Daniele Granata

Reward-guided inference-time search over the Proteína-Complexa sequence-structure codesign model yields validated binders across 127 targets, kinases and a free carbohydrate, with codesigned sequences outperforming ProteinMPNN redesign.

This paper predicts antibody non-specificity using protein language models and biophysical parameters, showing that heavy variable domains and isoelectric point are key predictors.

The paper proposes a molecular mechanism for TRPV1 activation through conformational changes and validates it with bioinformatics, structural analysis, and experimental data.

The paper introduces a co-evolutionary coupling analysis to identify structural and functional domains from protein sequences, showing that sequence-based methods can recover domain characteristics.