Good morning, and thanks for joining us to have a conversation with Zach Jonasson, CFO and CBO, and Alex Khan, VP of Finance of Absci Corporation. Absci is a public biotech company which is focused on developing novel protein therapeutics by harnessing the power of generative AI. Empowered by its proprietary Integrated Drug Creation platform, Absci has not only established collaborations with leading pharma and institutes, but also has started developing its own pipeline focused on cytokine biology. So the company's lead candidate, ABS-101, recently entered IND-enabling studies, and to discuss the company's development strategy in 2024 and beyond, I welcome Zach and Alex to this Fireside Chat. Zach and Alex, thank you for joining us and accepting our invitation to talk to our audience today. So for starters, Zach, could you kind of highlight, you know, at a high level, what is generative AI, AI, and how is it being used as a tool to help in drug creation? Absolutely, and first, thanks for having us here. We're more than happy to participate. When we think about generative AI at Absci, we think about it in terms of what it delivers, so generating differentiated assets against meaningful drug targets. And our approach is to integrate AI into everything we do, and so the roots of Absci are as a synthetic biology company, and why that's important is because the foundation of the company was around building a platform that generates data at high scale. And we use that data to train the AI models, and then we also use those assays in the lab at scale to do validation of the models. At the end of the day, we're using a generative AI approach because it's very complicated to design against a target. The complexity of the problem is such that it's not something that a human can rationally design with a pen and paper, even with a traditional computational approach. So we're really looking for AI to understand the fundamental language of protein design and protein-protein interaction. And that's where, again, our data capability comes into play because the proprietary assay systems we've generated in-house generate that functional data. So how do antibody CDRs, or how do antibody sequences interact with a given target antigen, in particular, the antigen's epitope? And measuring that in a high-throughput way at scale is what's allowing these models to learn that fundamental understanding of how antibodies interact with a given antigen to deliver the biology you're after. Very good. So in terms of the platform itself, the Integrated Drug Creation platform, can you highlight to us some of the capabilities that platform is delivering for you? And how do you plan to, you know, develop this further from here? Yes, and I'll highlight three key areas. The first is our models are all epitope-specific, so with traditional approaches for, what I would say, discovering antibodies, they are truly discovery. So you're using a mouse, an immunized mouse or a phage display. You're hoping it gives you an antibody of high quality and high affinity to a given target, but you have no control over what epitope on that target, or even if you get something out that actually has good developability. So in our case, our models start with the epitope of interest, so we can specify that epitope. So we're moving from discovery to more of an engineering discipline. And beyond just specifying the epitope, we then use the model to computationally look at every possible interaction with the epitope, and what we're finding is the actual antibody-epitope interaction, so the paratope, is very important for potency and, in some cases, MOA. So our models are now allowing us to test that because we can specify the epitope, and we can allow the model to look at all the interfaces. The second component is using the model, and this is the AI optimization model primarily, whereas what I was talking about before was the de novo model, where we generate the first antibody leads that target a given epitope. But in the AI optimization models, we're able to design in unique functionalities, so we've gone beyond just developability. So is this molecule designed to be stable, manufacturable, formulatable, all those elements that are key for success? But we've moved beyond that now to look at other features that could be meaningful clinically, such as pH-dependent binding. We've used our AI also to design Fc engineering that can deliver longer half-lives than YTE or LS mutations. We have yet to incorporate that into one of our programs, but it's just showing you where AI is able to be deployed in our, in our hands. And then the third aspect is efficiency. So with our lead program, ABS-101, I think we've talked about this before, we were able to just to get to a drug candidate, from target selection to drug candidate, in 14 months on a roughly $5 million spend. Over the last 12 months, we've carved about 15% of that timeline off. We could do that same program in twelve months today, and so we're continuing looking to leverage these efficiency gains as we move the platform forward. So now there are multiple AI-based drug discovery and drug development companies out there, but you have something called zero-shot generative AI model. How does that differentiate against what's out there? And then what sort of a validation do you have for the zero-shot generative AI? Yeah, absolutely. So to our knowledge, no one else can do this for antibodies, at least no one who's published or shown an actual molecule that's been designed this way. Several years ago, I'm trying to remember, the January of 2023, I believe, we put out our first manuscript showing the zero-shot de novo design of a novel antibody to the HER2 antigen. So we started with that work that early, and we've continued to move our de novo model forward since then. So most recently, as an example, in third-party validation, last week in an 8-K, we announced a technical achievement of a technical milestone in our partnership with AstraZeneca. And in that partnership, they gave us a very difficult transmembrane protein to work on, our target, and in under six months, we were able to use our de novo model to design antibodies with very good binding affinity to that target. So we achieved that milestone, and that program now is moving into lead optimization. So again, that's another proof point where we've done that with a partner, but we're continuing to use that de novo foundation model for our own programs internally as well. Then, the other unique factor with Absci is that you have a proprietary E. coli that you use, you know, to test some of the molecules that you've designed. So what's unique about this E. coli itself? And, you know, how does it help you to move your molecules from the computer to the lab? Yeah. So I come back to... I said we had our roots in synthetic biology, and that's really the E. coli platform that was developed. So in the early days of the company, they invented a way to engineer E. coli so that it could produce full-length antibodies and could do that very rapidly. And that's a number of innovations, but you can think of the E. coli that's been engineered here as being almost mammalian-like in its ability to produce antibodies, but still have the growth that you see with E. coli. And it also has an oxidative or semi-oxidative cytoplasm, so you get the disulfide bond formation, correct folding of proteins and antibodies in the cytoplasm. So that was a fundamental innovation that we then took and developed high-throughput screening methods around. So today, we can take a kind of a pooled approach, where we take a test tube of this engineered E. coli, and we can transform those, like, E. coli with a three million member antibody library, sequence library. So now we have a pooled approach where we have the E. coli in this test tube. We could have a billion E. coli cells in this test tube, each producing a different antibody. And then we can introduce an antigen in a flow-based experiment, where we can look at the binding of those antibodies for each E. coli is producing to the target antigen. And we can do that in a flow experiment that allows us to do it in high throughput and at scale. So what does this all mean? It means that we're able to generate high-throughput, scalable, functional data to train our AI models, and then we can take that same platform and do validation work. So once the models, the AI models, give us designs for antibodies, we can then synthesize those in the same platform and run them through that experiment to confirm the binding affinity and other properties that we're engineering for. So based on all this, you actually had quite a bit of success the last twelve months in terms of collaborations with companies such as Almirall, PrecisionLife, and AstraZeneca. And especially with AstraZeneca, last week, you announced achieving a major milestone. So what was that milestone, and when could we hear any update, you know, regarding additional milestones from with AstraZeneca? Yeah. So I talked about this previously. So we achieved a key technical milestone, and I think it's a really meaningful one because this was an opportunity for us to demonstrate our de novo zero-shot capability. Yeah With a partner, and we did it in a very rapid amount of time, six, six months, and it was against a very difficult target. One of the things we're always striving to do at Absci is improve efficiency, but also improve the generalizability of our foundation models. Here was a case where we were able to show that the model generalized to this difficult transmembrane protein, and we did it in a very rapid amount of time. I think that was a key technical milestone in that partnership. I should preface this and say, one of the reasons AstraZeneca, I think the key reason AstraZeneca selected Absci to work with, is they were looking for a company that actually could do de novo design using AI or just de novo design, period. We were one of the, I think, the only companies they found with true validation that we could do that, and we delivered in this partnership so far. In terms of next steps, that program will now go into lead optimization. It's hard for me to give you an exact timeline when we make a new announcement, but the next critical inflection point will be- Yeah ... achieving the lead optimization and having a drug candidate. So just like what you spoke about AstraZeneca, you know, and their interest in having a company that can do, you know, not only initial stages, but also help you or help them out in developing better drugs, what's the nature of... Is that a similar, is that the sort of a collaboration that these large cap pharmas want to get into, or are there, you know, what other things are large cap pharmas looking for, especially when they are looking at AI companies, you know, to help them in their R&D? Yeah, I think, there's a couple things they're looking for. I'd say top of the list is an ability to address difficult targets, and that's what we're doing with AZ, right? We can certainly address less difficult targets, too, and we can do that in a more rapid time frame than you could with traditional means. But that addressing these targets where they're difficult, you can't really get at them with traditional means, but where the biology is known, and it's understood to be an important target, that is a, I think, a really big value proposition where we can collaborate. I think there's, From some other companies, there's interest in our ability to design fast follower. Mm-hmm. Molecules that have significant differentiation, which translates back to picking the right epitope, engineering in higher potency and half-life and other properties that could be meaningful clinically. So it's not just large cap pharma, but you're also started collaborating with, you know, large medical institutions like Memorial Sloan Kettering, where you recently announced, like in mid-August, you announced a partnership there. So what's that. You know, what are you looking for in a partnership such as these compare when you compare them against large cap pharma? And also, can you remind us how many partnerships that or partnership programs that you want to develop in 2024? Yeah. I'll let Alex talk a little bit about our guidance, but I would say we're, as a company, very excited to be working with Memorial Sloan Kettering, and it's a global leader in cancer research. I think the one word I would say to summarize why is synergy, right? They bring a tremendous amount of expertise in cancer biology, so think about the target biology. We bring the expertise in designing an asset that's going to be highly effective, developable, and get at the biology that they're trying to understand. Then secondly, MSK is fully capable of doing phase I clinical work, so the synergy is on both ends, really. And the other point I would make is we're fully aligned on what we're trying to do here, which is to design an asset against a novel target and advance it through phase I, phase I proof of concept, and then work together to find a good, a good transaction, so a good pharma partner who can take it the rest of the clinical journey. And so I think we're well aligned, and the synergies are palpable. We're already getting started on that partnership, looking at novel targets with the MSK teams. So I... Yeah, we see a lot of potential here. Yeah, I mean, to Zach's point, finding these sorts of partners with whom we could have these really fruitful collaborations with is very important to us and to them, and we try to look for areas where we can have these synergies between, for example, our technical capabilities and a partner's domain expertise, so we can help each other fill the gaps and really have a, you know, a very meaningful collaboration. And so we saw last year, toward the end of the year, we signed a number of partnerships with, some pharma companies like AstraZeneca, like Almirall, and these co-developments with PrecisionLife. And then for this year, we've so far announced the up to six program co-development with Memorial Sloan Kettering. We do expect to have a few more additional partnerships this year. The guidance is for new partnerships with at least four partners, so looking to have a few more announced by the end of the year as well. Perfect, so it's not just external partnerships. Now, you have your own pipeline, you know, especially looking at cytokine biology. Of that, you have named three of them, ABS-101, 201, and 301, you know, especially with 101, you know, you're planning to enter, you know, you're currently doing IND-enabling studies, so what's the status of them, and when would... You know, in terms of clinical development, how far would you take it, and when do you plan to get into clinic? Yeah. I'll talk about 101, and Alex can talk about 201 and 301. With 101, we plan to wrap up the IND-enabling studies over the course of this year, and we're on track to initiate our phase I clinical work early next year, with an interim data readout from phase I planned for the second half of next year. We remain on track there, and as you probably noticed, a couple of weeks ago, we released some NHP data. Mm-hmm. Which demonstrated a two to three X extended half-life versus our clinical competitors. It also showed that we had very good biodistribution. We think that that may translate to an advantage where we wouldn't be required to have a loading dose clinically. And we also released press release that we're able to formulate this up to 200 mg/ mL, so we're well on track for a subQ formulation. Yeah, and beyond ABS-101, we also have in the pipeline ABS-201, which is a potential best-in-class dermatology program, and ABS-301, which is a potential first-in-class novel IO target that we discovered using our Reverse Immunology platform, which we could you know discuss a little bit as well. What's nice about the portfolio is it really does show the sort of breadth of areas that we can go after, including to Zach's point ABS-101 in IBD, and for you know the other programs in dermatology and immuno-oncology. So for ABS-201, we look to have development candidates selected toward the end of this year, and for ABS-301, we look to have the completion of our mechanism of action validation studies done by the end of this year as well. We expect to have some data to share on either one or both of those at our R&D Day on December twelfth, and if not both by then, possibly one then and one early next year. Also looking to announce potentially then one or more new programs to enter the pipeline as well, as we continue to make progress on those. Very good. So, in terms of catalysts or say, let's say, over the next 12-18 months, you know, what should investors be looking out for? We have quite a few coming up. So I would say, obviously, initiation of our phase I clinical work for ABS-101, and then the interim readout in the second half of 2025. And then to Alex's points, announcing the DC package and sharing that publicly at our R&D Day for ABS-201, and then the in vivo validation on ABS-301, and we'll have a DC not long after that as well. And then beyond that, we'll be announcing additional partnerships and progress points in those partnerships. So, you know, at least from the beginning of this year, we have been seeing, you know, huge leaps in terms of technology and, yeah, you know, technology developments within the AI space. So, as one of, you know, the pioneers in this space, what do you think is the growth from here, you know, let's say over the next three to five years, in terms of AI being a useful tool in drug development? And, you know, what should we be looking out for? You know, I think that right now there's a lot of noise in the space, and some of that is just because a lot of groups do their validation in silico. You really have to do the validation in the wet lab, and that's what we're set up to do. We have a 77,000 sq ft facility with proprietary assays to do that work. So I think as we advance the field, what we're going to need to see from other competitors, too, is just more validation of what the models are producing. And then the next step, which we're already on this journey, is producing assets using those AI capabilities that are differentiated, and that's what we've done with ABS-101. It's what we're doing with 201 and 301. It's what we just did with AZ in the partnership, right? So it's these important proof points on how you're leveraging the AI platform, which in and of itself is a tool to develop these differentiated assets. Thank you. Thank you very much, and, thanks for joining us. Good luck. Yep. Thanks, Vivek. Thank you. Thank you.