Everybody to our LSMT, our healthcare virtual conference here at KeyBanc. My name is Scott Schoenhaus. I'm the healthcare technology analyst. Pleased to announce or pleased to have here Absci CEO founder Sean McClain, as well as Zach Jonasson, CFO and CBO. Gentlemen, thank you both for attending our healthcare conference. Maybe to kick off, Zach and Sean, give us a little bit about your background and the background of Absci for new investors that are new to your story in this fireside chat.
Yeah. Thank you, Scott. I can kick off and hand over to, to Sean. So I'm Zach Jonasson. I'm the CFO and the CBO. Quick bio, I have a PhD in cognitive neuroscience from Harvard, degree in statistics. I've been a founder several times over, the CEO or CBO, but the last 12 years before joining Absci, I founded and built a venture capital firm, where I focused on investing in the life sciences segment, and in particular, in the intersection of AI and biology, AI and drug discovery, and so spent quite a bit of time researching that space. And, I was actually the lead investor in Absci's Series A, and I've been on the board since 2016, and I would say I'm absolutely thrilled to join the company last year, full-time in operations.
Absci is, in my view, having done a lot of work in this space, very well differentiated from other players in the AI space. They have their own data generation capability that drives the AI, which is truly unique, and have a tremendous AI team as well as management team, so really thrilled to be part of the team as of coming on board last year. I'll hand it over to you, Sean.
Yeah. Thanks, Zach. Yeah, as Zach mentioned, we're a data-first generative AI drug creation company. We actually originally were not an AI company. We originally had developed technology that could scale biological data, and in our case, the data we were able to scale was protein-protein interactions, essentially how antibodies interact with a target of interest. And we all know that, you know, antibodies work by, you know, binding to a particular region of the target or epitope of interest, as well as, like, the overall affinity that essentially dictates the efficacy, or the biology of an antibody. And, you know, why is data important, you know, in biology and when it comes to AI?
Well, if you look at, you know, these large language models that are out there, these chatbots, they're trained on the whole internet. But if you look at the publicly available data for biological data, it's a small fraction of that, and you need these scalable wet lab technologies now to ultimately be able to train these models to have, you know, similar outputs or usability or applications that you do with these large language models in biology.
I firmly believe that those that are gonna win in this space are gonna be those that focus on this data generation, whether it's curating it themselves, buying new data, having hospital partnerships, but ultimately having that data to use as training as well as for validation is extremely important. Essentially, we started combining this, you know, in 2017, 2018, when these transformers started coming out.
It's like this idea of, like, well, if you could take all this data we're generating, combine it with generative AI, you could go from this paradigm of searching for a needle in the haystack, this kind of trial-and-error process, to actually creating the needle, in our case, a biologic, and starting to apply engineering principles to biology, being able to, for the first time ever, hit an epitope of interest. No one has been able to, you know, design an antibody that hit the epitope of interest, and this is what's actually unlocking these new partnerships that we have.
It's allowed us to, you know, ink a deal with Merck, with AstraZeneca, with Almirall, because now we're able to start going after these kind of undruggable targets like GPCRs and ion channels and unlock new novel biology, but at the same time, being able to shorten the time it takes to get into the clinic. We showed with our TL1A asset, our lead program, that we're able to get into the clinic within 24 months. Normally, that takes 4-6 years. Additionally, we're gonna be able to, you know, dramatically lower how much it costs us to get into the clinic. It'll be $13 million-$15 million to get into the clinic. Normally, that's $50 million-$100 million.
And so you can now start to, instead of investing, you know, $50 million-$100 million into one asset, you can invest that into five to 10 assets and actually have higher probability of success with, with those, again, engineering in the biology you want the first go-around. And so this is a really new, exciting, era that, that we're entering into, and I, I truly believe that generative AI is gonna have a, a huge impact within this industry. But ultimately, we need to figure out ways to continue to generate this, this scalable wet lab, data to train these models in order to continue to have the, the, the success that, that, that we're having. And so, that's kind of a little high-level overview of Absci and kind of the, you know, where this intersection sits of AI and biology. Scott, you're, I think you're on mute.
Sorry about that. Thank you, I was. So, very helpful. Maybe taking a step back further, Sean, maybe talk about how you are situated in the market. You know, I cover several AI or tech-enabled drug discovery players. You're focused on large molecules, biologics. You really there's not as much a robust public data set on the large molecule side as the small molecular side, which inherently makes it more complex, but also differentiates you in the market. So maybe just for new investors, kind of, where you are in this marketplace of AI to drug discovery. Obviously, I view you as the more data-forward, AI-forward player in this space. Kind of just talk about the marketplace, 'cause I think there's confusion that you're all competing against each other, when that's not really the case.
Yeah, that's not the case at all, and I mean, you hit the nail on the head. I mean, we're focused on large molecules and antibodies versus small molecules. I would say a lot of the first-generation AI drug discovery companies that emerged and those that are, you know, publicly, you know, traded, are really focused on small molecules. And there's really a reason for that because you can go take a, you know, a 100- or a million member small molecule library off the shelf, you know, whether it's a CRO, large pharma, biotech, and go and screen that, and use that as training and data.
But when it comes to antibodies, you know, chemists don't, you know, make antibodies, you know, actually, living organisms make antibodies. And so for every antibody you want to test and screen, you have to figure out how to produce it in a mammalian cell or a CHO cell. That's traditionally how antibodies are made, and traditionally, that just wasn't scalable. You know, you can maybe produce, you know, thousands of different unique antibodies in a given week that you wanted to test and screen, but that ultimately just isn't enough data to train these generative AI models. And so we actually engineered a very simple organism, a prokaryotic organism, E. coli, to produce antibodies for the first time ever.
And what this allowed us to do was actually solve the production or scalability problem of these antibodies, where instead of producing thousands in a week, we could actually produce, you know, tens of millions, hundreds of millions of antibodies. And we do this because we can take a test tube of our engineered E. coli, we can take a 100 million member antibody library, throw that in the test tube, and, you know, every single E. coli in that population is then making a new or different antibody. And so you now have 100 million unique antibodies in that test tube that you can then go and test and screen, and that's where we developed our ACE screening assay, where we could interrogate, you know, every E. coli producing a different antibody, and look at how it's interacting with a target of interest.
You know, what epitope is it binding to? What, what's the affinity? And that's the data then we use to train our AI models. And one thing that's also important is, not only were we using this technology for the training, we're also using it for the validation as well. So when you know, when we test our models to see how accurate they are, you know, we can test up to 300, or we can test up to three million unique AI-generated antibody designs in a given week. And that whole kind of cycle is about a six week time period.
And so we're able to rapidly iterate on the model designs and architectures, and I think that's really what has been differentiated, or for us is that scalable wet lab technology to produce and screen these antibodies of interest. And you know, and again, I think that's why originally there was a big focus on the small molecules, 'cause the access to data was a lot easier to achieve than in biologics.
That's helpful. So I want to talk about your zero-shot generative AI technology. Maybe talk about what that means. You know, it's, it, it seems to me very much of a breakthrough technology. Maybe talk about how this was developed, and, and the applications, and what you're seeing so far with that technology.
Mike, you want to take that?
Sure, yeah. I mean, to me, this is one of the, you know, fundamental, like, breakthroughs that we've had at Absci. So the de novo models, essentially, you're able to access this massive sequence space. So I think Sean likes to put it this way: typical campaigns where you're trying to generate an antibody with a mouse immunization protocol, for example, you just sort of get what the mouse produces, and you're kind of crossing your fingers that it binds to the right epitope, that it has the right profile for developability. Here, computationally, we access this massive sequence space, so we can really look at true diversity beyond what you would see in any kind of animal model, and design in properties that we want. So the de novo piece is really important.
It's where we essentially feed the model a target epitope, and the model can access this massive sequence space on the order of 20 to the 55, and generate idealized designs for binding affinity to the given epitope, and then we can go verify that in the lab. And so the de novo piece gives us the starting point for the best potential binding antibodies, the best potential CDR sequences to bind to a given epitope, and then we take that through a second AI model, which then further optimizes that for all the properties we want to see for developability, for example.
Great. Sorry, I was just trying to unmute myself there. I kind of want to move into your internal pipeline and your partner programs. Where do you want to start first, Sean or Zach? Do you want to start on your pipeline, or where do you want-
Yeah.
Okay, let's talk. Let's start on your internal pipeline, your lead asset, your TL1A molecule. You've just entered it into IND preclinical studies. My understanding is there's other assets, clinical stage assets being deployed by Merck, Roche, Sanofi for the in this area. So let's talk about how your antibody, your molecule compares to them from your early readouts.
Yeah, absolutely. This molecule has the opportunity to be a potential best-in-class asset and have differentiation compared to the competitive molecules that are out there. The first is on the dosing side. We do believe that we'll be able to go from once monthly to once quarterly dosing by extending the half-life. We did show this in the mouse PK studies that we discussed earlier, or we released earlier this year. We'll have NHP data readout on that this year, which I think will be a key catalyst that is coming up. And then the other aspect of differentiation is on the potency.
We were able to show that we, you know, could have, you know, superior potency when compared to the competitor molecules. And then, you know, the last piece is actually differentiation on the potential immunogenicity profile. If you look at the Roivant and Merck data that's out there for immunogenicity, the ADA response for the Roivant molecule was substantially higher than what you saw with the Merck molecule.
And when we looked at kind of T-cell activation assays for our molecule versus all the competitor molecules, they were, you know, roughly the same, which then led us to believe that it was complex driven, where the antibody was binding to was really driving the immunogenicity. It was B-cell mediated versus T-cell mediated. And so what we did was we engineered the molecule to bind to a similar epitope as the Merck molecule to help ensure that, you know, the lower immunogenicity based on that, this hypothesis, and also have a higher potent molecule, 'cause the Merck molecule is kind of the least potent out of all the competitor molecules.
So we're able to hit a similar epitope but have higher potency. And I think that this shows the ability to kind of how you can use this epitope specificity from the de novo model to achieve new novel biology that you couldn't have achieved any other way. And then additionally, since we're able to search that larger search space, as Zach was, you know, pointing out with the de novo model, we're, we're able to actually then, you know, carve out new IP landscape for this molecule as, as, as well. And I think, so you're seeing differentiation on, on the, the, the biology side, on the, you know, on the preclinical side, but additionally, you're seeing differentiation on from an IP perspective as, as, as well.
And not to mention, we were able to generate all of this within a 14-month time period, so we were able to get to a drug candidate in 14 months. We'll be in the clinic within 24 months, and again, this compares to that four-six years it normally takes. So you're getting, again, differentiated biology, plus being able to have you know huge time savings when you know driving these assets into the clinic.
And so just follow
Just to add one other point.
Yeah, go ahead, Zach.
So in addition to all the biology that we've engineered, we've also introduced some half-life extension, so we've engineered the Fc portion. And so we've really looked at enabling greater patient convenience on top of engineering and the enhanced biology.
Yep.
Great. Just following up on, because you mentioned the time frame. So when do you expect an IND for the TL1A molecule and then initiating phase I? Do we have specific time frames we can expect for that?
Yeah. So we'll have the IND early 2025, and then the phase I will, you know, kick off shortly thereafter. And then we do plan on having an interim phase I data readout the second half of 2025.
Great. Sticking with your internal pipeline, let's talk about some other assets that you're excited about, and then we'll move on to the partner programs.
Yes, absolutely. So the other, you know, assets that we have is ABS-201, which is a not, or which is a, I would say, a derm target that is definitely underappreciated. This would be a, you know, potential best-in-class asset. There is one other competitor molecule that is in the clinic. We'd be second to the clinic on this, and we plan to have a drug candidate elected by the end of the year, with data coming out shortly thereafter. And then the last one is ABS-301, which came from our reverse immunology platform. This is a potential first-in-class IO target. It's a novel target that came from a patient that had an extraordinary immune response.
And we're looking at the tertiary lymphoid structure. The reverse immunology is really based on this, these tertiary lymphoid structures that show up in various different cancer cells. And the B-cell repertoire in those, you know, TLS structures is very different than what you see in peripheral blood. And we're able to take these samples from patients that have had an extraordinary immune response and be able to de-orphan and discover novel targets, and this is where this target came from. We've done in vitro validation on it, and one of the things we're excited about is that it stimulates the innate immune system over the adaptive immune system.
Right now, we're currently in in vivo validation on that, which we plan to have completed by the end of the year. We'll have a drug candidate shortly thereafter. I think if you take a look at this, you kind of see a theme that it's a holistic pipeline. You know, we have some best-in-class assets, but also some you know, higher risk, higher return, assets as well, and we're gonna continue to kind of focus on this you know, diversification on both first-in-class as well as best-in-class.
And then these partnerships that we have as well allow us to create more diversification as well, partnering with, you know, leading experts like, you know, Merck and AZ with on oncology. You know, we don't have to build out, you know, a full translational team on the oncology side. We can partner with, you know, with these leading, you know, oncology companies and develop differentiated assets there. And so we see this as, again, as a way to continue to build a robust, diversified portfolio.
That's helpful. And maybe, Zach or Sean, maybe talk about the monetization or financial implications of focusing on sort of more narrowed partner programs, but also, you know, focusing on your internal pipeline and taking these things further out, in the clinical trial phase process, and how you, you know, hope to monetize this. You know, what you're seeing, for example, on the TL1A side, the economics on that molecule.
Sure, I can, I can kick off that discussion. I think we're really excited about advancing more and more internal programs, Scott. It's y ou know, by taking some of these programs forward to proof of concept in man, we think the value inflection is very significant. And going back to Sean's comments earlier, we've validated with TL1A that we can do this in an incredibly short amount of time, so two years to get to an IND, spend of roughly $15 million. So, you know, mapping to what pharma does, there's a 5-6x multiplier on kind of what we can do with the same kind of capital investment.
So I think investing more in those internal programs, given that we have a world-class discovery team to help us decide on target selection, it's led by Andreas Busch, makes a ton of sense. And we still wanna keep maintaining and building out some partnerships with pharma. We think there's great diversification there. We look for partners like Merck and Almirall and AZ that bring really great expertise in a given indication space, great target biology expertise that we can leverage, and look for those synergies. But I think more and more you'll see us developing internal programs where we can bring them into IND phase I validation, and then look for really strong partners that could take it for the rest of the development phase and into commercialization.
Yeah. So, so you're ultimately still partnering. It's just a question of, of when you're partnering. I think now, now you're able to take these to value inflection points, you know, faster and cheaper than, than ever before, and you have much higher upside partnering out of phase I versus a phase II, or, you know, phase I or phase II versus partnering at the target phase. And so again, it's just, it's still partnering, it's just pushing out the partnering thing-
Yeah
Y eah, later.
Yeah, that makes sense. Zach, following up on your point about Andreas and his team, is there a shift to strategy on the internal pipeline to developing molecules that already have, you know, competitors in the field? Because you know there's commercialization or there's validated commercial. How to phrase this? There's already kind of been kind of this marketplace assigned, and you can theoretically use your technology to prove efficacy against competing molecules. Is there, is that a legitimate idea or thought that your strategy has shifted towards that, or is that false?
I mean, I wouldn't say we've shifted to it. I'd say we've been very strategic in what we've done here in designing our portfolio. So the two lead programs are fast followers. So to your point, there's validation on the biology, on the target. So we've really reduced the risk there for the lead programs. Now those are markets that are gonna be carved up, and so we've used the AI to design best-in-class features, so to position those molecules really well for market share. But on the back of that, we also wanna explore first-in-class, so novel targets where, particularly with our AI platform, if we do this the way we believe we will, we'll be able to create tremendous amount of IP coverage and composition claims.
Because we're doing all this validation of different antibody structures in our lab, and really go forward and potentially try to lock up the really juicy targets there that are first-in-class, where there's no competitor ahead of us. So we're, we have a mix to give us some diversification, but there's a lot of strategic thinking in which targets we go after, and also diversifying that portfolio of internal assets, so we risk-adjust it across validation versus novel.
That makes sense. I think we're running up in 10 minutes here, so my last question here before I'll turn it over to the audience. And by the way, audience, if you'd like to ask a question, there's a chat box in the bottom of your screen. Please type a question in, submit it, and I'll see it on my end on my portal. So you just had a really successful offering. I think it netted over $80 million in proceeds. I don't know the exact number, what it netted out to.
Oversubscribed. I think that brings your current cash balance almost to $200 million, nearly $200 million cash on the balance sheet. Where are you looking to deploy this capital? Maybe talk about near term, mid-term, longer term. Cause I do believe that you said this should provide you liquidity up through the first half of 2027, if I got the timeframe right.
Yeah. That's right. Where we're looking to deploy the capital is, one, advancing TL1A into the clinic, being able to have the interim data readout in the second half of 2025. Additionally, we're gonna be investing into, you know, 201, 301, getting those to IND, and then additionally continuing to invest in building out more, you know, pipeline assets that you'll see more of in the coming year.
And then, you know, finally, we're continuing to invest in the data generation capabilities that we have, data to continue to train our models, and then additionally investing into our AI team, our AI research, and that additionally includes compute as well. And so those are kind of the main areas of investment, but it's really driving these assets to value inflection points and continuing to develop our AI models.
Great. One question from the audience is: "Do you plan to deploy any capital around M&A, or any bolt-ons, or any tech pieces that you might not already have in-house?
I think it's a great question. We're always looking for technologies and teams that could bolt on to, you know, give us new data generation capabilities, give us new AI capabilities that we do not have, and so we're always on the look and the hunt for that. I would say as the platform stands today, we aren't actively pursuing any opportunities. However, again, if there is technologies that do present themselves or companies, we'll always consider that. And that's been an effective strategy for us in the past with the two acquisitions that we've done with Denovium and Totient.
Another question is: "Can you talk about, large pharma and biotech end markets? Are you seeing more appetite, demand for your technology, or, large pharma coming in to wanna partner with you versus last year?" I guess the question here is, have these end markets been improving, worsening, staying the same?
Yeah. Just a quick comment here. You know, one thing that we've heard consistently, and this fits really nicely with what we're trying to do to build our pipeline out into the early development and into phase I , is that most of the late-stage assets in the market have really been picked over, and so we're seeing the pharma's lens shift to earlier and earlier, so now looking a lot more at phase I and IND-ready assets. So we think, we think we're timed really well over the next one-two to really capitalize on that, and that's another big reason why we want to invest in our own internal pipeline.
Yeah.
And I would say, you know, I guess the other thing to point out is we have a, I think we have a nice business development pipeline going into this year, so we're looking to do some more partnerships at the creation phase as well.
I would say that pharma is also in the show me phase,
Yeah
J ust like investors are, and I think that that's, like, what has really driven the success that we've had with the pharma partnerships that we have closed with AZ and Almirall, is really taking this, you know, de novo model that we had, which, you know, showed a lot of exciting promise, and then apply it to an actual asset, and that's what really drove, you know, both the AZ partnership, the Almirall partnership. And it's like, okay, if you can apply it, you know, to TL1A, we can now go apply it to these, you know, kind of low-hanging fruit, you know, targets, you know, like GPCRs and ion channels that could unlock, you know, huge market opportunity for pharma.
And so I think that they have been very diligent in who they partner with, and it's the companies that have shown the proof that the AI is actually creating differentiated assets. Those are the companies that pharma is, you know, jumping on for partnerships, and so I think you're gonna continue to see that, but it's gonna continue to kind of be this, I think, show me environment.
Yep. Well, we'll leave it there. Thank you so much, Sean and Zach, for participating in our conference. If investors have any more questions, feel free to reach out to me, and I can connect you with the Absci management team. But thank you both. Thank you all for attending.
Yeah, absolutely. Thanks so much, Scott.