Vincent Mallet

Adding Boltz2 distogram edges and features to protein and RNA graph networks beats MD-correlation graphs on binding tasks and improves stability, interaction and RNA predictions without simulation.

RNAmigos2 improves RNA virtual screening by combining semi-supervised graph learning and large-scale docking data, achieving high accuracy and speed.

This paper introduces CrAI, a fully automatic method for detecting antibodies in cryo-EM densities, achieving faster and more accurate results than existing methods.

InDeep predicts functional binding sites in proteins for drug design on PPIs, outperforming existing tools on both PPI and conventional targets.

The paper introduces new RC-equivariant architectures for DNA sequence analysis, showing they outperform existing models.

OptiMol optimizes molecular binding affinities using a hybrid structure- and ligand-based approach, improving docking scores and compound generation efficiency.

RNAmigos predicts small molecule ligands for RNA binding sites using augmented base pairing networks, showing improved performance over baselines and highlighting the importance of non-canonical base pairs.