Sorbonne University

Protein-protein interaction prediction with partner-specific interface localization from sequence. Screens one million pairs in under two hours.

End-to-end protein structure alignment from 3D coordinates, matching TM-align quality at LDDT 0.56 and beating Foldseek on SCOPe40 search.

Protein conformational motion prediction from a single structure, using an SE(3)-equivariant GNN trained on ensembles mined from the PDB.

Retinal encoding models that predict ganglion cell responses to visual stimuli, with pretrained checkpoints across four species and two modalities.

Predicts intrinsic and soft disorder per residue using LoRA adapters on frozen protein language models, released with the SoftDis database.

Liquid-liquid phase separation predictor that scores proteins and residues from sequence and designs phase-separating peptides by gradient descent.

Protein motion prediction from sequence alone, mapping language model embeddings to continuous 3D displacement vectors with a lightweight CNN.