%20(2).jpg)
Ideagen Radio
Ideagen Radio
Thras Kayardis: AI Revolution in Drug Discovery & DeepCure's Innovative Future
Join us as we unravel the fascinating journey of Thrasyvoulos Karydis, or Thras, the brilliant mind behind DeepCure's technological revolution in drug discovery. From the serene landscapes of Corfu, Greece, to the bustling halls of MIT, Thras's path is a testament to the power of curiosity and innovation. With a robust background in electrical engineering, computer science, and physics, Thras ventured into the biotechnological frontier, merging AI with molecular biology to spearhead groundbreaking advancements in immunology and inflammation treatments. Discover how his transformative experiences at the MIT Media Lab ignited the spark that would become DeepCure, a company at the cutting edge of therapeutic development.
In our conversation, we explore the synergetic power of AI and human expertise in overcoming the traditional constraints of drug design. Thras shares real-life narratives of initial challenges turning into triumphs, illustrating AI's potential to reshape the landscape of drug discovery. Thras discusses DeepCure's ambitious goals, including upcoming clinical trials for DC9476 targeting rheumatoid arthritis. As DeepCure continues to push boundaries with their innovative platform and open-minded approach, find out how they plan to repeatedly design transformative therapeutics and make a lasting impact on patient lives. Don't miss this insightful episode that promises a glimpse into the future of healthcare innovation.
#ideagenglobal #globalpartnerships #deepcure
Thras's LinkedIn: https://www.linkedin.com/in/thkarydis/
Learn more about DeepCure here: https://www.deepcure.ai/
Watch the entire Global Partnerships Summit here: https://www.ideagenglobal.com/2024-global-partnerships-summit
Welcome to IdeaGen TV. Today, I am incredibly honored to have with us Thrasybulos Karidis, also known as Thras, co-founder and CTO of DeepCure Thras welcome.
Speaker 2:Pleasure to be here, George. Thank you so much for the invitation and excited to have a conversation with you.
Speaker 1:It's very exciting. You hail from the great island in Greece of Corfu from the great island in Greece of Corfu. We were talking pre-interview about how we hosted a summit global summit there recently and, wow, it's just amazing to know that you're from that incredible island. So I wanted to start with that and then deep dive into our conversation. Today we're gonna be talking a little bit about your role as CTO of DeepCure and how DeepCure is working on new drugs for immunology and inflammation, which obviously is a global problem that's estimated to contribute to 50 percent of deaths. Could you please introduce yourself, of course, and then, for our global audience, describe the incredible work, if you can, which is so much. But in this short interview, can you please describe what you're doing at DeepCure?
Speaker 2:Definitely. Thank you, george. So I'm Thrasylbulus carides, or Thrasybulus carides, as George said, started from Greece and my background is in electrical engineering, computer science and physics. And then I did my master's and PhD at MIT, where I work on the intersection between molecular biology and AI, and throughout the years that I was at MIT, I met a lot of great academics, but also, as part of the Media Lab, we had a lot of partnerships with pharmaceutical companies. So I ended up getting exposed both to small-scale and large-scale pharma, worked on different positions in startups, large-scale pharma, worked on different positions in startups and then, as part also of my background, I was um teaching, prototyping and fabrication in the global fablab movement, so I've taught, you know, students from cambridge in may to rwanda, in africa, to bhutan, indonesia. So I had a very extensive software and hardware exposure while I was at MIT and also exposure to a lot of the real life problems that industries like the pharmaceutical one face.
Speaker 2:And around 2018, we decided with a lab mate and our professor there to start DeepCure, and I'm now at the role of the CTO and leader of the platform team at DeepCure. Deepcure is a biotech company and we're working on inflammation and immunology, which is a huge disease area with a lot of medical need, and we're developing small molecule therapeutics. We currently have our leading asset on its way to the clinic it was nominated as development candidate and then we have also our discovery pipeline, working on extremely impactful targets like STAT6, with the potential to produce transformative therapeutics for inflammatory indications. As part of my role at DeepCure, I'm actually leading the platform component, which includes not only artificial intelligence and machine learning components that's where we started but also spans across physics and automation. So our goal is to design an end-to-end pipeline that starts working from challenging targets for impactful diseases and produces therapeutics quickly and effectively in this what we call the next generation drug discovery process.
Speaker 1:What we call the next generation drug discovery process Incredible, and one of the you really focus a bit on the media lab. I know Nick Negroponte was one of the founders of the media lab. In fact, oddly enough, I was having a coffee with his brother, the former ambassador John Negroponte, and we were talking about that when they were cutting the ribbon for the opening of the Media Lab, they had their mom there and the mom said my God, what, what is this place like? What is this like? What is he doing? And who would have known so many brilliant minds would be? It was so ahead of its time.
Speaker 2:The media left, right and and to hear that you were part of that.
Speaker 1:I just wanted to zero in on that for a moment, because it's uh, it's really phenomenal. I'd like to to go a bit further. Um to us about deep cures from formation. Could you tell us a little bit about that as well, about deep cures from formation?
Speaker 2:Could you tell us a little bit about that as well? Definitely, definitely. I'll use actually the perfect segue that you just described the MIT Media Lab as a place to be, and the Media Lab was always a pioneering institution with respect to combining the foundational knowledge across different sciences like physics and computer science and transforming that into products that can actually be deployed in the real world. And our group in the Media Lab, together with Professor George Jacobson and now my co-founder, fritz Schreiber, was working on exactly that combination of fundamental physics and computer science, translating that into the design of different therapeutics like, for example, antibodies and small molecules, specifically on the small molecule side, so designing explicit chemical compounds that can interact with proteins, which is the targets in the human organs that we want to modulate to affect disease. We had some really exciting results Around 2017, 2018,. We built a platform that would very, very quickly search very large databases of chemicals and identify the most promising ones that can work on any given target that a pharmaceutical company would be working on. And, as part of the sponsorship part of the Media Lab, we got communicated this result to big pharma executives and I remember very vividly one of them being so surprised and so excited about what we were showing him on the screen, that we thought to ourselves, okay, this is actually. This goes beyond an academic project, this is now something that we can deploy in real life and help the drug discovery process to make it faster. The drug discovery process to make it faster.
Speaker 2:Now, of course, that was just the beginning and it gave us kind of the push to graduate in a sense, from the Media Lab and start DeepCure.
Speaker 2:But since then we've been realizing, and kind of it's now part of our DNA, that drug discovery is so much more than a faster AI algorithm. Drug discovery is an entire process. It involves multiple organizations, multiple components, both software, hardware, experimental, the entire what's called the design, make, test, analyze cycle. So as part of building DeepCure, we built essentially a new, we built essentially a new biotech organization that gives equal attention to both the therapeutics aspect, the sciences biology, the chemistry, pharmacology but also the computer science, the automation experts, the machine learning scientists that could enable the end-to-end design of a platform that produces therapeutics that can go to the clinic. And I think we're one of the few companies of our cohort, let's say of AI and drug discovery companies, that pay that equal attention, which led to the fact that we now have our first candidate all the way to the clinic and enabled us to truly work on novel and challenging targets and indications well, that is incredible to hear thrust, because there's so much related to this whole area of immunology and inflammation.
Speaker 1:And then you know, the fact that you all are tackling it directly is just profound, and I know that, um, the benefits and the potential benefits to humanity are beyond what we can even fathom, uh, during this conversation, so that that is just amazing. Um, I'd like to ask you about that question though, drus is is what is in, in your opinion, the importance of immunology and inflammation and why is that driving you and Deep Cure so quickly to attack this issue?
Speaker 2:Great question, George, and it's very close to our heart actually, and I've been exposed to this throughout the past few years as part of being in this in DeepCure as an organization, and I've been astounded to find out that recent studies show that inflammation actually and specifically chronic inflammation, systemic chronic inflammation might be the underlying cause for almost any disease you can imagine you know cardiovascular diseases, diabetes, Alzheimer's, even through neuroinflammation, and it's such an unmet clinical need. It's one of these cases where we understand a lot of the biology behind it. We understand the different pieces of our immune system, which is actually highly complex, of our immune system, which is actually highly complex, but we have been. It's hard for us to actually design drugs for it and figure out ways so that we can modulate the disease because of all the intricacies. You can imagine our immune system as being, you know, a system where it has a thousand different switches and you have to just flip the right ones so that you can actually stop the over-inflammatory response that has no reason to be there and actually not stop the proper immune response to, for example, a viral or bacterial infection. So for us, the talent was the drive, but also the opportunity as a company that specializes in solving hard chemistry problems, Because the if you would say that this therapeutic space on inflammation has been focusing so much on biologics, so antibodies which address part of the disease but have difficulties, like, for example, acquired resistance or the fact that someone needs to go like, for example, acquired resistance, or the fact that someone needs to go, you know, a couple of times a month to get injections and there is not much attention to the small molecule space.
Speaker 2:And there is not much attention because it's very hard to design these drugs. You have problems like two proteins coming together and forming protein protein interactions, which are notoriously hard to design. A therapeutic that actually can can block them. Or you have fundamental science challenges on the way that these, that these chemical compounds, the potential drugs, can be both effective but also safe, Because chronic inflammation is a disease that you're going to have potentially for the rest of your life, so you have to be taking these drugs for very long periods. So there need to be both effective and efficacious and this has been a challenge and that's where I believe DeepEar can have a real impact in helping to design this next generation therapeutics, and that's a huge drive. So there is both the identification of a really big problem and an opportunity that our technology can help design therapeutics for.
Speaker 1:And so that's to hear, the trajectory and the reason behind your work is really so inspiring. And so what has been the issue, that overriding issue with traditional drug discovery for these types of diseases that you alluded to?
Speaker 2:Exactly so. The main issue has been that, as I mentioned above, those targets, so those proteins involving our immune response are so indirectly connected to each other that they pose novel challenges that we haven't tackled before with the traditional methods. To be able to design the next generation of drugs. And one part, you know, to solve, one part of this big challenge is actually the chemical space that you are, that we have available, you know, at your arsenal, to be able to find molecules that satisfy all the different requirements to you know, target only the specific part of the protein that you want and avoid anything else that could cause a harmful adverse effect to the patient. And this exploration of the chemical space has been very limited by the traditional method. We tend to look, as scientists, we tend to look to the successes of the past, and the success of the past has been on what's called in the industry, the low-hanging fruits. You know we have designed molecules that you know were much easier to to to become safe drugs, say, therapeutics. Now, with inflammation in knowledge, that becomes so much harder.
Speaker 2:A second big piece there is the fact that sometimes we're not even asking the right problem.
Speaker 2:When a traditional medicinal chemistry pipeline starts, you usually start with a structure of your protein, a visualization, a model, if you would say in the computer, of where that protein interacts with another protein or another molecule.
Speaker 2:And people have been focusing on the interactions that are known, the interactions that either a prior molecule or molecule we have, a designed molecule or a molecule we have in our body goes, but they tend to ignore all the rest of the protein, they don't look beyond what is known and you end up in this trap of design molecules that have the same limitations, the same problems to be developed as what has been in the past. So that's, those are two of the fundamental challenges of the of the design process of drug discovery and, of course, last but not least, there is a very big talent of the translation of the virtual to the physical world. How do you actually I mean, you can design the perfect molecule, but how do you actually make it in real life? And chemical synthesis, which is for the field of drug discovery, is fundamentally based in manually processes. You know, chemists, actually pipetting corollaries, liquids and making molecules together is very inefficient and is a very big bottleneck for innovation in that space.
Speaker 1:So, as you look at this, you've talked about the drug discovery and you know how it applies to traditional drug discovery, etc. Would you be able to walk us through a successful case where your AI driven platform led to a significant breakthrough in drug discovery?
Speaker 2:Exactly exactly, and I think that for us, we have seen the potential of AI even from the beginning of the company. But I would love to describe a moment that actually made us say, actually really realize the true impact that we can have. And it goes exactly to the point I was making before about the restrictions and biases that exist in the current chemistry process. We were working in a target that had a very challenging set of requirements to design a drug. It had to be very efficacious to the protein that we're going after. It had to be extremely selective to a very closely related protein and then also that molecule that we designed had to be able to enter the blood brain barrier. It had to be brain penetrant. All these requirements pose a very hard challenge and a very unique profile of the molecule that we're looking for. So we build our AI models and searched the world's largest chemical database at the point more than 20 billion molecules and we synthesized the top results and we actually didn't get any hit. We didn't get any molecule that would actually fit the profile of the drug that we're looking for. So then we said, okay, what is the underlying issue here? And we looked at the scores and we realized that we can design molecules and we can synthesize them with custom synthesis. Design molecules and we can synthesize them with custom synthesis for that maximize the scores of these AI models that actually are a perfect fit, or they seem to be a perfect fit for what we're looking for.
Speaker 2:And that's when, you know, we got this amazing result where we synthesize 12 molecules and seven out of the 12 were excellent binders, as we say. They had excellent efficacy against this protein and then actually four out of the 12 met all the requirements that we had for the design. So we realized at this point, which was an aha moment for us, that AI models are not a problem. We realized that our AI modeling actually works and it can enable the design of these molecules, and we then focused a lot on the second piece, which led us to develop an automated robotic chemical synthesis lab that would enable the rapid design, test and analysis of small molecules. So we realized on this project that by expanding the chemical space, by removing the restrictions and limitations that existed in the types of molecules that we can search and we can also make in real life, we can have a tremendous impact and drive forward the programs much faster than before.
Speaker 1:Ross, the process you described is just incredible, and I'd like to go even further, if we can. We have heard so much inspiration and so much positivity in the drug development space as it relates to the AI-driven platform that you all have developed at DeepCure, and so we hear a lot. We've heard so much about AI. Right, we hear it. It's ubiquitous. The most remote conversation somehow touches upon AI, it seems like today, and we also hear that AI is replacing or may replace someday scientists, is replacing or may replace someday scientists. However, however, we have also heard that DeepCure believes in scientists, actual living, breathing scientists and AI working together. Kindly tell us more about that.
Speaker 2:Really interested in hearing the DeepCure perspective. Great, great question, george, and I think it's sort of very close to our heart and the DNA of the company. As I mentioned before, deepcure is built at the exactly at the intersection of the various disciplines that are required to be able to have a functioning drug discovery company drug discovery and development company, and these include, of course, not only AI but also all the other technologies and all the other important team members and people that contribute to the design and development of a drug. Specifically around the AI vs human question, the way I view it is that we should establish a co-creative process between the AI models, the AI platform, the AI systems that we develop and the scientists that surround it and are driving the project forward.
Speaker 2:So, instead of, you know, ignoring the years of experience and knowledge and expertise that the scientists have in that space, what we need to do is just make sure that we remove all the elements that we don't want, so, for example, the different biases that get introduced in the process, as I mentioned before and focus on an interactive conversation, if you would say, between the ideas that an AI model proposes for the design of these molecules and the experience and ability to quickly evaluate and interpret the results that a scientist working many years in drug discovery has.
Speaker 2:So we focus a lot on elements like the explainability of our AI models and interpretability. So display to the, to the science, is not only a result that comes from a black box model, but build a or have a process that builds conviction that actually explains to the scientists why this molecule was selected, was created by the AI model, and what are all the insights that can be generated on why this would be a great molecule. So together, hand in hand, the humans, the medicinal chemists, the structural biologists, everyone involved in this process can drive the project forward, iteration by iteration.
Speaker 1:Quicker and more efficiently perhaps and that's really what we're talking about is getting to that solution set to help humanity, and that thrust is incredible. I want to shift just a bit to a related topic about the focus on partnerships. Thras, tell us why and how are partnerships helpful for DeepCure?
Speaker 2:Great. So you know, drug discovery and development is a team effort. There are so many different components to it, starting from the identification of the targets and the disease areas that you want to work on, all the way to getting a drug to the market and for DeepCure. Partnerships are essential both at the technology level and also at the therapeutics and pipeline level. In general, the goal, as we just said, is to make the design, make, test analyze cycle as fast as possible. For that we had an excellent partnership with a company called BioCero who helped us create the world's most advanced automated chemistry platform. Helped us create the world's most advanced automated chemistry platform Very quickly.
Speaker 2:We were able to integrate our software expertise on designing molecules and handling all of the different reaction data that we had in our models with a very advanced robotic system that would very quickly integrate different equipment and result in a very efficient process in making thousands of complex molecules within a month. At the same time, the partnerships on the therapeutics front are equally as important. We are a company that can solve really hard chemistry problems, so we are actually looking for, you know, all the design specifications, all the problem statements that can be very impactful for real patients. So I want to highlight one of our recent partnerships with one of the world's most advanced rheumatoid arthritis centers and the University of Leeds, where they help us, you know, design the perfect studies and the perfect clinical trial around our lead candidate for rheumatoid arthritis. It's extremely important to have these interdisciplinary partnerships so that you can utilize the best of your strengths and the best of the strengths of the partnering organizations.
Speaker 1:And as we approach the conclusion of this incredible interview, Rás, what does the future look like for you and your work at DeepCure?
Speaker 2:Great question, george, and the future looks very exciting. I'd say, especially for me, that you know we started with the hope of, you know, one day getting a real therapeutic to patients, actually impacting patients lives, and we're now getting closer and closer. I'm really excited that in 2025, we're going to be entering the first clinical trials of the company. We're going to have our first year first in man studies, and this is with our leading program, what we call DC9476. And it's going to be alongside inflammatory indications like rheumatoid arthritis. But then also, I'm extremely excited about the potential of our platform to do this again and again.
Speaker 2:So DeepQ was started with the hope that it's not going to be, you know, one drug and then that's it. Both Kfir and myself, as the founders, we were excited about the opportunity to unlock core capabilities in the, in chemistry that can help us design transformative therapeutics on various biological indications. So, as I mentioned early on in the interview, we're working on a target called STAT6, one of the most challenging in the area of inflammatory indications, and we have some very exciting preclinical results that hopefully will translate later on in advancing our program. That hopefully, will translate later on in advancing our program. So I'm excited at the potential both of the going directly to the patient and actually seeing the molecules we designed help people, but also of the even greater opportunity to do this again and again and essentially prove that we have made an impact in the way that we design therapeutics.
Speaker 1:And Trast. I want to express a deep note of gratitude to you and to your entire team at Deep Cure for what you all are collectively doing to change the world. It is profound. We wanna invite you back soon to update us again on how things are going and what is new, because there's so much and you're moving so fast and we're so excited to hear more in the future about all of the incredible work you're doing at DeepCure. How can folks find out more about your work at DeepCure?
Speaker 2:Thank you, george, and they can visit our brand new website that was very recently updated. This is at deepcureai. They can also find us on LinkedIn, and we're very active these days in participating in several conferences, both in the therapeutics area and in the AI indirect discovery area. I was just in four conferences last month and have a lot of upcoming ones, so we're always happy to chat with people. We're always happy to learn about new ideas, new ways that we can help the lives of patients.
Speaker 1:Geras Gariidis, co-founder and CTO of DeepCure, changing the world, helping humanity. Thank you so very much.
Speaker 2:Thank you very much, George.
Speaker 1:It's my pleasure.