The moment has finally arrived. @RecursionPharma has combined with @exscientiaAI, giving rise to a new TechBio powerhouse! All the essential elements are now in place: an unwavering commitment to challenge the status quo; an interconnected end-to-end platform combining purpose-built multidimensional data w/ massive computing power, automated labs & precision chemistry capabilities; partnerships with leading pharma and a culture that breaks down the wall between biology and technology and encourages bold action. With this combination, we have 10 clinical and preclinical, 10 advanced discovery programs across rare disease, infectious disease, and oncology, and more than 10 partnered programs with leading pharmas, including @sanofi, @Roche/@genentech, @Bayer and Merck KGaA, Darmstadt, Germany. We’ve received approximately $450 million in upfront and milestone payments with the potential for over $20 billion in additional milestones before royalties. We built this in a little over a decade, and our technology is only improving as we increase proprietary data, further develop our foundation models, and accelerate our automation and unbiased discovery. This moment is both a culmination of all the work that has come before – and the beginning of an accelerated push to deliver on our promise to decode biology to radically improve lives. I’m incredibly grateful to the Recursionauts and Exscientians who have been part of this journey and to our counterparts in Oxford who bring such a complementary culture and approach that will enable us to lead, and significantly advance, the TechBio industry. Here’s to the next chapter of this exciting story!

A new article in @NatureBiotech looks at the rise of #foundation models in #drug discovery, noting that unlike others using an #LLM framework, “Recursion is pursuing a radically different approach with its Phenom family of foundation models.” As @ImranSHaque describes, our latest model can interpret cell response to treatment with new drug candidates & accurately interpret image formats that differ from those used in the training data. “His team is currently focused on extending Phenom to live-cell imaging data, as well as other cell types such as neurons.” nature.com/articles/s41587-0…

Announcing positive Phase 2 topline results for our first-in-disease treatment for cerebral cavernous malformation (#CCM) – the #AI drug discovery that launched our company. REC-994 met its primary endpoint of safety & tolerability & showed promising MRI-based secondary efficacy endpoints. We’ll meet with the FDA as soon as practical to discuss additional clinical studies. Learn more: ir.recursion.com/news-releas…

1/ 🚀Today, we’re excited to launch Polaris! 🚀 Polaris is a platform where the ML community working on drug discovery problems can easily share and access datasets and benchmarks. Get started with just a few lines of code: polarishub.io/

Today's benchmarking landscape for ML in drug discovery is complicated. We want to help the community develop methods that matter!🌟 Join us for our launch party at ICML on July 25th to learn more: lu.ma/wj1agv8o

“This is a once-in-a-lifetime opportunity.” In a Fireside Chat w/ @RecursionChris, Jensen Huang shared his vision for the future of healthcare – driven by #AI, #supercomputers, and a data-driven approach to #biology. #downloadday youtube.com/watch?v=Sr_n3gVe…

This blog post from NVIDIA talks about the expansion and our collaboration with NVIDIA: blogs.nvidia.com/blog/drug-d…

Our new supercomputer, BioHive-2, clocked in at #35 on the TOP500 list announced today. Check out the timelapse of what building a supercomputer in less than 3 weeks looks like. 😀

This work is the culmination of years of efforts in building powerful, expressive and scalable molecular GNNs. And we finally have it! A foundational GNN of 1B of parameters that achieves SOTA on a large diversity of tasks

People ✅ Data ✅ Compute ✅ Capital ✅ Pulled the last pieces of plastic off of BioHive-2 today as we finish up our Top500 benchmarking work. HUGE thanks to the dedicated team @RecursionPharma for the incredible push to get this across the line… and a big thank you to our partners @nvidia who pushed hard shoulder to shoulder with us… exciting stuff! Lots of models to be built on top of what we believe is one of the world’s largest biological and chemical datasets fit for the purpose of ML/AI.

Big news on the People front as well, with Najat Khan joining the Recursion exec team and Board of Directors! Najat is incredibly talented and interdisciplinary with her unique blend of expertise in life sciences, data science & business, she is the perfect addition to our team.

On the compute side of things, I went to help setup our new cluster today with @bertonearnshaw! For some reason they didn't trust us to rack the DGX nodes so we settled for some garbage duty and I read the DGX H100 Quickstart Guide... going to get these new nodes training soon!

Christmas has come *early*, yay for priority access! If you want to use these H100s to decode biology or help run the cluster with our amazing HPC team please apply! recursion.com//careers

You have to have skin in the game! Reflecting back on #JPM2024 last week and this shot has to be a personal highlight. I was comparing company tattoos with NVIDIA founder & CEO Jensen Huang when @RecursionPharma Board Chair @RMartinChavez dropped his full sleeve Turing machine! Stay tuned for a new reflection each morning this week!

At yesterday's standing-room-only #JPM24 talk by @nvidia, Kimberly Powell announced the release of our #phenomics foundation model, Phenom-Beta, on NVIDIA's #BioNeMo platform. Learn more about the power of Phenom-Beta in our latest blog by @bmabey: recursion.com/news/nothing-s…

Attending #NeurIPS2023? Don’t miss our spotlight poster, "Masked Autoencoders are Scalable Learners of Cellular Morphology" Paper: openreview.net/forum?id=jasx… Where: GenBio workshop, Poster Session 1 in Room 265 - 268 When: Saturday, Dec 16th, 11:30 am - 1:30 pm CST.

We published an article speaking a bit more about this model and the training of it... we intend to share more in paper in the coming months linkedin.com/pulse/does-imag…

While these reconstructions are visually impressive what is most exciting is the performance of these models on useful drug discovery tasks (e.g., see our Benchmarking work rxrx.ai/research#biological-… ) We’ll be sharing more about this model and the results soon! #showmethemodel 2/2

We have trained a phenomics ViT foundation model and the results are exciting. We’ve been using masked autoencoders as the pretraining task, and check out how well the models are able to reconstruct images! 1/2

We're at #ICML2023 all week and would love to meet up with folks interested in ML for drug discovery! Hit us up at our booth or one of the posters below 👋