Schrödinger, Inc. (SDGR) Earnings Call Transcript & Summary

And I think we are live. Welcome, everybody. My name is Chris Shibutani. I am a member of the Goldman Sachs equity research team. I'm joined by my colleague, CJ Zopf, and the 2 of us are very pleased to have joining us today to present at our conference, the 42nd Annual Goldman Healthcare Conference, still virtual, hopefully, live next year, Schrödinger. The company has done so much to really create an opportunity for investors to think more broadly about investing in the life sciences with so many tangencies relevant to therapeutics investors. So I think this will be a very interesting discussion, and we're going to take it from a bit of an educational approach and have a management team tell us a little bit about their story. Here to join us, we have Ramy Farid, who's President and CEO; Karen Akinsanya, who is the Head of the Biomedical Sciences and the Discovery Group; and Joel Lebowitz, the Chief Financial Officer. So Ramy and Karen, I think you're going to present us with a summary of a presentation. And then we'll chase you with some Q&A. So take it away.

Thanks a lot, Chris. Sounds great. Thank you. Yes. So let me start with a little bit of background. So it's, of course, very widely known that drug discovery and materials design are incredibly hard. The large majority of programs fail to deliver a product. And of course, we're all very familiar with costs that our skyrocketing due in large part to these very high failure rates. So -- and even when programs do succeed, the compounds often have issues. In drug discovery, that's often the result of a therapeutic window not -- being too narrow, not large enough, due in large part to poor properties. So an obvious solution to this problem is to develop computational methods that allow the exploration of vast amounts of chemical space to identify high-quality molecules. So as what's outlined here on this slide, if we could prospectively compute all the relevant molecular properties with sufficiently high accuracy, designing drugs and materials would have, of course, a much higher success rate, obviously be faster and cheaper. And I think this is the most important thing, and I'll emphasize this throughout the talk would yield much higher quality molecules. So where are we now? That's in some sense of vision for the future. So at Schrödinger, we've made really major breakthroughs in the science, and we've developed for the first time technology that can compute key molecular properties with extremely high accuracy, and in fact, accuracy, the approaches that of experiment. Its technology now makes it possible, as you can see here, to explore hundreds of billions of molecules computationally. And the technology is having a really truly transformative impact on drug discovery and materials design. So let me give you sort of a highlight of that. And then we'll get into a little bit more details on programs when Karen talks about our program. So in traditionally run drug discovery, it's typical for about 5,000 molecules to be synthesized, and when successful, it can take 4 to 6 years to get to a development candidate. And it's still often the case, as we're all familiar with, that these development candidates have issues. By leveraging our platform, at scale, and that's a really critical point, we -- and our internal programs, again, we'll talk about that shortly, and our partners that we're collaborating with are able to explore billions of molecules computationally; synthesize far fewer, only about 1,000 molecules on average, and in some cases, again, we'll talk about this, far fewer even than 1,000 molecules, it's taking about half the time to get to a development candidate. And again, and I think this point really needs to be emphasized, with extremely high-quality molecules. Again, we'll touch on that in a second. So we're leveraging this unique platform in several ways. That's what this slide shows. So we have a software business, of course, where we license our platform to pharmaceutical companies, biotech companies, materials companies, universities and government labs worldwide. We're also leveraging our platform in a number of drug discovery programs in collaboration with pharma companies and biotech companies. We also have internal drug discovery pipeline, again, Karen is going to talk about that very shortly. But let me also mention actually in the next slide, let me also just mention before handing it over to Karen, that we have a track record of revenue growth from all of these businesses, and we recently reported Q1 revenue of $32.1 million, up 23% from Q1 last year. So let me hand it over to Karen, and then I'll come back with a few closing remarks.

Thanks, Ramy. So as you've heard from the slide so far, the platform has been deployed actually broadly across the industry. We have a number of collaborations that have been established over the last 10 to 15 years. And you can see here on the left side of this chart that our early collaborations with Agios have actually resulted in FDA-approved drugs. In addition to that, there are a number of companies here, Morphic, Nimbus, that we have had a role in cofounding over the last 10-or-so years. And in those relationships, we've been working with these companies to deploy the platform at maximum scale to identify and deliver development candidates, as Ramy described, over a 2- to 3-year period. And now you can see some of those assets have now moved into clinical studies as far as Phase II and several now entered into Phase I over the recent years. In addition to biotech companies, we also collaborate with large companies. So you can see here companies that also buy the software are working directly with us to again leverage the technology at maximum scale on a number of programs across many different disease areas. And I think it's important to note that many of these collaborative projects are pursuing targets where there is very strong rationale. There's often proof of concept, if not approved agents. And I'll just briefly mention that the Phase I asset in the Morphic portfolio is a small molecule version of what is an antibody drug against alpha4beta7. And in the case of Nimbus the Phase I asset there is a Tyk2 inhibitor. So you can see just 2 examples of where we've leveraged the platform to go after specific design challenges, making a small molecule version of an antibody or a highly differentiated, selective or improved version of an agent that has passed proof of concept. So I'll move on to say that we, over the last few years, have established a very significant capability internally in drug discovery and early development. We use that group to really pursue programs with Takeda. As you can see on this slide, we've also got our internal programs, some of which actually partnered with BMS towards the last half of 2020. So we have 3 programs that are at advanced stages: our MALT1 inhibitor is in IND-enabling studies; CDC7 and WEE1 are close behind; our HIF-2 alpha and SOS1/KRAS programs, which were earlier actually did transact BMS last year as well as 3 other targets. So in terms of timing, these programs have moved very quickly over the last couple of years. And we're excited about moving them forward in 3 different areas of oncology: MALT1 in relapsed and resistant lymphoma; CDC7 actually in hematological tumors as well as solid tumors; and WEE1, which has been the subject of a lot of discussion recently in gynecological cancers as well as solid tumors. So we're excited about moving those programs into the clinic over the course of next year. And I'll pause there. Do you want me to present this one, Ramy, or...

Was that okay, Karen? Yes.

Yes. So you just heard me say that we've selected a development candidate in our MALT1 program. We are now building the capability of full clinical development. And as I mentioned, we're going to be submitting INDs next year with our first expected in the first half of 2022. And finally, as those programs move forward, we're replacing them with additional drug discovery programs across additional disease areas, including immunology.

Great. Thanks, Karen. Good. So I just have a couple of slides just to sort of close things out. So we're obviously very excited about what the future holds. And to give you a sense of what may be the next decade might look like, it's kind of useful to see how we got to where we are now. So of course, we're all very familiar with the fact that computer performance has been increasing exponentially for a long time, notice this is a log plot. So that's what the point of the gray bar there showing that computer performance has been increasing exponentially, we're very well aware of that. But what might be less well known is that actually the number of proteins in the human genome for which we know the 3-dimensional structure to some cutoff or resolution has also been increasing exponentially. The number of compounds that we can explore computationally, not surprisingly, since it's probably largely tied to computer performance, but also improvements in the science is also increasing exponentially as has the number of properties that we can compute with near experimental accuracy. So what does this mean? This means we can reasonably predict over the next decade or so that computers will be in this is kind of an extraordinary thing, more than 100x faster than they are now. And that really can have profound impact on the kinds of simulations that we can imagine doing, and we're already taking advantage of that in a number of really challenging scientific problems. And we've also -- it's not unreasonable to project that maybe 50% of the human genome will have structures for a reasonable resolution, high-resolution structures, for maybe 50% of the human genome. We're estimating that we'll probably be able to predict with very high accuracy as many as 10 molecular properties. And if you remember from the first slide, that's a significant portion of the number of properties that need to be optimized for drug discovery. And this may well lead in the next decade of being able to reliably discover, that's, again, a very important point, high-quality molecules in as little as a year. We continue to see the amount of time that it takes to get to a development candidate decreasing. And as we see these kinds of exponential advances, I think that's a reasonable expectation over the next decade. So it's clearly an exciting time for computer-driven -- or computationally driven drug discovery, and in general, molecular design. So on the last slide, let me just close by highlighting our growth strategy. So we'll continue to advance our computational platform to fully realize the vision for the future that I just described on the previous slide. We are the clear leader at the moment, and we intend to maintain that leadership position by continuing to aggressively invest in the science that underlies the platform. We'll also continue to license our platform to pharma and biotech companies. We learned a lot from doing that -- obviously, that generates revenue, but we also learned a lot from those kinds of interactions, which, of course, gets fed back into the software. We'll continue to advance our collaborative drug discovery programs, and as Karen emphasized, initiate new ones. We'll continue to license our platform to material science companies. There are lots of synergies between material science and life science. The underlying physics is the same, actually, of course. We'll continue to advance our internal drug discovery pipeline, and again, as Karen said, initiate new programs. And we expect actually in the future as well to initiate material science collaborations as we did a number of years ago on the drug discovery side. So -- and a key point I was kind of highlighting this, as going along, is that these multiple business areas and these growth opportunities really have -- there's extraordinary synergies between these areas. The advances in the drug discovery programs is a tremendous validation of the platform, and that's really helping the software business. I already mentioned the synergies in the other way. And we expect to continue to leverage these extraordinary synergies. So hopefully, that gives you a nice overview of the business. We're trying to keep this short, so we can spend time in the Q&A. So happy to hand things back over to Chris, and I'll stop sharing as well.

That's perfect background. And I think we were thinking about this discussion that we have is hopefully create a reference point often for people who are new to the story and just really understand foundationally. Something very important to bear in mind is that you guys have been around for several decades. So I think often in people's minds, the notion of the class of 2019-2020 IPOs are companies that are like three guys and a venture fund and an idea, and all of a sudden, they assembled in Kendall Square somewhere. And that's not your background, that there's tremendous heritage, there's a lot of foundational proprietary work, established business relationships globally. I think there was something like roughly 55% of your business in the U.S., but you have established commercial relationships everywhere. So just to understand a broad perspective, I think, just thinking about two real parts, right? There's kind of the software part of the business, which has really developed over the last several decades; and the more recently more visible aspect is doing it yourself, having your own proprietary drug discovery pipeline and really taking -- using your own home cooking and serving it up for yourself. And so maybe we'll split the discussion along those lines and I'll just start more on the original software side. I think with our audience for this conference, we have a lot of therapeutics-directed kind of investors and your computationally driven drug discovery, and they recognize that it's kind of like a service, it's a tool. We're doing things smarter, quicker, better, et cetera. But perhaps could you just sort of outline specifically because I think there's a phrase that you always use, which is accurate, obviously, that you have a physics-based software design. And can you just clarify sort of what are the capabilities that distinguish this and stand apart from maybe more traditional structure activity relationships and some of the other machine learning-based approaches. I think for everybody it's a big bubble of a concept of computationally driven, but maybe help identify some countries within that continent.

Absolutely, yes. it's a really important question. So -- and in some sense, it's tied to what you were saying at the very beginning about how long we've been around. These methods that we developed and I'm about to describe to you are -- people have been working on these for decades. I'm trying to -- this idea of being able to actually compute the property of molecule isn't new. It's been around for a long time. But it turned out to be an enormously hard problem. And what happened is that the large majority of companies, and again, I'm going to explain what it is specifically, let me just give you a little bit of background, abandoned the idea of it ever working. At the time, computers were too slow. The -- our understanding of molecular interactions was just too nascent. And it was really easy to transition very rapidly in the context of not being able to get these things to work to sort of more straightforward machine learning knowledge based. I mean that's really the way to think about machine learning, AI, it's knowledge based. We had started to accumulate enough knowledge. We can -- and the idea was, well, If we can just develop large enough training sets using the experimental data we have right now, right, all the knowledge, accumulate all that knowledge, we can somehow extrapolate that to an understanding of new chemical space. And it turns out that doesn't work. And now that's where the physics-based methods come. So first, what does it really mean when we say physics based? What it means is really the way to think about it is that we are rigorously computing the property, accounting for all of the terms that are required to really be able to predict the property accurately. So give you an example. If you're trying to predict whether a molecule binds to a protein, you don't only have to -- it's not sufficient to understand what that molecule looks like when it's bounded the protein. You actually have to understand how the molecule behaves in water when it's not bound And you have to understand how the protein is -- behaves in water before the molecule binds. That's just as important as what these 2 things look together. And you have to understand something called entropy which is related to protein motion. So in other words, what you have to do is you have to actually simulate the molecule binding the protein coming out of water and into the protein. And you have to do that at an animistic level of detail. You have to model every atom that's involved in that system, and you have to simulate that molecular motion for a long enough time to build up the proper statistics to get the right answer. So that's what physics-based means. It means essentially getting the right answer for the right reason. And it's a rigorous method. The only requirement is a decent starting point. You need a structure of the protein molecule complex as a starting point to be able to simulate that entire product. And when -- and it turns out when you do that and you do it right and you have the right underlying physics that describes all these motions and you simulate for a long enough time, which requires faster computers, you can get the right answer. You can essentially recapitulate experimental data or you can make predictions prospectively and get the right answer. And here's the key, and this is why I was talking about machine learning at the beginning. With these methods, you can explore in theory all of chemical space. Chemical space is essentially infinite. It's infinitely diverse. And that means that a machine learning method is going to be severely limited to only knowing about what's already known. You can't learn anything new, by definition. With these physics-based methods, these first principles rigorous methods, you can extrapolate a new chemical space, you can explore enormous amounts of chemical space and that's really what's required to solve these really challenging multiparameter optimization problems that we all face in drug discovery and materials design.

And so when we think about what distinguishes your software, your approach, your product offering, we often think about things in terms of what is the secret sauce, what [ intellectual ] property, what barriers to entry are there so that the argument is -- you've already mentioned how difficult it is actually to accomplish what you have and how long it took. Anything in particular to note once you're in the fundamental sort of what are the barriers to entry question?

Yes, yes. Absolutely. So the size of chemical space is estimated to be around 10^50. In other words, the number of ways you can combine carbon, oxygen, nitrogen, the organic elements plus all the other elements you normally see in drugs is estimated to be around 10^50. Now that should give -- that's a really big number. That's like approaching a number of atoms in the universe, just to give you a sense of that scale. What does that mean? That means that the -- in order to understand how in principle any arbitrary molecule out of that 10^50 behaves in water or behaves when it's bound to a protein requires generating a set of parameters that describe molecular motion. It's called the force field. So that's what the field refers to. I know it's a very unusual name. But that force field, which is the set of parameter to describe atomic motion or molecular motion, is incredibly complicated. Requires running a massive number of quantum mechanics calculations actually to parameterized that force field, so that when you run a simulation you get the right answer. That's something that we have put a really significant effort into it. There is not only just the raw compute power that's required to run the calculations, but there's a lot of complexity in developing a set of parameters that can capture or explain essentially all of chemical space, again, thinking about the diversity of it. You can see how daunting a task that is. And again, that's something we've been working out for a long time. And that's something that is allowing us to run these simulations and actually get the right answer. Now there's other things related to actually the molecular dynamic simulation themselves and the free energy methods that now utilize this force field that also has quite a bit of complexity in it. That's also something that has taken really this whole thing, collectively, thousands of person years of effort to really get to this point. Then there's one final thing. Remember, I said the input to these methods is a structure, a 3-dimensional structure of the protein, right? Where is that going to come from? So of course, it can come from experiment. It can come from a crystal structure or a cryo-EM structure. But those structures are usually too -- in fact, essentially always too low resolution to just take them right from experiment and go right into the calculation. These structures generally are missing hydrogen atoms, that makes up almost half of the number of atoms. They don't -- the structure of the water around it. Usually, these structures are missing atoms and missing residues. There are lots of computational methods that have to be applied to basically produce from a crystal structure a structure that's high enough resolution, high enough accuracy to be able to run these simulations. That's also -- there's a lot of technology behind that, and that's stuff that we've developed that's also proprietary to Schrödinger, essentially the technology used to generate the inputs to these physics-based methods.

No. That's very helpful. And I think what's very clear from your response is how folks want to really geek out on the technology that this avenue to do that. I always think about, particularly when I'm training people who are investors, sort of thinking about there's the science, there's the business and then ultimately the stock. And I'd love to be able to kind of touch on each of those. And maybe, Joel, I'll turn to you a little bit in terms of understanding the business side a bit. So when you provide quarterly updates since you've been a public company, there are revenues, in the tens of millions in revenues, and your product is you're selling software. So talk to us a little bit about just some of the basic components of that revenue line and what you think people should really be paying attention to?

Sure. Thanks, Chris. Appreciate that. So actually, just taking a step up from that. Not just software, we actually have revenues from the drug -- emerging revenues from the drug discovery side as well. But in terms of the 2 business segments, drug discovery and software, I'll start with software. About 2/3 of our software business is on-premise and the rest is a combination of what's classified as hosted, a small portion services that support the on-premise deployment and some maintenance. I think the key factors that have been driving growth over the last year since we've been public and before that as well have been a steady upsizing of customers increasing the deployment of our solutions. And I'll give you some examples of that. So in 2020, we -- software grew at 39% for the year. And the primary driver of that was a concentration of companies that decided to make major decisions to increase their deployment of our solutions at a very large scale, much larger than they had previously. And the result was a real surge in the growth rate that we saw in 2020. Large portion of that is driven by the on-premise that I referred to later. So that's a good number to look at. And I think that we also saw that reflected in the metric -- the annual metric that we provide, which is our annual contract value. That grew in 2019 at 19%; and in 2020, it grew at 22%. So we're seeing this pattern of acceleration in 2020. And I think in many ways, that showed what we're capable of. And we're really confident that we can repeat that kind of cycle of growth over the next several years, perhaps not in a -- definitely not in a straight line, as we've talked about because typically following some of those decisions to upsize significantly, there's a period of assessment to ensure that there's a significant impact on their program. But we're really confident that we can repeat that cycle because we're seeing the benefits of the large-scale deployment of the software on our own programs and how fast they're moving. And just remind you that we're deploying the software at very large multiples. Very large multiple, higher level than our -- even our largest software customers. So there's a lot of room for our largest customers to continue to grow on that side. I'd also -- we're also very excited about the early stages of our -- and development of our material science business. We haven't disclosed a lot about that yet. But I think over the next couple of years, you're going to hear more about that, and we see a lot of opportunity there. On the discovery side, I would look at the continued kind of multiyear pattern of growth there. It's not the kind of revenue that is amenable to quarter-to-quarter smooth growth rate prediction. But I think if you look back a few years and then look at the revenue now and then look at the opportunity that we have going forward with the addition of the BMS arrangement, which gives us an opportunity to capture very large sets of milestones, a lot of which are precommercial, so in the relatively near term, I think you see a trend there. And I think it's important to look for that over, say, a longer period of time, but you want to see certainly directional momentum there, and we're confident that we're achieving that. So -- and then ultimately, we do expect, as we drive value in our internal programs, revenue contribution at a pretty significant level in one form or another from our internal programs, whether they be out-licensing or some other type of arrangement that will allow us to capture that value.

Right? I think so many important points that you had there. And if you look at the stock performance, it represents as much in education of the investment -- investor community in terms of understanding kind of the rhythm and the cadence. I mean you guys aren't selling hip implants. There's not a seasonality. There's is kind of a longer-term arc. So every public company is required to make quarterly disclosures. I think perhaps people should be thinking about things over sort of accumulated rolling 9-month periods or whatever, the things that you talk about, the, I guess, the ACV or sort of like the annual contract value, the percentage of which you penetrate. And again, I think we talked about you 50%-plus in the U.S., 20% in Europe, you're a very well-established business. And there's a level of understanding, there's proof points that go back to Agios signing up to this product, and those products have been approved and commercialized for a while. So I think from people's perspective, to understand that this is very well established but the quarterly metric is probably just not the right way to think about this. And then to be clear, the material science is an opportunity that goes beyond your traditional drug discovery pharmaceutical customer, and that diversification of the base, I think, seems as if it will serve you well in terms of expanding what the rhythms of that kind of customer will be like. Probably the sentinel event that defined why you chose to go public, and maybe I'm drawing the conclusion, was your decision to develop your own internal drug discovery effort. And I'll turn it over to my colleague, CJ, to talk to Karen a little bit about helping us understand what we should be keeping an eye on. CJ?

Thanks. So maybe first to start, can you give us a sense of what sort of philosophy of target selection is enabled by the discovery platform that you guys have built.

Yes, certainly. Thanks, CJ. I think that one of the themes that I touched on in the presentation was the idea that with the platform, we can really solve very big design challenges. And so when we think about picking targets, it's sort of a 3-pronged approach. Number one is the target amenable to our software. The other is what is the design challenge we're trying to solve. And what I mean by that is there are obviously molecules that are out there. We described earlier on turning an antibody that has not just proof of concept, but market presence into a small molecule so that it's more convenient for patients to take for a chronic indication. In other cases, it's coming up with a much improved therapeutic index because you're able to dial out off-target activity and make for a drug that has great properties that will be very well accepted by physicians and patients. And so we look very carefully at those design challenges when we're picking targets. And then the third element, of course, is how well validated is the therapeutic concept. Is there genetic evidence, human genetic evidence? Is there biological rationale pathway precedents or information from the clinic that would lead us to believe this is going to be a very important molecule target to go after? So those are the principles with which we proceed when we're selecting targets. And for the 5 programs that we kicked off over the last 2.5 years, each of those was in a space where we felt that there was a really strong rationale to pursue those programs. And here we are now 2 years later, having solved a lot of those design challenges with target product profiles accomplished and moving forward into the next phase.

Are there any sort of chemical challenges that show in your platform is uniquely positioned to be able to overcome in terms of the structure of the target that you're going after.

Yes. I mean I think we have a very strong track record working on kinases, just as one example, where it's often very challenging to dial in selectivity. But that's not necessarily representative of the capabilities the platform really allows for. And what I mean by that is if you look across that portfolio of collaborative programs as well as our internal programs, you'll see a representative pretty much from every class of targets. So GPCRs, kinases very complex proteins like the integrins. Really, the key here is being able to accurately predict the properties of molecules as they interact with proteins at either a binding site, a protein-protein interaction interface. It's really the accuracy of these predictions that allow you to get to very selective molecules. But on top of that, as we know, you can come up with a very interesting molecule that then has a lot of liabilities with regard to drug-like properties. And drug-like properties encompasses things like solubility, permeability, the sort of cleanliness, if you like, of the off-target profile when it comes to ADME properties, brain penetration. I mean these are all the characteristics that one needs to dial in while holding potency in order to get a very attractive drug molecule. And so being able to predict so many of those properties with the physics-based platform allows us really to come up with what we believe to be somewhat unique molecules that have novelty, potency, selectivity, but also great drug-like properties.

Got it. So you highlighted a couple of collaborator programs with Morphic and Nimbus that have progressed to the clinic and starting to show some of that validation. For the -- your own internal pipeline that you're working on, I think MALT1 is your lead program at the moment. So can you give us a sense of what were the chemical challenges that you overcame in discovery there? And sort of what the clinical context is that program will be entering into?

Yes, certainly. So just to start with why MALT1. MALT1 is a target that is in the BTK NF-kappaB pathway for short. NF-kappaB has been known to be a driver of many types of cancer. And in the case of ABC DLBCL and the number of other lymphomas, including CLL, NF-kappaB is overactive. And so the idea there was in settings where BTK is actually no longer working, relapsed or resistant CLL or ABC DLBCL, if you can further dampen down NF-kappaB signaling, you have an opportunity to really recover antitumor activity in those patients. So that was the premise of going after MALT1, which is downstream of BTK and upstream of NF-kappaB. In terms of a design challenge, NF -- MALT1 is a protease, and traditional protease inhibitors at the orthosteric site were peptidic large molecules with very poor drug-like properties. A few years ago, it became apparent that there was an allosteric site that one could go after to create drug-like small molecules. And so we used our approach with great structures to be able to go after allosteric small molecules. But another challenge was the idea, again, that you could get very potent inhibitors that have great drug-like properties and have very limited off-target activity. And so that's what we've said about designing and actually in record time, I think we had the development candidate by 14 months after initiation of the program, after synthesizing just 78 molecules actually. We had what we believe to be the most potent MALT1 inhibitors described publicly so far with great drug-like properties. And in vivo, we've been able to show in patient-derived tumors regression of ABC DLBCL, both in ibrutinib-sensitive, but also more importantly, in the ibrutinib-resistant setting.

And this more potent, more selective profile, does this allow you to take any sort of different clinical strategy than others might have taken before? And when might we start to see, I think, some of that early proof of biology data.

Yes. So MALT1, I guess, at some level, is still relatively new mechanism in the clinic. There isn't a lot of clinical data that has been presented. In fact, we're expecting to see data this coming December at ASH. So in terms of clinical strategy, having a molecule that is well positioned to combine is, in our opinion, very important. What I mean by that is having a low-dose molecule so that you have the opportunity to maximize efficacy and doing that in a way that allows you to combine very quickly with BTK inhibitors, for example, or indeed with venetoclax. When you have a combination agent that has its own safety issues, you don't really want to add to those safety issues with a second molecule. And so having a very clean molecule going into the clinic that you can combine very readily with BTK inhibitors, with venetoclax, that's our strategy to go into initial clinical trials to demonstrate proof of biology, first of all. There are biomarkers, IL-10, BCL10 cleavage that we've been able to establish are great opportunities to see activity in this pathway. So next year, when we open up a clinical trial, we'll be looking for proof of biology. And proof of concept, I think, is really demonstrating responses in patients, which should follow within the next year over the course of 2022. As you know, there are others in the clinic. Janssen is in the clinic and they've already announced their expansion cohort. So we hope to see data from that trial establishing proof-of-concept for this mechanism soon.

And for the CDC7 program, Takeda has been working on this for a little while. What did you see there that made it seem validated, but with room for improvement? And how do you think you guys can carve out your niche in terms of monotherapy or combination approaches?

Yes, absolutely. So CDC7 actually has been around for a while. It's about 10 years, I think, since the first molecule in the hands of BMS and [ Navion ], I think it was or Exelixis went into Phase I. The challenge in that space is being able to get, again, extremely pertinent molecules. These are kinase inhibitors where we believe actually in order to shut down the cell cycle, you need not just potent molecules. You need molecules that are able to do that over a 24-hour period to really shut down the cell cycle, push these cells into a replication stress and into apoptosis before you then back off. And so that was the goal for the program, again, getting to the point we have really potent molecules that have great PK properties as well as other good drug-like properties. And we've been able to accomplish that. And in terms of differentiating from other molecules, it's been really hard to dial in those broad set of properties, the potency as well as the drug-like properties. You mentioned Takeda's molecule. They, at ASCO, I believe it was in 2019 presented monotherapy responses in a number of different solid tumor types. The Phase II data hasn't yet been published. But those monotherapy responses, I think, gave a lot of people renewed interest in CDC7. From our point of view, being able to get these molecules not just into the cell, but also into the nucleus where effectively they are doing their work means you need really permeable soluble molecules in order to be able to, as I said, damp down CDC7 and keep it that way for 24 hours or longer. And so the PK/PD profile of our molecules is showing that not only can you do that, you can get these molecules into the nucleus, really shut down CDC7. You can do that and withdraw treatment in a sort of dosing holiday perspective and maintain antitumor activity to the point you're getting regression in solid tumors and also in AML. So Takeda did not, as far as we're aware, pursue any HemOnc indications or tumor types in their trial. We've now looked at a broad range of tumor types and we've identified sensitive tumor types, including AML, that because of high levels of replication stress, you're able to see really profound antitumor activity. And so our plan is to pursue that into the clinic as well as look at some other sensitive solid tumor types that we've identified.

Great. I think we're almost at the end, if not, we're a little bit into over time. So we'll wrap it from here. I think we've had a very helpful and thorough introduction to the many dimensions of the company. And I would say to investors at this point that this trio of folks who are here have tremendous credibility. Ramy, you've been at the lead of this company now for almost 2 decades, many roles that you've played have included on the business strategic front. You're obviously very able to speak on the technical aspects. But certainly from the history that includes spinning off companies and efforts that you've been involved with the formation and whether it's Morphic or Nimbus demonstrates an ability to make some of those strategic decisions. Joel, you and I go way back. I know that you were a student of Judy Lewent, one of the most prominent CFOs from Merck. So tremendous capability, credibility, have sat on the side of the large pharma partners. We're very much your customers there. I think investors should get comfortable with that. And as Karen with her mellifluously intones about the incredible science that you're working on, I think what's interesting is that this particular time, when you look at the stock, there have been a couple of episodes where people have had a little bit of trying to get comfortable with the notion of how the revenues flow on the software side. But after a couple of laps around the track, it's just I think people should be getting more comfortable. And I think AACR this spring was an important debut to begin to really unveil more about what's happening with the drug discovery. WEE1 is a topic that CJ and I are looking forward to continuing to dig further into and it's something that I think will have greater visibility. So for the therapeutics investors who have joined us for this presentation, the waters are certainly much safer and the nature of the approach on the proprietary drug discovery side, in our opinion, looks like it's going to really continue to emerge. So this is, I think, a really good opportunity to begin to do more work and to be attentive to the fact that there's a solid revenue business on the software side, which we therapeutics people are kind of not so great at figuring out. But at this point, there's enough of a track record to see behaviorally but that there's opportunities. And one final shout-out, Karen did a wonderful long-form interview with Luke Timmerman that's available from January 2019. If anybody needs, for whatever reason, conviction that she's been there and done that from all the [ path ] and the people that have advised her at Merck, I think that's a very compelling listen as well. So thank you, all of you, umlaut included for Schrödinger, and we look forward to continuing our dialogue.

Thank you, Chris. Thank you, CJ. It's always a pleasure.

Take care. Bye-bye.