Ivan Valer'evich Oseledets

This study benchmarks 11 protein-ligand interaction prediction tools, showing classical and deep learning methods perform similarly on drug-like ligands, while co-folding tools excel on complex ligands and all struggle with generalization to unseen pockets.

The paper introduces MULAN, a multimodal protein language model that integrates sequence and structure information, showing improved performance on downstream tasks compared to sequence-only and structure-aware models.