Recursion

November 20 Update Call

Planning to attend the Superbowl of machine learning conferences? NeurIPS is coming Dec. 10-15 to the Vancouver Convention Center and researchers from Valence Labs powered by Recursion will be on site presenting over 10 accepted papers and hosting a “Night of Science” TechBio social with NVIDIA on Thurs., Dec. 12. ?? RSVP for the Night of Science: https://lu.ma/biikt7ox ?? ?? Check out our NeurIPS papers: ?? “ET-Flow: Equivariant Flow Matching for Molecular Conformer Generation” East Exhibit Hall A-C #2509 Dec. 11, 11am - 2pm PST https://lnkd.in/gySDH2JB Nikhil Shenoy Dominique Beaini ?? “How Molecules Impact Cells: Unlocking Contrastive PhenoMolecular Retrieval” East Exhibit Hall A-C #1110 Dec 11, 11am - 2pm PST https://lnkd.in/eWpuH9aq Philip Fradkin Dominique Beaini ?? “Propensity Score Alignment of Unpaired Multimodal Data” East Exhibit Hall A-C #1000 Dec 12, 11am - 2pm PST https://lnkd.in/g72FeTgH Jason Hartford ?? “Targeted Sequential Indirect Experiment Design” West Ballroom A-D #7106 Dec 12, 11am - 2pm PST https://lnkd.in/geU5Su57 Jason Hartford ?? "QGFN: Controllable Greediness with Action Values” West Ballroom A-D #6402 Dec 12, 11am - 2pm PST https://lnkd.in/gVVahKZc Emmanuel Bengio ?? “Amortizing intractable inference in diffusion models for vision, language, and control” West Ballroom A-D #7101 Dec 12, 4:30 - 7:30pm PT https://lnkd.in/gkizXqn9 Emmanuel Bengio ?? “On the Scalability of GNNs for Molecular Graphs” East Exhibit Hall A-C #3103 Dec 13, 11am - 2pm PST https://lnkd.in/g54cDfnz Dominique Beaini ?? “SAFE setup for generative molecular design” AI4Mat workshop in West Meeting Room 211-214 Dec 14, 8:15am PST https://lnkd.in/g8KSmv3S Emmanuel Noutahi, PhD ?? “Graph Classification Gaussian Processes via Hodgelet Spectral Features” Bayesian Decision-Making and Uncertainty workshop Dec 14, 8:15am PST https://lnkd.in/gRVbkqjU Mathieu Alain Emmanuel Noutahi, PhD ?? "Benchmarking Transcriptomics Foundation Models for Perturbation Analysis" AIDrugX workshop in West Meeting Room 109, 110 Dec 15, 8:15am PST https://lnkd.in/geu_mj5e Emmanuel Noutahi, PhD Alisandra Denton ?? “Score-Based Interaction Testing in Pairwise Experiments” Causal Representation Learning workshop in East Exhibition Hall C Dec 15, 8:15am PST Workshop: https://lnkd.in/g9tv6rga Jason Hartford ?? “ViTally Consistent: Scaling Biological Representation Learning for Cell Microscopy” Foundation Models for Science workshop in West Meeting Room 202-204 Dec 15, 8:15am PST https://lnkd.in/eb7wFmMz ?? “Towards Scientific Discovery with Dictionary Learning: Extracting Biological Concepts from Microscopy Foundation Models” Interpretable AI workshop in East Ballroom A, B Dec 15, 8:15am PST #NeurIPS #TechBio

?? Announcing OpenQDC: the largest public quantum mechanical dataset. ?? ML models using quantum mechanical data can provide critical insights for drug and material discovery -- expanding our understanding of protein folding, allosteric sites, and binding mechanisms, among other molecular behaviors. But to achieve accuracy, researchers need high quality QM datasets – and these are not easily accessible. ?? Now, researchers at Valence Labs powered by Recursion have released OpenQDC – Open Quantum Data Commons – the largest collection of publicly available QM data designed for Machine Learning Interatomic Potential (MLIP) algorithm development and training, democratizing access. ?? OpenQDC aims to unify and standardize existing, well-known datasets to advance the future of MLIP research. Researchers collected publicly-available datasets and computed essential missing metadata necessary for accurate data processing –? including energy, distance, force units, and isolated atom energies. ?? The 40 QM datasets cover 1.5 billion geometries across 70 atom species from over 250 quantum methods all consolidated into a single, accessible resource. ?? The data is meticulously preprocessed and standardized for MLIP training, covering a wide range of chemical elements and interactions relevant in organic chemistry. ?? OpenQDC allows researchers to utilize multiple datasets in previously impossible ways to advance MLIP research – accelerating discovery in molecular dynamics, and helping to drive novel algorithm development. ?? Learn more and access here: https://www.openqdc.io/ #quantum #ml #mlip #data

A look ahead at our first-in-class and best-in-class programs. In a new story in Genetic Engineering & Biotechnology News, Alex Philippidis looks at the recent combination deal and how it positions us as a "trans-Atlantic AI powerhouse" with "a pipeline of more than 10 clinical and preclinical programs, 10 programs in advanced discovery phases, and more than 10 additional programs partnered with biopharmas." ?? In oncology, it includes REC-617, a CDK7 inhibitor being developed to treat advanced solid tumors. We'll report Phase I monotherapy safety and pharmacokinetic and pharmacodynamic (PK/PD) data for REC-617 on Dec. 9 during the American Association for Cancer Research special conference in Toronto, and hold a related webinar on Dec. 10. ?? In rare disease, it includes REC-994 for symptomatic cerebral cavernous malformation (CCM). REC-994 had encouraging Phase II readouts in September and we'll share related data in the first half of 2025 in a conference and scientific publication. ?? Najat Khan, PhD, Chief R&D and Chief Commercial Officer says: "We really went through a thoughtful discipline process around portfolio prioritization, strategically. Where do we think we have the highest percentage of opportunities to win? And that’s where some of the decisions have been made.” ?? Read more: https://lnkd.in/e_BsiScq #TechBio #pipeline #drugdiscovery #pharma #patients

"We saw an opportunity to take the two leaders of this space, put them together and really catapult ahead of the rest of the industry." - Chris Gibson Yesterday, cofounder and CEO Chris Gibson appeared on Bloomberg Television's "The Close" to discuss the combination deal with Exscientia and why Recursion is positioned to lead the TechBio space. ? On our moat: "The real moat is the data. We generated or partnered on more than 50 petabytes of data...We’ve been making this investment in this build phase for a decade... No matter how much money you have to put into the AI space, data takes time. Cells take time to grow [and] we’re well ahead of anybody else." ? On transforming the industry: "I think there’s going to be an Amazon or an NVIDIA of the biopharma industry, and we intend for it to be us." Full episode here: https://lnkd.in/e3DyABvQ #techbio #ai #ml #pharma #drugdiscovery

Announcing our deal close on the big screen -- Nasdaq MarketSite in Times Square! #RXRX

Join Chris Gibson, Najat Khan, PhD, and Dave H. live for our Update Call today, 11/20 at: ? 5:30am MT? ? 7:30am ET? ? 12:30pm GMT?? Submit your questions at https://bit.ly/4fssA7I #techbio #biotech Exscientia

The future of TechBio has arrived. Today we announce that our business combination with Exscientia is officially closed – and two world-class, clinical stage TechBio companies have become one Recursion. We’re entering a new stage, for our company and for the TechBio industry. This combination allows us to move with even greater speed, precision, and efficiency to leverage AI, automation, and scientific expertise in order to develop new treatments for patients in need. ?? Highlights: ? Pipeline: Our combined pipeline has 10 clinical and preclinical, and 10 advanced discovery programs across rare disease, infectious disease, and oncology -- an unprecedented scale achievable only through the speed and efficiency brought by our combined AI-enabled platforms. ? Partnerships: We have more than 10 partnered programs with some of the largest pharma companies in the world – including Sanofi, Roche and Genentech, Bayer, and Merck KGaA, Darmstadt, Germany – and have received approximately $450 million in upfront and milestone payments. These partnerships could yield over $20 billion in additional milestones before royalties. In our partnership with Sanofi, focused on creating best-in-class therapies in inflammation and oncology, milestones have been achieved for three programs already. In our oncology partnership with Bayer, we have 2 joint projects rapidly advancing to Lead Series nomination, and another 20+ multimodal data packages in development.? ? Platform: At Recursion, we’ve been relentlessly focused on innovating early-stage drug discovery, driven by 60+ petabytes of proprietary data, millions of automated wet lab experiments per week, and machine learning models to develop Maps of Biology capable of identifying novel targets and developing promising new drug candidates. Now, with the combination, we’ve added precision chemistry capabilities driven by AI and automation to design and test highly optimized potentially first-in-class molecules for high-interest targets. ?? How We’re Improving Pharma Averages: ? 3X Speed: Traditional pharma takes 30 months to develop a validated candidate; we’re doing it in as little as 10 months. ? 10X Efficiency: Traditional pharma produces 2,500 molecules before they find a compound for testing; we’re producing just 250 on average. ? Up to 80% Cost Reduction: Traditional pharma costs $25-$35 million to reach an investigational new drug (IND); we’re getting there for $5-$10 million. ??Learn more: https://lnkd.in/e7K5_qrQ #ai #techbio #combination #partners

It’s Tea Time at LONDON BIO. Taking place alongside the Jeffries Healthcare Conference, the Afternoon Tea on Tues., Nov. 19 will feature Chief R&D and Chief Commercial Officer Najat Khan, PhD, with Bharatt Chowrira, CEO and Board member at PureTech Health, and Sutha Satkunarajah, Senior Director of Flagship Pioneering UK aboard the historic HMS Wellington, discussing the latest in pharma and TechBio. The discussion will be moderated by Yasmin Siraj, investor at BACKED VC. BACKED VC and Jack O'Meara co-founded LONDON BIO over a year ago to bring together a diverse set of biotech and TechBio practitioners to share their personal journey and war stories in discovering, developing and scaling category-defining therapeutics. This event is currently oversubscribed. But if you'd like to sign up to be considered for the next one - you can do so here: https://lnkd.in/ewGgfnax #FutureofHealthcare #TechBio #LondonBio

Two recent posts in BioPharmaTrend.com highlight our publicly available foundation model, OpenPhenom-S/16, just launched on Google Cloud's Model Garden. 1. Andrii Buvailo, Ph.D. describes how OpenPhenom-S/16, built on microscopy data, has set a new "gold standard" for the industry, outperforming CellProfiler, as part of a story on 19 companies pioneering AI models in pharma and biotech: https://lnkd.in/eVRamQeD 2. Roman Kasianov shares additional insights -- including how the model was trained using self-supervised learning on over 3 million microscopy images from datasets like RxRx3 and JUMP-CP and can "replace traditional microscopy analysis tools that require significant tuning and struggle with large-scale datasets." https://lnkd.in/eVNKA2B7