ISPRED-SEQ: Deep neural networks and embeddings for predicting interaction sites in protein sequences

Abstract

AbstractThe knowledge of protein-protein interaction sites (PPIs) is crucial for protein functional annotation. Here we address the problem focusing on the prediction of putative PPIs having as input protein sequences. The problem is important given the huge volume of sequences compared to experimental and/or computed protein structures. Taking advantage of recently developed protein language models and Deep Neural networks here we describe ISPRED-SEQ, which overpasses state-of-the-art predictors addressing the same problem. ISPRED-SEQ is freely available for testing athttps://ispredws.biocomp.unibo.it.