Today in our reading group: Multimodal language models for mapping the genotype-phenotype relationship

Join us at 11 am EDT / 5pm CEST: portal.valencelabs.com/logg

We are looking for a talented and motivated postdoc to work in a collaboration between Frank Noe and my group at Freie Universität Berlin on the development of machine-learned coarse-grained protein models. Please email [email protected] if interested.

How to GNN for your atom point cloud. I am quite impressed by this presentation: youtu.be/rUedGw3K9Uk?fe…

'Smooth, exact rotational symmetrization for deep learning on point clouds'

Tomorrow in the reading group we discuss 'Smooth, exact rotational symmetrization for deep learning on point clouds'! arxiv.org/abs/2305.19302

Join us at 11am EDT / 5pm CEST on zoom: portal.valencelabs.com/logg

There exists a paper discussion recording 'Stability-Aware Boltzmann Estimator (StABlE) Training of NN Interatomic Potentials | Sanjeev Raja'
youtube.com/watch?v=hCVXXd…

Oi, another summer edition of Molecular ML conference is happening in Montreal. Would be cool to have you there :)

You can submit posters and register here if you feel like it portal.ml4dd.com/submit-paper

The Molecular Machine Learning Conference (MoML) is back MilaQuebec this year on June 19th!

Come and present your work - we are now accepting one-page abstract submissions.

Deadline is April 29th. More details here: portal.ml4dd.com/moml-2024

Our method for molecular linker design with diffusion models (DiffLinker) got published in Nature Machine Intelligence 🥳

nature.com/articles/s4225…

Thanks a lot to our amazing coauthors Hannes Stärk Clément Vignac Arne Schneuing Víctor Garcia Satorras Pascal Frossard Max Welling Michael Bronstein Bruno Correia

Fun adventure of our lab in small molecule design rdcu.be/dErPa pleasure to work with a bunch of cool people that made it possible Ilia Igashov Hannes Stärk Arne Schneuing Clemence Vignac Víctor Garcia Satorras Pascal Frossard Max Welling Michael Bronstein

Is anyone interested in faster FID calculation for high-dim embeddings?

I first used github.com/mseitzer/pytor….
For 2k dim: 6 sec
For 16k dim: 1h 32min

Then use 'Newton-Schulz iteration' for matrix sqrt
For 16k dim: 16 sec (and lower sqrt error)

Let me know if anyone wants this

'Stability-Aware Training of Neural Network Interatomic Potentials with Differentiable Boltzmann Estimators' arxiv.org/abs/2402.13984

the paper we discuss tomorrow in the reading group with Sanjeev Raja.

On zoom, anyone can join, 11am EDT, link here: portal.valencelabs.com/logg

Cool paper to check out. i recommend at least glancing at fig 1 and some results

We wrote a little review on diffusion models for protein structure generation and docking.

Again, we tried to not just make it a list of papers, and instead to give some interesting framings and takes.

You judge if it worked :)

Thank you so much to 100+ of you who tuned it! It was a great pleasure to discuss this little field with all of you! 🤗

Slides are here: chaitjo.com/publication/du… (to be updated soon)

'A Hitchhiker's Guide to Geometric GNNs for 3D Atomic Systems' arxiv.org/abs/2312.07511

Tomorrow in the reading group with Chaitanya K. Joshi Simon Mathis and Alexandre Duval 🤗

Join us at 11am EDT / 5pm CEST on zoom: portal.valencelabs.com/logg

This is pretty neat.
MasiF with glycans and ligands.
Can screen for proteins interacting with protein-ligand complex.
And then use that to design a binder by screening for protein fragments and scaffolding them together.

WATCH: #JClinic faculty lead Regina Barzilay in NOVA | PBS's AI Revolution. AI Revolution is now streaming & premiering tonight @ 9pm ET on PBS 📽 Stick around to catch our fantastic collaborators at Mass General Cancer Center: Lecia Sequist, MD, MPH + Florian J. Fintelmann!
pbs.org/wgbh/nova/vide…

🥳🥳🥳 We are excited to share that AI for Science workshop will be held again with ICML Conference 2024, Vienna! This time, we focus on scaling in AI for Science (as a new dimension to theory, methodology and discovery)! Tentative schedules can be found: ai4sciencecommunity.github.io/icml24.html