Top Tweets for #Recursion

For machine learning models to be lifelong learners, they must adapt to new tasks without forgetting previous capabilities. For continuous-time generative models, like flow matching and diffusion models, the challenge isn’t only absorbing new information, but also remembering what came before. In a new work, we introduce trajectory replay, a continual learning technique that is designed to remind the model what it already knows. Traditional techniques, like generative replay, use generated samples of the historical data distribution to retrain the model on what it’s previously seen. But these methods can be expensive, and errors in the generated samples accumulate over time. By contrast, trajectory replay involves retraining the model on the velocity of the previous model’s generative path, without ever generating a full sample. This approach is cheaper and more robust to errors, which is crucial for lifelong learning. For more information about trajectory replay and how it can be used to support the continual learning of perturbation models in new biological contexts, check out the links below. Paper: https://t.co/2gS6NkbYnr Code: https://t.co/18CI86WnRX

"This is only possible because we control the design, make, and test steps within one connected system rather than handing off between disconnected tools and teams." Once molecules are designed in silico, they must be synthesized and tested in the real world. In Part 7 of our platform series, VP of Pharmacology Martin Redhead, Ph.D., explains how these candidate designs move directly into our automated DMTL (Design, Make, Test, Learn) labs. Recursion’s automated chemistry platform goes far beyond simple one-pot reactions. We execute complex, multi-step medicinal chemistry at scale. Advanced automation handles the workup, purification, and digital sample tracking for thousands of compounds, creating a digital record of every single experiment. Our in-house AI agents take this a step further by actively helping scientists plan efficient synthetic routes. Because we control the entire design, make, and test lifecycle within one fully integrated system, we completely eliminate the friction and time lost handing off data between disconnected teams. 🚀 Follow along over the coming weeks as we continue this series, covering how Recursion maps biology, optimizes precision drug design, and closes the gap between patient and discovery. #TechBio #DrugDesign #ComputationalChemistry #DrugDiscovery #MachineLearning

Recursion Announces New Clinical Data Across Two Potential First-in-Class Programs, Sanofi Partnership Milestone, and Oversubscribed $200M Offering Today, we announced important progress across our clinical pipeline, partnerships, and capital position—reflecting growing evidence of what our integrated AI-enabled drug discovery platform can deliver. 🚀 REC-4881 | Deep, durable polyp reductions in FAP In an expanded adult population (18+), 87% of patients achieved polyp burden reductions after 12 weeks (47% median), with 85% remaining below baseline after 12 weeks off treatment. In the upper GI—the area of greatest unmet need—92% achieved reductions (55% median), including two patients with near-complete reductions (≥98%). Of those with reductions, 92% remained below baseline after 12 weeks off treatment. With no approved pharmacologic therapies for FAP, we are targeting registrational trial initiation in 2H 2027, pending FDA discussions. Full data will be presented November 2 at CGA-IGC. 🚀 REC-1245 | Encouraging early activity in heavily pretreated cancer patients Our potential first-in-class oral RBM39 degrader demonstrated three partial responses (two confirmed in endometrial cancer; one unconfirmed in liver cancer) and 91.7% disease control (11/12) at doses ≥45 mg. No dose-limiting toxicities or treatment-related serious adverse events were observed at the selected 45 mg expansion dose. 🚀 Sanofi | First development candidate optioned Two challenging targets, one oral molecule. A potential first-in-class candidate for inflammatory disease, enabled by Recursion’s AI platform and small-molecule chemistry capabilities alongside Sanofi’s expertise. Phase 1 expected in 2027, marking the sixth progress-based milestone across our collaboration. 🚀 $200M Oversubscribed Follow-On Offering | Extending Runway Into 2029 Broad demand from leading healthcare specialists, long-only institutions, and growth investors, including Janus Henderson Investors, Invus, Longaeva Partners, ARK Invest, and Driehaus Capital Management. The financing supports advancement of our clinical pipeline, continued AI engine innovation, and strategic partnerships. Multiple independent proof points. Across diseases, programs, and partnerships. Powered by one integrated AI-enabled engine. Thank you to our employees, collaborators, clinical investigators, patients, and investors who make this progress possible. Links below 👇 Business Update: https://t.co/mauWHLYph4 Sanofi announcement: https://t.co/NXtdrzWiGr Public offering: https://t.co/nUr3XMg7Zg

"Our integrated system narrows millions of molecules down to a short list that's synthetically accessible. Not just promising in silico, but actually makeable before a single one is sent to the labs." Traditional drug design often leans on decades of precedent and familiar chemistry. In Part 6 of our platform series, SVP of Drug Design Donovan Chin, Ph.D., explains how Recursion breaks that mold by getting a full computational design cycle up and running in about a day. Our AI-native engine vastly increases the aperture for our chemists. Instead of relying solely on known binding pockets, they can efficiently explore entirely novel and complex chemical space. By leveraging our integrated data, the team can forecast non-obvious design hypotheses and predict how to synthesize molecules even when nothing like them currently exists. This process begins by scoring binding affinity using advanced 3D structural models like Nesso-1. From there, the team runs hundreds of integrated models covering ADME, potency, advanced simulations, and AI retrosynthesis. This unified system rapidly filters millions of potential molecules down to a precise shortlist of high-quality candidates that are physically ready for the lab. 🚀 Follow along over the coming weeks as we continue this series, covering how Recursion maps biology, optimizes precision drug design, and closes the gap between patient and discovery. #TechBio #DrugDesign #ComputationalChemistry #DrugDiscovery #MachineLearning

"Across our programs, we've advanced quality novel candidates after synthesizing only around 300 compounds, roughly 90% fewer than the industry average." Now that we have mapped the biology, we need to design the exact molecule to target it. In Part 5 of our platform series, SVP of Drug Design Donovan Chin, Ph.D., and VP of Pharmacology Martin Redhead, Ph.D., break down how we transition from biological target discovery to navigating complex chemical space. Just like our approach to biology, we solve chemistry problems through a continuous feedback loop between computation and experimentation. Before a single molecule is ever synthesized in the lab, our team uses Centaur, our integrated drug design environment. By combining hundreds of machine learning and physics-based models into a single platform, we can iteratively design novel molecular structures starting directly from target properties. This generative de novo design approach automatically updates our models so they improve with every single cycle. Our AI-native product engine has already produced over 10 development candidates. By holding our chemistry to the same rigorous standards as our biology, we can synthesize a fraction of the traditional compound volume while delivering superior novel molecules. We believe this focus on quality and novelty over sheer volume will define the true impact of AI on drug development. 🚀 Follow along over the coming weeks as we continue this series, covering how Recursion maps biology, optimizes precision drug design, and closes the gap between patient and discovery. #TechBio #DrugDesign #AIDrugDiscovery #DrugDiscovery #Chemistry

To predict binding affinity, would you defer to lessons from physics or AI? No matter which option you choose, you’re unlikely to get where you’re looking to go. Check out Daniel Ramirez’s op-ed where he offers a third option, one that has the potential of going further than any one field alone. Link to op-ed: https://t.co/Fe0wl8ohav