High-Throughput Chemistry and Engineering with Kadi Saar

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. Hey, Cady, how are you? Great to have you on the podcast, and I'm just really excited to talk to you. It's about your career and research on everything. I think I found you a superstar. So how are you?

Hi, hi, Josh. Good to, good to talk to you. I'm doing very, very well today. How about you?

Just kind of a busy Sunday. So this is, I think it's a Monday. So you recorded.

So I'm up in the north where the heatwave has just, just arrived, you know, the 2 weeks a year when it's here. So I'm very much under the heatwave at the moment. But other than that, very good.

In England?

I'm physically located in Estonia at the moment where my family is. I think England just had a heat wave. Yes.

Jeez Louise. In San Francisco, there was like 100 degrees one day and it was like so brutal. I had to go work out and I felt like I ran like 6 miles that day and it was just brutal. But let's talk about you and I just find you so fascinating, your whole, just your whole personal journey and then your research. I'm a huge fan of your research in Alpha Leads Lab at Oxford. And then like, just really excited to see what you do. Is Alpha Leads in Cambridge? He's not Oxford or?

Exactly, he's—

He's Cambridge, he's Cambridge.

Actually, he did his PhD, I believe, in Oxford, yeah. But now, as you know, I've been working with him.

Okay, cool. This is, I'm so glad we have video on. I just saw your body language. I saw your response. I just saw it. I just saw it. I just thought, oh, I said something wrong. But maybe to start off, maybe we can talk about how you got involved in science, how you got excited as a kid, and then we can go from there.

Sounds good to me. And thanks once again for inviting me here. It's really, really great you're doing this series.

OK, so, Cat, how did you— what's the most important question I like asking people? What got you excited about science? When was that moment, or maybe moments?

Interesting. I'm not sure I can actually point back to one, one moment. I think I probably was one of those kids who was just always kind of excited about science and technical subjects and the world, world around me, and then just a curious, curious kid. So I'm not sure I can point down to one moment. I think my interest just grew over time. I come from a family where both of my parents are trained as medical doctors, so maybe that to some degree also influenced the fact that I kind of, you know, science was always around me. But when I went to school, I noticed myself being quite good at math and technical subjects. And then when, you know, other scientific subjects like chemistry and physics came around, I kind of felt like they were just another another, another application of mathematics. And I think that's how I got interested in this. Yeah, I don't know if I have anything else very, very original to say here, but yeah.

Did you have any— do you have any books laying around? So like my mom's an anthropologist and so I would have books laying around as a kid. I would just read the books. But you had certain things like that were like in the house that just like, you know, your parents just had as like a throwaway, but then you just got rope into it somehow, um, or maybe, maybe just you're a genius and so just, you know, Cady's a genius, your parents knew it, let Cady do her thing. And do you have any hobbies as a kid, like collecting things, or, you know, do we, you know, any, maybe that, that too?

Uh, interesting. I think we did have lots of, uh, books around and they probably felt like such a natural thing to have that I I didn't even pay any attention to the fact that I had them around. But you know, if you're even thinking about back, like, you know, the TV that people have on as parents when you're growing up, I do remember there were oftentimes like trivia or quizzes on them on TV, right? So we wouldn't have like, you know, completely, like, probably we wouldn't have that many kind of, you know, reality shows or films on, but like kind of, you know, educational background on. So I think that's exactly probably why I I can't point back to one moment, but I think it was just always around me, wherever I was, and always, you know, access to information through my parents that whenever I would ask a question, they were there, they were happy to answer it, and so on. Yeah, so kind of, yeah, again, nothing very specific to point at, but I think the full environment that I was in was just kind of in a good geared up for doing science on some level. Cool.

I think in America, at least— I'm going to talk to your parents. Maybe your parents will write a book, a parenting book, and it'll be a bestseller in America. And your parents write some book about how to raise a kid, and like every parent in America will buy it. But okay, so let's take a segue into the reality TV show. So you were on a reality TV show, Estonia's Most Talented, or Estonia's Got Talent. How did you— I think your talent was your arithmetic machine, or is that the term? How did you get on the show? And how did you realize you're so good at multiplying numbers?

Yeah, I think actually here, probably what I need to thank is the Estonian educational system on some level. But you know how I ended up at the show, that's I think kind of by longer story in the sense that when you go to school, Estonia is kind of organizing a lot of, I guess, competitions for various different subjects. Some of them are in math, but you also have competitions in, say, linguistics where, you know, people across the country are coming together and are solving exercises. And most recently, you know, almost any humanities subject like philosophy and history as well. It kind of started with more scientific subjects. And then, for instance, in math and the mental arithmetic that I was doing, like you know, across the country, I don't know, half of the kids maybe would be taken to a computer class because back then kids didn't have computers at home, right? And you'd go there and you, you know, work with this program and try to do the math. And I think the first time when I did it, then, you know, I'm getting the rankings and the rankings are showing to me like, you're number 3 in your country. And I was like, oh wait, I'm just doing it for the first time in my life and I'm that good. Maybe I should put more effort into practicing this. And that was when I was like maybe 8 or 9 when it first started happening, and then I participated at the competitions every year and so on. And then interestingly, when the TV show Estonia's Got Talent got started, they had, you know, lots of interest from across the board, singers and dancers and so on, on coming there. But the organizers had kind of, you know, heard of this mental arithmetic competitions that kids were, kids were doing, and they got in touch with the organizers and were asking like, is there Is there anyone whom you would recommend who would come to the show to like spice it up a little bit compared to what we have at the moment? And somehow they are going to organize— actually, he first recommended my brother's name, who was really good at this as well. But my brother was like, no, I'm not going to go into the TV like some reality show, not for me, probably exactly for the fact that they were never playing for us at home, right? He was like, what is this reality show? But I was like, well, if you don't want to go, I can go instead of you, right? And yeah, I went first the preliminary round and then they did the final afterwards and it worked out very Very nice. And I was just kind of amazed that people were interested in a skill like this because you could also send your votes for, you know, singing and everything else that there was as part of this show. And what I found even more interesting perhaps was the fact that the TV show itself took place on the 30th of December as I was doing my math skills, and then everyone had a chance to call in and vote for whomever they liked, and they ended up winning it and getting the prize. So, in this Got Talent, it was the first first essentially series or first year when they were doing that TV show all together, so the inaugural one. And then about 5 days later, when the student president was giving a speech, like, he actually pointed back to this example when he was saying Estonians have always been a nation who has loved, who have loved education. And that was, for instance, amplified or exemplified when just, you know, 5 days ago the public voted someone who was using their brain as the, you know, top talent of the country. So it was just really, really great to see that all the country actually liked it and the appreciated it.

Okay, you have groupies? Okay, cool. You have like, probably have super fans of you. So like, I don't want to put you on the spot, what's like 978 times 100? You know that?

Can you do that? 100, yeah, so just add two zeros and that should be it.

Exactly, I don't want to put you on the spot. I don't want to add, I don't want to give you some complicated number that you have to like, maybe you lost your edge. But maybe not. But maybe we've established that, Cady, you're special and in a good way, and you're really talented. This is my opinion, my opinion. You're a very special person and very talented.

Everyone is special just in their own way, but I guess that's a longer discussion. But that's a fact. Everyone is.

It's okay. You have this kind of interest in mathematics and science. And maybe you can talk more about your journey to Cambridge and how you got to the Li Lab and how did you start doing work in computational chemistry in particular, given you have a skill set, you can do anything in my opinion. So how did you get involved in Cambridge? How did you get involved in research and what was your journey towards where you're now, at least from a scientific perspective?

Yeah, very, very good questions. And I actually didn't find these choices easy at all. I think When I still think back to the days when I was at high school, I was considering professional sports as a potential career as well. And actually, when I was on that same Tony Scott talent show, I think one of the aspects that maybe even helped me to win, or, you know, somehow made my strong even case, was exactly the fact that I was in parallel a professional sports person as well. And then you have this human calculator whom everyone was expecting to be an old man turning onto the stage, but then you have this like 15-year-old girl who was actually 2.

Okay, we're going to link the YouTube video. When I saw the YouTube video, it blew my mind. So we're going to link it and so you can click on the link and watch Cady multiply numbers. It's incredible.

Maybe, but I mean, I did put the organizers on the spot as well by saying that there was a human calculator coming in and then there wasn't a gray-haired old man.

No, it's a little girl who shows up. It's crazy.

Anyway, so going back to your question, Sports was always a very big part of my life, and I was very certain I will just go to the US and play in probably an Ivy League college where sports would help me to get in and do this. But then I think I kind of started asking myself the question that, well, will that essentially postpone the time when I need to make a choice if I want to invest more time into science and technology and math and my essentially schoolwork? Versus investing that time in sports. That, you know, it's really great because I could do these things in parallel. You can't do that well in— that so well in parallel in the universities in the UK. And I told myself, like, you know, it's really attractive at this moment, but, you know, 4 years down the line, I will again be facing the same choice. Like, do I want to go more towards professional sports, or do I want to go more towards science? And, you know, let's just make this choice slightly, slightly earlier. And that's why I decided to go to Cambridge instead. But like a big part of me, I feel, especially during the first year of my university studies, was kind of regretting this a little bit because UK is asking you to specialize much earlier. So this fact combined with the fact that I couldn't play tennis as much as I could have in the US, I think the first year in a UK university was actually mentally quite hard for me. But then it forced me to maybe make choices more than I would have in the US. Then at the end of my first year, I was thinking, do I want to do chemistry, do I want to do physics? And it was, you know, you had some freedom, but you kind of had to choose what you're specializing in. And then I heard about this possibility to study chemical engineering, which to me sounded like a combination of chemistry, physics, and math. And I thought, I'll go with this. In reality, afterwards I found out that chemical engineers learn surprisingly little about chemistry for them it's literally like A plus B reacting and producing C and D with a reaction constant of k1 and k2. But what exactly A and B are, we were paying surprisingly little attention to. But it was quite heavy on maths, I think, in terms of the physical sciences training that I got through that. That wasn't bad. And then during my undergrad, I actually really enjoyed the design projects that we did at the end of the course. My group was designing an ammonia plant And again, like, you know, a lot of math and science is coming together. And I thought I could even, you know, maybe do this for career, like go and design chemical plants. That could be quite good and interesting. And then I think that was kind of another point in my life where I needed to make a hard choice and just didn't know how to make it. And probably a few factors that were contributing was one, I think my family was actually advising that, you know, if you've already gone to university and are you're on this path, why don't you just go out and do that PhD as well? And I think what really, really helped was the fact that during one of my summer internships, I had been working with a person, Professor Thomas Knowles now. Back then he wasn't a professor, but who I just thought was a really, really great person to work with, whom I enjoyed working with, whose research I liked and who I felt was really supporting me as a person as well. So I think that this kind of two aspects together, that I actually had a good environment where to go to, like Thomas group and my parents encouraging this direction of doing a PhD were the ones that kind of shifted me towards this academic path. Otherwise, I probably would have just, you know, gone and worked in some of the major kind of, you know, chemical engineering companies, maybe even oil companies, for doing this work. But yeah, this, you know, little things came together that actually affected the outcome.

Cool. I think, yeah, I think Probably that tennis decision was really hard. I have friends who are like, growing up as kids, they were really good at golf or something, and then they got to college and they kind of gave it up, and it's like, it leaves a major hole in your life. But in these transition periods you're describing, you're alluding to your family helped you make the decision. What kind of, you know, when you're in those transition periods, like everyone is in their career, what's your framework to think about how to make a choice? Is it based on the people you work with? It's probably a combination of things too. Is it based on what you're really good at? Is it based on what you want to learn? What's your high-level framework to make a good decision in a transition period? I think a lot of scientists have that issue. Yeah.

Oh, you think scientists predominantly, not other disciplines?

I think the whole world, the whole world, but I think scientists, in particular scientists have a huge issue of like, they get the PhD, then what? I think one of the hardest transitions ever because like you're trained, you're like essentially an Olympian scientifically in a certain field, and then that you're probably not going to become a professor.

Yeah, I actually didn't find the choices at that level anymore as hard. I felt like, you know, the younger I was, the worse, you know, the less experience I had with making choices, and probably choices were also kind of, you know, it could have taken me more different paths. Like now we are kind of, you know, to some degree it's defined where you are. And sure, you know, you can kind of pivot to different areas as well. But I felt the decisions I had to make at the age of 15, 16 on some level were somehow harder because they could have taken me somewhere completely different.

I think when you're a PhD, you have sunk costs where like you spent like 6 years spending, worrying about chromatin and then you're done. Do you keep on doing that or do you get a job somewhere? And it's just like really rough actually. It's like a really rough transition because it's like someone playing, being a professional tennis player. I'm just a professional chromatin study—

I kind of don't have at this moment anymore, right? So they're kind of more confined in space, right? The choices that I have.

Yeah, and it's like a bummer when you realize you studied some really esoteric topic that maybe doesn't have a value, or maybe it's just nice to do something else. I think it's my own experience and seeing other people, it's pretty hard. And so it seems like you've made a series of good decisions. That's fascinating.

Coming back to the question of making choices, I think you can't spend all the time thinking about what you're going to do. But I think that there is value in kind of being a little bit strategic about it. So kind of, you know, a year before, if not knowing what you want to do, at least starting to actively think about this question and trying to kind of, you know, interact with and reach out to people whom you think can help you in that process is, I think, quite important. And, you know, when it actually comes to making the choice, I think the The people whom you would have on either side, you know, if you make choice A, choice B, the people who would be with you as part of the choice is kind of a very, very important factor as well in the decision-making together with how much you would like it and you would enjoy it. But oftentimes they're kind of linked, right? You know, you may enjoy something, but if you don't like the people around you, then you probably don't like what you're doing either, right? And vice versa, even if they are not— maybe if it's not your top choice. If it's your bottom choice, it's a different story. But if it's your, you know, top 3 choices, then already the environment the one that is making a huge, huge difference, right?

I agree.

I think that none of the choices I've had to make have been as hard as the one to, you know, give up on professional tennis at the age of 15 and a half, 15, 16. So every time I can point back to this saying, I was able to make that choice, I can make every other one as well.

That's so sad. It could have been— oh, that is a hard decision though. I, I like— I mean, you should How long were you playing tennis? How old were you to start playing tennis? Like 5 years old?

Or I think I started when I was like, uh, no, 6 or 7, something like 7 probably when I started. But, you know, invest a lot of time into, uh, into this. The people whom I was playing together with, like, you know, there are some of them are top 10 ranked people in the world, uh, at this moment, right? And, you know, people who are in the finals, semifinals, I've played against many, many of them during my junior, junior days. So essentially, I know you invest a lot of time time into this and give up on a lot of other things in life. So maybe that's what making, you know, you mentioned Sankost before, so maybe there was even an element of this in that.

Oh man, I'm convinced. Exactly. I'm convinced. We tell them you're a superstar caddy, so I think you're going to become a top 10 biotech person or something, the equivalent. You're going to become the equivalent of top 10 tennis, but in biotech. It's going to be so exciting. And so maybe to make the transition toward I think a really good point you made, do things you like the people around too. You want to like the people you work with, and so if you don't like them, do something else. And so with this broad skill set you have, how did you pick chemistry? Was it just a passion you've always had and you entered Cambridge, say I want to do something in chemistry and physics, and you kind of alluded to that through kind of just studying? I meet people who like learn have skill sets to do machine learning, AI type of work, and they get pulled into a bunch of different directions. And so I have a lot of respect for you to say, hey, I'm gonna focus on— I'm gonna be the best computational chemist in the world. What was the kind of the— how did you— was it just based on the people, or was it based on maybe something— maybe was it based on a problem you want to solve, or is there something— yeah.

Yeah, I mean, and interestingly, I don't know if I should necessarily define myself as a chemist, because I think like chemists are certainly one of the one of the, like, the keywords that has been there at every stage in my career. As I said, my undergrad was in chemical engineering, my PhD was in biophysical chemistry, the kind of postdoc work now with Alpha has been more on computational chemistry. So it has always been there but has never been the only, only thing. But I mean, I think probably it is defined by what you ultimately want to do, what interests and excites you. And I feel in my case this has been very much essentially helping to advance healthcare, and that is somewhere on the interface between chemistry and biology. And again, you know, with my parents being medics, they had access to essentially, you know, patients and practical medicine, but they were never that much on the, you know, technology R&D side of this. And I think That's essentially what excited me. Like, you know, I wouldn't be doing exactly the same stuff as them, but, you know, it's still in the area of things that I'm interested. And I think chemistry is probably exactly, you know, some people define it in a way that chemistry is everything that involves molecules. And I was kind of thinking about biology as well during my undergrad, but I just felt I wasn't made for it in the sense that I had so much like learning and details that I was unable to grasp. And I kind of decided to go more through the route of specializing my undergraduate education in physics, chemistry, engineering, and then learning, learning the biology aside, which I sometimes, you know, these days wish I knew more of. But I've just picked up as much as I, as I, as I could. And I think it's also hard to just be an expert at every single topic, right? So you kind of have to make some, some choices. And somehow I felt that my For my personality, the idea of going through physical sciences and kind of learning biology on top fitted me better than the alternative way of doing it.

Cool, yeah, I totally agree. Kind of picking a field based on your personal experience, your parents in the medical field and maybe having interest there, and then also, you know, you have to kind of do things you like and you shoot it from a personal perspective.

It kind of comes back to some of the points you made earlier that you kind of have the broad direction you're going towards. But then when it comes to choosing your precise path within that, then it may make sense to optimize for, you know, what path works the best for you, where you have the best people supporting you. I think you shouldn't change the, you know, final goal that you want to go towards just because of some external factors, because there needs to be an element of you as well in what you want to achieve. But like details, i.e., you know, through what course you're getting there or through, you know, what lab you're getting there, I think these already can easily come down to other aspects as well.

Cool, okay. Now we set the stage for your career. I think that's probably more important than the work. I think it's kind of where you're coming from and what drives you. So how did you choose to join Alphalyse Lab? And maybe you can talk about the research you've done so far, or you did. We can talk about the pre— I mean, I think the preprint you put out around kind of like high-throughput structural biology was really interesting. But you talk about— really, first question, how did you— Why did you go to Nathalie's lab of all groups?

Yeah, I think that's another interesting story of just, you know, stars nicely aligning for you. So after my— when it came towards the end of my PhD, I was applying for various different fellowships, both in parallel. And I wrote my research proposal, what I want to work on, and send it to various different funding bodies, seeing who's interested in funding it.. And the fellowship I got funding from was called Schmidt Science Fellows, which is by Eric Wende Schmidt Family Foundation essentially. And they very much believe in the idea of interdisciplinary science and kind of bringing different fields together. I would say in a different way from most other funding bodies who are also emphasizing the importance of interdisciplinary science, but they're often doing it by saying that, you know, people in field A and field should be writing a grant proposal together, right? And, you know, work on a problem together. They actually believe in kind of investing in individuals and giving them the chance to work at a few different fields and, you know, pivot after their PhD to different fields. So I got a fellowship from that organization, and I think the reason why I wanted to go— my PhD was kind of relatively experimental work, but then I was in lab, I was doing experiments in a variety of different questions somewhere related to kind of building high-throughput techniques for looking at proteins and protein aggregation and protein interactions, microfluidic techniques for this. And then I also— that was in Thomas Knowles' lab, who I mentioned before, is a great, great mentor and advisor for me. And then I also spent some time in Dave Weitz's lab in Boston, where I was kind of working more on high-throughput techniques at looking at single cells. And Dave's lab is one of the ones where everyone, like, you know, whatever hour of the day you're there, 2 AM or whatever, 1:00 AM, there is always someone in the lab doing experiments. And I was also there for just a short while, so I wanted to make the most of my stay. And when I was there at 1:00, 2:00 AM, you know, doing all these experiments, I was saying to myself that if I'm producing all of this high-throughput data on that particular project— was actually on cyanobacteria and getting cyanobacteria to produce electricity and build photovoltaic cells out of this, which was a very, very interesting project. But, you know, if I'm there and producing all this data, better knew how to analyze this as well. And I just, you know, felt like I have to educate myself more in the computational area. So I had to find that for myself. And I had got funding from Schmidt. So everything was in a great situation. I just started looking around the Cambridge, UK ecosystem and reached out to actually a few people because I knew I wanted to stay in the UK because of family reasons. I couldn't move too far, but across Cambridge talking to a variety of people and met with them. And they somehow just liked Alpha and his way of thinking, his way of approaching problems. And he was very welcoming towards me as well and said, yeah, sure, happily would have you in my group. And that's how we started working together when my Schmidt Fellowship started.

Cool. Okay, great. I think that's kind of like doing things that you're excited by. And then focusing on who you would like to be around.

And so I think the same story describes you. I had the broad direction that I knew I wanted to do, but then the details of how you're going there, you're choosing based on people.

Absolutely. It's okay. What was the kind of the project, or do you have a set of projects in the lab, or maybe one? And maybe you can talk about the output so far.

Yeah, I think I took a relatively broad angle in the beginning. My interest, you know, when you're pivoting to new build, you kind of want to get broad exposure to what is going on. And actually, one of the great things about— it's never easy to go to a new field— and one of the good things about the Schmidt Science Network was the fact that there were 20 of us doing that together. So we would have a WhatsApp group where we would exchange messages about the pain, about learning.

Was Peng Greensight in your cohort, any chance?

He was a year above me, actually. Yeah.

Oh, you're above me.

Okay. I interacted with her as well. She's a great person. I know her startup company is doing really, really well at the moment.

Cool. Yeah. We're an investor in Big Hat. She's awesome.

Her and Mark are awesome. I have during my career also asked for advice from her, especially when I was considering about spending more time working in an industrial environment. She was able to give me good advice. So you choose good companies to invest in, Joshua Elkington.

Yeah, I know. Mark and Peng, good taste in people. Katie, you're, you're, you're in the same league.

So, okay, so you have a broad set of projects in the Lee Lab, and it had to be molecular dynamics-based approaches for property prediction of small, small molecules. But I think my kind of heart probably was more in the direction of proteins and biology to some degree as well. So one of the key projects I ended up working, working on was exactly how to, how to make use of of, you know, experimental data, where experimental data in our case was essentially crystal structures of proteins and small molecules and the binding poses, and how to make use of especially kind of, you know, relatively small amounts of data to solve this problem. And like, look at how small molecules are binding to proteins the best, because I think, you know, there are lots of approaches that are focusing on doing that purely in silico, and then there are also approaches that are kind of, you know, focusing on specifically working with problems where you have a lot of data available. But to me, on some level, the interesting middle ground seems the one where there is just, you know, a little bit of data available to you, and how can you make use of this in the best way possible to essentially learn the most. And actually, it was a good, you know, if you're working in a computational group, it's always challenging to get access to interesting data, and, you know, you almost need to convince experimentalists that they should do the experiments you think they should be doing instead of the ones they think they should be doing. And that's really, really challenging, right? But the time when I started working with Alpha was the same one when— well, actually that happened slightly later, but when the COVID pandemic hit us and there was a— I don't know, again, if you've heard of this or some of the listeners have heard, but COVID Moonshot, they're very interesting program that brought together a lot of scientists across the Globe, whose goal was essentially to develop an open-source antiviral against SARS-CoV-2, focusing on the main protease specifically. And as part of this, there was actually the Diamond Light Source Centre in Oxford that was generating a lot of interesting structural data, small molecule binding to proteins, and Chen Jilares' group in New York were collaborating with a lot as well. So I think there was again, like, you know, a lot of interesting people, uh, with different ideas working on the same, uh, same problem. And I think that's maybe, you know, how I ended up spending the most time working on exactly this problem rather than the MD work-related specific, uh, problem where we were, you know, trying to break also very important properties like, say, solubility of small molecules and permeability of small molecules, which I think I still need to write the paper, paper up on. But, uh, I got some of the work done, but I think it was literally again, you know, the community effect and lots of other exciting people being around there thinking about these questions, how I ended up working mostly on this. And then in parallel, actually, interestingly, Thomas Knowles, who was my, as I mentioned before, PhD advisor, got in touch with me and said that, you know, some of the computational approaches we've been developing, let's try to use these more for proteins and specifically for liquid-liquid phase separation as well. So another kind of big part of my fellowship work together with Alf and Thomas was on exactly building a kind of predictive models for understanding how proteins are forming condensates. So I think, yeah, these were essentially the few key contributions from my Schmidt Fellowship work, using structural data to predict small molecule binding to proteins and then predicting protein phase separation related behavior.

Cool. Let's talk about the first part, using kind of high-throughput crystallography to predict new molecules. That's kind of the In my opinion, it's beautiful work you did. And so, uh, you just— we'll put the link to the preprint where, um, you essentially use— I forgot how many structures. I think you'd use like 100 protein ligand structures for a COVID protease, and then you screen like 10,000 compounds. And maybe you can talk about that research, but from my understanding, the key breakthrough— and I love to learn more about how you how you figure out how to rank these molecules differently? Because from me reading the paper a few times, the key, the key, the key trick for you to actually discover useful chemical compounds from a virtual screen based on, you know, 100 protein ligand structures is you, you compare the compounds pairwise first and then did a bioactivity ranking. Maybe talk about the research, but how did you— what was it kind of— that seems to be the big breakthrough or the big kind of the intellectual leap you had to make or intellectual creativity? How did you get there? I'm always fascinated with that.

I would definitely want to credit Alpha here a lot. I think he was the person who first suggested that let's do it pairwise. But conceptually, I feel there is a lot of beauty to this idea. If you're thinking about machine learning models, people are often building classifier models, but then all that is telling you is, essentially if or what class something belongs to. And if you want to find a compound that would be more potent than another compound, it's almost like, you know, if they're in the same class, it's very hard to say if or not it will fulfill that constraint, right? And then that's why there is attraction towards building regressor models where you're actually predicting the precise value. But with such models, it's just so hard not to end up learning noise because, you know, any measurement system, you know, and that's the experimentalist To me, my speaking has some level of noise to this, and often it's really hard to even determine what the level of noise is. But we kind of had a look at this experimental data and we're asking ourselves that where do we feel noise essentially begins or noise finishes and signal begins and kind of define some threshold. And then we were asking the questions that are two compounds more different from each other than that threshold? And that kind of feels almost like a marriage between the two approaches, that is a regressor-based approach and the classifier-based approach. You're still building a classifier model and you're not learning the noise, but you're actually getting rankings. And yeah, and I think, you know, the other beautiful aspect of this was like when we used the available data from Diamond Light Source and we're using this essentially to train the model before starting to do it in silico. Screen. But then you have more than one compound that you know is potent. So when you're doing things in silico screen, you can actually rank these 10,000 compounds relative to all of the top ones. And you can only focus on these which come out as the top ones according to, or like, you know, when compared to every, every single hit. So I feel that's kind of another really good way of reducing, because often the outcome of in silico screens they have so many things that are potentially interesting, but that is really helping you to reduce that down and in a somewhat orthogonal way in the sense that oftentimes the hits are coming from different scaffolds, right? So you're not going to buy just to watch one scaffold if your key leading compounds are from different series.

So I think this research in the context of like docking and just virtual screening elsewhere, like most virtual screens of small molecules for like a given protein target, generate a lot of low-quality hits that don't match experimental data. And then you like, you discover some molecule, think it's really potent, and then you actually put it in an animal and you spend all this money, you're like, oh, it's useless. And so you have to go again, you keep on, you put garbage in, garbage out. And so your research is kind of a better kind of virtual screening method that actually can match experimental data. And so when you think about this research, this ability to combine high-throughput crystallography with machine learning to then produce better virtual discovered small molecules. What's the bottleneck there? Is it the modeling? Is it the actual data generation? And what do you see the next steps for this type of research?

Yeah, great question. And I mean, I'm very much looking in the direction of the key companies in this area at the moment. Like, you know, Isomorphic, Charm, all the other key companies that have been formed in this context to see how they will push the boundaries because probably whatever I will say is the boundary they've already solved internally, right? But I mean, I personally, and again, it's probably because I'm trained as an experimentalist during my PhD, but I almost feel that there is limited value in not using experimental data to seed the models. Of view if you can, because, you know, sure, conceptually it's really beautiful that we can do things in silica. If you're thinking about the, you know, full drug discovery pipeline, then, you know, it's long and, you know, sometimes even synthesizing the protein, sure, it requires some resource and it can require a few— these days oftentimes, you know, a few months' worth of work. But if the full pipeline is like, you know, 10 years, long, it's actually not that big a problem to spend some time on synthesising this, right? And actually getting as much information out of real data as possible. So I think that, but you know, you don't want to spend infinite resource on this. So exactly, screening a small number of molecules against this and making sure that you're integrating as effectively these learnings as possible. And, you know, probably screening for a few different types of proteins, but then you also have cases where proteins are not even folded and you don't have the same binding pockets available. And that's actually one of the current questions I'm focusing on a lot when I'm working as a staff transition, that, you know, what do we do if we have not structured proteins and how do we still do anything useful in that context, right? But coming back to your previous questions, I mean, I think there is a lot that can be done on the computational side, but I mean, I personally think that just bringing in some experimental mental capabilities, even if that's not your primary, primary focus, as soon as possible. And if that's accessible to you, why, why not make, make use of this?

When I read your preprint, I'm like, wow, this is such a great intellectual breakthrough for me, where the value of a diverse set of protein-laden structures and where that can actually help build better models. I think, I think there's a lot of work left to be done there. And so maybe we can make that—

this, this work kind of forward in the, in the sense that already where, for instance, in this, in the context the same data, we're looking into how to generate even less data or screen even fewer compounds. So, you know, which, given the 200 structures we had available for us, you know, how would we come up with a protocol that if we were only to screen 50, how would we still get good performance? Because like, I don't know how detailed they looked at the paper, but we're essentially comparing what happens as a result of the number of compounds in the training data, kind of below 100, we don't really get a good performance.. But, but, you know, that really illustrates to you that it's probably not the absolute number, but the way how these compounds are chosen. So that's kind of been one of the ongoing directions. And then actually also, interestingly, the people who were kind of part of the COVID Moonshot Consortium recently got a bigger grant from NIH to kind of, you know, use the same framework in the context of other viral diseases as well. And I think we are going to have similar data against different systems as well so that we can really validate if the learnings are kind of specific to just this case or are they more, more generalizable. So I think that's something I'm actually quite excited about, getting more data for that initiative.

Cool. And then let's get to the penultimate part. It's a fancy word. Okay, so your transition, but maybe talk about your time in research. And right now you're the head of machine learning or you're the head of something at TransitionBio. You know, how did you, you know, you bring some of the work you've done maybe in the virtual screening part, and how you're kind of using it in condensates. How did you get interested in condensates? You know, what are you doing at Transition right now? What are you excited about? What are the new, what are the problems you're trying to wrestle with right now to bring, let's say, machine learning and chemistry to condensate biology, which is really exciting and new?

Very, very, very good and valid, uh, valid questions. I think the, the reason why I'm, why I'm excited about the condensates is just because how, you know, very, very simple concept of, uh, or, or a simple idea of condensates are actually playing such a broad role across biology. Like, you know, people are seeing condensates. And we can have a separate discussion if we're not, you know, how liquid a condensate needs to be, or can it have some, you know, gel elements to this as well, or, you know, when does an aggregate start and when it's still a liquid-liquid condensate and when it's a gel. But, you know, in principle, this concept is kind of relevant across so many different domains, from stress responses to gene regulation to signaling and many, many other functionalities. And I think it's also kind of, you know, taking— again, if you're looking at drug discovery and development, oftentimes things fail because of biology, not because of us being unable to find a small molecule that would be very good, that would bind very well, right? But because it's just, you know, not acting on target or the target— modifying the target not having the effect we think that it does. But that's like condensates are kind of on some level taking a step further. Your target is no longer one protein, but it's kind of a system, a complex system that is playing a broader role. And, you know, even your screening can't happen against just a single component, but it kind of has to happen on a broader scale. So I think the reason why I got personally into this idea, and I know it's a very, very controversial field, although I think that the controversiality is a lot associated with the terminology and, you know, how you're trying to resell an old concept and so on. And I'm not, you know, arguing that in any way. I think the interesting question is, can we use this to treat diseases? And I feel on some level it's kind of a step towards the right direction because you're actually looking at a system as your target rather than just a single protein. But anyway, that's why I like condensates. And I think I The reason why I'm very involved with the transition now is because to do this work well, you need a lot of resource and you need a lot of data. And we were talking about COVID Moonshot before as a good way to, you know, a good initiative that was creating a lot of data. And I was mentioning some of my earlier projects that I wasn't as excited about in a computational chemistry context, which I think was partly because I was like, 'Yes, I've built this great model, and what now? Like, what next?' I kind of, you know, wanted to see the goal as well. And that's why, you know, you're just working in your own corner building a model. I felt that that wasn't for me, but I kind of, you know, wanted to be— wanted this to help the bigger ecosystem. And yeah, I think that's what I felt with the Condensate work as well. There is just limited amount of data that is available. Creating data is often expensive, especially at the scale where it can actually learn some something as complex as biology. And through activities at Transition, I think that's the best way how we can make progress in that direction.

Cool. Yeah, I'm really excited to see what you publish and do. And I think Transition is very lucky to have you. So very lucky to have you. And so I think condensates are very interesting biology. And I think, like you said, if you generate enough data, you begin to figure out which diseases they're influencing, because they are. And then also trying to figure out where they actually exist and at what time. And kind of condensates at a high level is where biology meets physics. I think that's one reason why you're really interested in it, because you kind of have to actually generate all this data. And there's actually physical components to these communities. And kind of, yeah, that's a whole other hour conversation on the history of condensates, because it's a fascinating— I come from a C. elegans background.

Initially.

Ah, yeah. So condensates, kind of these granulars, they were discovered in these worms. And so I'm kind of, I've been, I come from that angle where I kind of know the history of the field and only kind of now do we have more powerful enough tools to image these condensates and then probe them. And so I think it's the next decade of the field that's going to be really, really new.

I feel it's a good excuse on some level to learn biology, right? You know, you're focusing on learning about the broader biology around it. Everything as well, because you just can't treat it in isolation. It's just so associated with everything. Like, you know, it's sometimes really hard to say condensates are there in the beginning or if they're formed as a result of, you know, some other pathway interacting with that, with others. So yeah, I sometimes say it's an excuse to learn biology.

Oh, interesting. Okay, that's a good point. Okay, that's pretty good. Okay, yeah, condensates are a way for you to learn biology. I like it. Okay, maybe to wrap it up, to the end, you've had this incredible career. From a freaking awesome arithmetic machine to tennis player, to then really freaking incredible world-class chemist, and now you're doing condensates. What do you see the next 5, 10 years of your career? Do you want to, what are your big goals? Do you want to be part of a drug approval? Do you want to start a company? Do you want to, what kind of, let's see the big things you're pursuing. Which direction do you want to go into? And what excites you too?

Yeah, yeah, I mean, I think my near-term future will be quite focused exactly on biotech and drug development and kind of, you know, curing, contributing towards human health and our understanding of human health and disease. And I, you know, I'm actually very happy that I'm involved with the transition at the moment because I'm actually closer to having an impact than I would ever, ever do if I was purely just in an academic context. So I feel that's exactly the right thing for me to focus on at the moment, as well as, you know, all the activities that are hopefully building on the COVID Moonshot work that we were doing before. And as I mentioned, looking in the context of other viral diseases, what's going on. So I think these are the near-term questions that are interesting me, and I want to do it exactly together with the people who are similarly excited about these questions as I am. And I just feel really, really privileged to have such people around me. But then, you know, when I'm looking outside at the heat wave that is here, and I don't remember being as crazy when I was still growing up, and looking at the 3 young kids I have at home and what planet I'm leaving behind for them, I feel like, you know, right now I'm kind of relatively focused on what's going on and I don't want to leave it, leave it. But if I, for some reason, things weren't going as well as they were, and I, or, you know, at some stage in future I want to move on, then I'm quite, quite certain I want the focus to be on probably still with the biotech hat on to some degree, but on maybe even bigger, bigger problems like climate and our environment and even agriculture too, to some degree. I don't want to kind of pin down one, one specific problem. But yeah, you know, maybe I didn't know if I listen to this in 10 years from now.

Cool. I didn't know that. I learned something new today. But yeah, I think Transition is doing really incredible work. And so, you know, it's a really great place to be.

And then long term, right? So, and who knows, maybe there will be just too much interesting things to solve in there where I am at the moment. I will never, never make this move. But, you know, just obviously looking at what's happening around, around us, you know, I think One thing is that I'm working on myself, but in terms of encouraging other people, I think that that's an area where a lot could be done and should be done. And if we can get more human resource, best resource on the planet to think about these questions, then that's the best outcome of this conversation with you at the moment. If any talented person listening to this is getting more and more excited about looking at these questions, since then.

Cool. OK, that's exciting. I think a lot of people are going to find this useful, not only right now, but over the course of the next decade. And so we'll do a version 2 in a few years. And then we'll do updates. We'll do updates, Katie. And then we'll do a whole little 40-decade— we'll make a whole documentary one day. 4 decades from now, we'll have a whole documentary, a little canvas of conversation. And see your career. And I know for a fact you're going to do incredible work. So like, I know, that's why I wanted to talk to you right now.

You should have a thumbnail and focus on those who are either the most consistent in terms of what they're saying or the—

I only like talking to superstars. It's just how I spend my time. And so thanks for taking the time to talk to me right now. And, you know, I really appreciate it.

Thanks again for inviting me, and even more importantly, for doing this. I think that's really, really great that you're actually thinking about the younger generation who's finding it really, really hard to make choices between things. And I know that's, you know, in your mind, one of the primary audiences for this. So I think it's really, really valuable.

This is going to serve as a resource for up-and-coming scientists, engineers to make great decisions, and you know, I think your career is kind of an example, one to kind of be inspired by, at least in my opinion. And so, and I think there's, you know, all you need to be honest, as long as this reaches one person, that's cool. It's good enough for me. As long as like one really talented person gets something out of it, that's like cool. And they, as long as they email me and I can invest in their company, boom, totally worth it. And I, you know, and have a lot of, totally worth it for me. And if other people get something out of it, that's cool too. But Cady, I really appreciate it. I know it's really hot, so try to get hydrated, but I really appreciate your time here.

Yeah, no, thanks. Thanks a lot for your time as well. And I'll also say the fact that make sure that you do a good 100 of such podcasts because I always feel that it's very dangerous to follow one person's story. But if there are 100 you hear, then you just pick out an element from each of them that works for you because I think every person's different. Totally agree. Keep, keep up the hard work, Joshua. Appreciate it.

Thanks.