Raúl Fernández-Díaz

This study shows that chemical foundation models can capture glycobiology principles, including immunogenicity and protein-glycan interactions, by fine-tuning atom-level representations from general chemistry.

BioMetAll v2.0 improves metal-binding site prediction with metal-specific scores and backbone geometry analysis, enhancing benchmarking and understanding of metal-protein interactions.

The paper examines how proteolysis in seeds is influenced by germination, homogenisation, and species mixture, finding increased proteolytic activity and peptide release.

HemeFinder predicts heme-binding sites in proteins with high accuracy and speed, demonstrating its utility in identifying heme interactions and pathways.

This paper develops AutoPeptideML, an AutoML tool for peptide bioactivity prediction, showing that automated methods can improve model trustworthiness and efficiency.

The paper introduces a new metric, AU-GOOD, to evaluate model performance on out-of-distribution data in biochemistry, with applications to small molecule and protein property prediction.