Dominique Beaini
@dom_beaini
Solving Life with AI
Definitely a career highlight to be awarded a best paper award 🤩🤩 But even more exciting is that the 2nd and 3rd awards are given to colleagues Kian and Mohit, a podium sweep from @valence_ai / @RecursionPharma 🥇🥈🥉
(1/n)🚨You can train a model solving DFT for any geometry almost without training data!🚨 Introducing Self-Refining Training for Amortized Density Functional Theory — a variational framework for learning a DFT solver that predicts the ground-state solutions for different…
The supervision signal in AI4Science is so crisp that we can solve very complicated problems almost without any data or RL! In this project, we train a model to solve the Schrödinger equation for different molecular conformations using Density Functional Theory (DFT) In the…
(1/n)🚨You can train a model solving DFT for any geometry almost without training data!🚨 Introducing Self-Refining Training for Amortized Density Functional Theory — a variational framework for learning a DFT solver that predicts the ground-state solutions for different…
Very excited by this work! Simultaneously learn the DFT and sample the Boltzmann distribution by using two GPUs concurrently. Amazing stuff!!
(1/n)🚨You can train a model solving DFT for any geometry almost without training data!🚨 Introducing Self-Refining Training for Amortized Density Functional Theory — a variational framework for learning a DFT solver that predicts the ground-state solutions for different…
Data is what matters most!
1. We often observe power laws between loss and compute: loss = a * flops ^ b + c 2. Models are rapidly becoming more efficient, i.e. use less compute to reach the same loss But: which innovations actually change the exponent in the power law (b) vs change only the constant (a)?
What’s needed for a virtual cell to succeed in drug discovery? A new perspective paper from Recursion and our AI research engine @valence_ai lays out our vision for a virtual cell as a system that can reliably drive the discovery of new drugs via an iterative loop of: predict,…
one click - select a gene two clicks - select a cell three clicks - here's all the biological interactions you should see We're living in the future 🚀🦠
1/ Introducing TxPert: a new model that predicts transcriptional responses across diverse biological contexts It’s designed to generalize across unseen single-gene perturbations, novel combinations of gene perturbations, and even new cell types 🧵
1/ Introducing TxPert: a new model that predicts transcriptional responses across diverse biological contexts It’s designed to generalize across unseen single-gene perturbations, novel combinations of gene perturbations, and even new cell types 🧵
🤯 We can now directly interact with biology via this new foundational knowledge graph. A must read 🧵
1/ Introducing TxPert: a new model that predicts transcriptional responses across diverse biological contexts It’s designed to generalize across unseen single-gene perturbations, novel combinations of gene perturbations, and even new cell types 🧵
Everyone’s chasing AGI - but what’s the #DrugDiscovery equivalent? Meet the Virtual Cell: our bold roadmap to curing any disease by simulating the building block of human biology Check out @valence_ai vision of curing diseases with GPU-based scientific discovery
1/ At Valence Labs, @RecursionPharma's AI research engine, we’re focused on advancing drug discovery outcomes through cutting-edge computational methods Today, we're excited to share our vision for building virtual cells, guided by the predict-explain-discover framework 🧵
It was a pleasure to do the closing talk for the @gsp_workshop !
@gsp_workshop closing this years workshop with the very entertaining keynote on graphs in drug discovery by @dom_beaini !
Are you working on a @NeurIPSConf paper related to molecules? There's still chance to submit a short 1-page abstract for the MoML 2025 conference on June 18th in #Montreal portal.ml4dd.com/moml-2025-post…
(1/3) The poster submission deadline for MoML 2025 has been extended to May 20th, 2025. Don’t miss an opportunity to share your work at this years conference. Submit here: portal.ml4dd.com/moml-2025-post…
Thrilled to release Boltz-1x! The physical quality of poses was a recurring piece of feedback we received. With Boltz-1x, we use inference-time steering to remove clashes, respect chirality and other physical properties. A few other goodies too. Work led by the amazing @NoahBGetz
🚀 Excited to release a major update to the Boltz-1 model: Boltz-1x! Boltz-1x introduces inference-time steering for much higher physical quality, CUDA kernels for faster, more memory-efficient inference and training, and more! 🔥🧵
(1/3) Speaker Spotlight: Dr. Smita Krishnaswamy We’re thrilled to welcome Dr. Krishnaswamy to MoML 2025 on June 18th - hosted @Mila_Quebec Dr. Krishnaswamy is an Associate Professor of Genetics & Computer Science at @YaleMed
Join our social event!
(1/7) The team at Valence Labs—powered by @Recursion will be at @iclr_conf in Singapore this week, where we’ll be co-hosting a TechBio social on Friday, April 25th. Join us at the top of the Marina Bay Sands. RSVP: lu.ma/nts2d8uj See below for a summary of our papers👇
MoML conference is back on June 18th!
(1/5) The Molecular Machine Learning Conference (MoML) is back @Mila_Quebec this June 18th. Join researchers advancing machine learning, molecular modeling, and therapeutic design. Tickets are free for students. Register Today: portal.ml4dd.com/events/molecul…
These tarrifs are meant to prepare for war! @realDonaldTrump wants to expand the USA borders, but he cannot do so while steel, textile, oil, cars, etc. come from abroad. This is a well thought plan to bring resources + manufacturing to America so they can start to invade others
It’s now clear that the @realDonaldTrump Administration computed reciprocal tariffs without using tariff data. This is to economics what creationism is to biology, astrology is to astronomy, or RFK thought is to vaccine science. The Trump tariff policy makes little sense EVEN if…
🚨 Call for a POSTDOC in machine learning for drug discovery at @UMontreal 💊 The position will be part of a large collaboration with @BrunLabCaulo Audrey Durand @dom_beaini Anne Marinier @IRIC_umontreal @Mila_Quebec. All details here: brunlab.com/jobs/open-posi… Please RT 💜
We're currently looking for a PostDoc position at @UMontreal and @Mila_Quebec to work on a multi-modal bacteria representation for anti-biotic discovery, including multiple modalities: - Phenomics - Transcriptiomics - Structural bio - Viability curves - ADME The work will be…