Gabriele Corso
@GabriCorso
PhD student @MIT • Research on Generative Models and Geometric Deep Learning for Biophysics • BA @CambridgeUni • Former @TwitterResearch, @DEShawGroup and @IBM
Excited to unveil Boltz-2, our new model capable not only of predicting structures but also binding affinities! Boltz-2 is the first AI model to approach the performance of FEP simulations while being more than 1000x faster! All open-sourced under MIT license! A thread… 🤗🚀
TT-Boltz is more than 1 minute faster again. The blue protein structure in the image was computed in 3 minutes and 20 seconds on a single Tenstorrent Wormhole processor. At Dev Day, it took 15 minutes, and the goal of 2 minutes seemed so far away. I was hoping for a speedup of…
Please run TT-Boltz on them. Please discover drugs on them.
There is a lot of satisfaction to seeing your baby out in the wild. Our Wormhole-based Galaxy is now powering some of the most exciting new cloud companies all over the world. @tenstorrent is about open source, open standards and multiple entry points to our software stack.
Wow, is Boltz-2 really that much faster than Boltz-1 and even ESMFold. Have to test it out soon. biorxiv.org/content/10.110…
📢📢 "La-Proteina: Atomistic Protein Generation via Partially Latent Flow Matching" Fully atomistic. Partially latent. Structurally precise. Entirely generative. w/ @tomasgeffner*, @DidiKieran*, et al. 📜 Project page & paper: research.nvidia.com/labs/genair/la… 🧵 Thread below... (1/n)
Truly useful autonomous chemistry agents are imaginable (but it's early) 🫣 AgentD turns GPT‑4o into a chemistry helper: 📑 pulls FASTAs ➜ 🧬 generates & refines SMILES ➜ 🧪 predicts 67 ADMET traits ➜ 🛠️ plugs into Boltz‑2 for 3D binding poses Two refinement loops lift…
🌽AlphaFold didn't "solve" protein structures for crops In 417 maize genes, orphan & domain-poor proteins drop confidence 25 - 43% AF2 leads geometry; Boltz-2 reaches AF3-like quality 60× faster; ESMFold is quickest yet clash-prone
We will be presenting our work on steering and scaling diffusion models at #ICML2025! Bonus: - FK steering can beat grad guidance?! 🤯 - Boltz 🧬adapted FK steering to BioML!!! x.com/GabriCorso/sta… East Ex. Hall A-B #E-1308 (Tue 15 Jul 11 a.m. PDT — 1:30 p.m. PDT)
Got a diffusion model? What if there were a way to: - Get SOTA text-to-image prompt fidelity, with no extra training! - Steer continuous and discrete (e.g. text) diffusions - Beat larger models using less compute - Outperform fine-tuning - And keep your stats friends happy !?
🧬 Boltz-2 is the new open source AI model from @MIT and @RecursionPharma. ⚡Did you know it's capable of predicting protein binding affinity with unprecedented speed, scale, and accuracy? #NVIDIABioNeMo #Boltz2 #DrugDiscovery
Can AI-predicted complexes teach machine learning to compute drug binding affinity? A new study demonstrates that machine learning based scoring functions for drug binding affinity can be effectively trained using AI predicted protein-ligand complex structures, provided these…
Human protein interactome structure prediction at scale with Boltz-2 biorxiv.org/content/10.110… #biorxiv_bioinfo
Human protein interactome structure prediction at scale with Boltz-2 １．This study presents large-scale structure predictions of 1,394 human binary protein-protein interactions (PPIs) using Boltz-2, an AI/ML model that has shown strong performance across biomolecular complex…
🚀 Open AlphaFold3-like model Boltz‑2 was just used to map 1,394 human protein‑protein complexes ⏩ from IntAct pairs to 3D models in a single sweep What’s the next scale up of this?
Excited to share our work on modeling the human protein interactome in 3D. This currently includes 1,394 interaction structures predicted with Boltz-2. Grateful for the open-access resources making this possible! Preprint: doi.org/10.1101/2025.0…
Excited to unveil Boltz-2, our new model capable not only of predicting structures but also binding affinities! Boltz-2 is the first AI model to approach the performance of FEP simulations while being more than 1000x faster! All open-sourced under MIT license! A thread… 🤗🚀
Boltz チームから、モデル改善のため低分子デザインの新規テーマの募集。何か標的分子があってアッセイ系を持っているようであれば、結合する低分子を Boltz チームでデザインするので活性評価してフィードバックしてくださいとのこと。
📢 Call for proposals: Boltz small-molecule design collaboration! 🧬 Can we help design your ideal molecule? Can you help us improve our open-source models? Please reach out or share with scientists you know! More details below! It has been great to see the level of excitement…
📢 Call for proposals: Boltz small-molecule design collaboration! 🧬 Can we help design your ideal molecule? Can you help us improve our open-source models? Please reach out or share with scientists you know! More details below! It has been great to see the level of excitement…
Boltz-2 now runs on Tenstorrent boards and we saved another 20 seconds. Clone TT-Boltz from GitHub and try it out. This is a pretty big deal, since Boltz-2 can predict binding affinity which is essential for drug discovery. A future where Tenstorrent computers become the best…
Excited to share our new ProxelGen paper! Completely different from RFDiffusion etc., we generate proteins as densities instead of point clouds. Turns out this works just as well and e.g. does better on some scaffolding tasks. arxiv.org/abs/2506.19820 (1/8)