Interesting, approachable, and well written!

The first paper citing Egret-1 was released recently, less than a month after we released our model... Sanketh Vedula, Alex Bronstein, + co-workers use Egret-1 embeddings to predict local protein properties like NMR shift + secondary structure.

Tune in today at noon ET!

Really cool work by Meital and Sanketh Vedula, excited to highlight this. Check out their blog post and paper here!

me in pursuit of favorable hyperparameters: I know, O LORD, that the way of man is not in himself, that it is not in man who walks to direct his steps. Correct me, O LORD, but in justice; not in your anger, lest you bring me to nothing. - Jer 10: 23-24 (ESV)

The long-anticipated g-xTB paper was just released on ChemRxiv. g-xTB is the next semiempirical method from Grimme and co-workers at Rheinische Friedrich-Wilhelms-Universität Bonn. (I've heard rumors about this work for almost two years.) Here's some quick thoughts upon an initial read:

Our new ChemRxiv preprint: AIMNet2-rxn: A #MachineLearning Potential for Generalized Reaction Modeling on a Millions-of-Pathways Scale! #compchem Collaboration with Brett Savoie group. Funded by Office of Naval Research (ONR) under #MURI program chemrxiv.org/engage/chemrxi…

update, they released the code and it's now live on Rowan! you can now run g-xTB optimizations etc on Rowan. (it's a pre-release, so gradient and Hessian calculations are numerical and a bit slow.) thanks to Ari Wagen and the rest of the team for yet another same-day sprint!

Chai Discovery (Joshua Meier Chai Discovery) just released Chai-2, a model for "zero-shot antibody design in a 24-well plate." The most existing result? Chai-2 successfully generated binders for 50% of the 52 targets they assayed (100x better than SOTA)! Some thoughts below:

you nvidia-smi post to flex your 5090 I’m downgrading my local workstation to sm75 to triage a Triton bug we are not the same

🚀Exciting news! We are releasing new UMA-1.1 models (Small and Medium) today and the UMA paper is now on arxiv! UMA represents a step-change in what’s possible with a single machine learning interatomic potential (short overview in the post below). The goal was to make a model

Few know this, but Rowan hosts a free 2D/3D structure viewer and editor at labs.rowansci.com/editor. You can go from SMILES to .xyz, edit structures, change dihedrals, add a periodic cell, and download the final structure—all totally free, no account needed. Here's a quick demo:

hydrogen-bond-donor strength workflow now available on rowan! nice work Corin Wagen

Great to see Rowan's redox potential and Fukui-index calculations used to help with total synthesis. Congrats Matt Carson Kozlowski Lab and co-workers!

New low-cost computational methods like NNPs and g-xTB run orders of magnitude faster than DFT and often give comparable accuracy. But can they accurately predict of protein–ligand interaction energies? To find out, we tested 11 new methods on the PLA15 benchmark.

Rowan is making molecular modeling actually usable, so I built an MCP server that lets you access their tools through natural conversation.

Boltz v2.2.0 released! Boltz-2 now includes full support for arbitrary contact conditioning, template forcing potentials, and several improvements contributed by members of the community 🤗

We often think about potential energy surfaces in 2D, but they're really 3D. Our summer intern, Isaiah Sippel, used Manim to create a short 3D animation of a Diels–Alder PES. The video illustrates the energy barrier between reactants and products, along with the connecting MEP: