Martin Pacesa

Imagine teaching immune cells to precisely target diseased cells and ignore healthy ones. 🎯 Using deep learning, we’ve designed de novo proteins that can distinguish small differences on the surface of cells through MHC molecules. This work is out now in @ScienceMagazine! 🧵

Girl with Bacteria

Dear Cryo-EM Community, Are you looking for high-quality cryo-EM data with fast turnaround times and competitive pricing? The DCI-Lausanne Facility is here to support your structural biology research with cutting-edge instrumentation and expert guidance.

🚨 Check out DrugFlow, our new generative model for structure-based drug design. DrugFlow provides an atom-level confidence score for each designed molecule, and can adjust molecular size on the fly! Additional details in thread 🧵 #ICLR2025

New paper (and #ICLR2025 Oral :)): ProtComposer: Compositional Protein Structure Generation with 3D Ellipsoids arxiv.org/abs/2503.05025 Condition on your 3D layout (of ellipsoids) to generate proteins like this or to get better designability/diversity/novelty tradeoffs. 1/6

The code & camera-ready version of our #ICLR2025 paper on "Multi-domain Distribution Learning for De Novo Drug Design" are now available 📚 Paper: openreview.net/forum?id=g3VCI… 💻 Code: github.com/LPDI-EPFL/Drug… (1/4)

🚀#AlphaFold Database update AlphaFold DB now integrates The Encyclopedia of Domains (TED) – a resource designed to systematically identify & classify structural domains within AlphaFold-predicted protein structures. ebi.ac.uk/about/news/upd…

If training NN has taught us anything, it is that we should overparameterize. The 'lottery ticket hypothesis' applies here too. We cannot predict which PhD student holds the winning ticket. Cutting PhD programs would decrease chance of important future scientific breakthroughs.

RNA-guided systems span kingdoms, from bacterial immunity to eukaryotic gene regulation. How did they evolve? We traced the origins of CRISPR-Cas13 and identified its likely ancestor: an RNA-based toxin-antitoxin (TA) system. Here’s what we uncovered: cell.com/cell/fulltext/… 1/8

New #preprint from the lab! Do you want to see the invisible? So did we! When DNA genomes get broken, cells somehow find related sequences to fix the break. But how do they find it? We developed a way to look at sequence *search*, not just sequence usage. doi.org/10.1101/2025.0…

Artistic representation of the AuraA prenyltransferase tetramer, as solved by cryoEM (PDB code: 8Y9E) #scivis #molecularart @dzine_ai @proteinimaging behance.net/gallery/219042…

That makes sense. If you’re the CEO of the alphabet you should make sure to include all 26 letters

Excited to share our new AllMetal3D model building on Metal3D. AllMetal3D predicts the location, identity and geometry of the major biologically relevant metal ions doi.org/10.1101/2025.0…

Here’s the full, no-nonsense list of phrases that drive me nuts - because they’re misleading, fear-mongering, or just plain wrong: Natural. As if nature guarantees safety. Cyanide is natural too. Organic. A marketing label, not a health guarantee. Bio-Sourced. A fancy way to…

I am thrilled to say that our work is now online @CellCellPress ! I've analysed protein structures for quite some time, but @Robb_Mallik, who co-led this work, took these analyses to yet another level - 🧵👇 sciencedirect.com/science/articl…

This is very unnerving