nice way to meet new people and get out of your normal comfort zone. Totally. It's a hard thing to do.

Very important though. One thing I'll say, what I love working with you, Colin, and I think a lot of other people do, is You're high integrity. So people know with the Farlow brothers, you know, high integrity. It's like a, you know, you know, you trust them, they're gonna tell you, you know, gonna be transparent about what they're doing. I think another reason why people like are attracted to Serenity is just like it's very high integrity organization. I think you've built that in your culture. And I think, you know, this podcast and a lot more in the future is, you know, you, you're already a superstar. Now it's about unveiling that. It's no, you know, very easy work. So I think the next few years, you know, you're going to be on some more panels and jet setting to— I don't know, what if some JP— you'll be JP Morgan. I'll go JP Morgan.

Okay.

Be on some panel, you know, a lot of panels. We'll be on, uh, famous, famous co-founder, and you'll be on, uh, CNBC or I don't know, TV shows and stuff, uh, Bloomberg or whatever. I always want to be in Bloomberg, actually. Let's do Bloomberg one day.

All right, well, I appreciate your optimism, Josh.

I'm so optimistic. You know how optimistic I am. I'm so, I'm so positive.

It's great. No, it's wonderful. It's, it's a, it's a good quality.

Uh, I'm your, you know, I'm your movie agent, so I got to, I got to be hyping you up. So, uh, uh, okay, we're going to publish this. This is a Friday. I'll send this on Monday, and, uh, you know, we'll get the hype started, hype cycle started. We'll do this again I appreciate doing this. Let me— I'm gonna press stop record right now, but, uh, and we'll talk.

Thank you very much, Josh. I really enjoyed chatting with you. Uh, it's, it's— I, I think, um, helping the, the community understand, uh, what's— how these companies are built is helpful. I think there are a lot of people who are interested in it. I think the community is still pretty small, and I think you've done a great job of trying to get that, that, that communication Appreciate it. So I appreciate it.

And this is going to be— this is going to be useful like 10 years from now. Like, they're gonna be like, whoa, we're gonna unearth—

it's gonna see how much we were wrong, how much, how much—

no, there could be so much. You don't know what we were doing. You know, take the Disney Vault, they're gonna be so inspired by Justin Fallow's story, and we'll do a follow-up. And you know, this podcast will last a thousand years. Aliens will come down and listen to it.

I'm scared about that, but I do appreciate— I would like the ability to come back in a and a little bit and, and see what else we've learned.

We're gonna do it. Okay, I really appreciate it. Let me press stop record right now, but I, I appreciate this.