Prediction of the hydration structures at the protein–protein Interface of dimers using deep learning

Hydration is crucial in various biological processes such as protein folding, ligand binding, and protein–protein interactions. Clarifying the role of hydration will improve our understanding of biological processes. Previously, we proposed a deep-learning model “gr Predictor” that successfully reproduced hydration structures obtained using the three-dimensional reference interaction site model theory. Because the gr Predictor was trained using a set of only single-chain proteins, in this study, we investigated whether it could predict the hydration structures at the protein–protein interfaces of dimers. This prediction was found to be successful. These results expand the applicability of the gr Predictor. • Hydration structure (HS) at the protein–protein interface (PPI) was investigated. • Our deep-learning (DL) model that reproduced HS of single-chain proteins was used. • Our DL model successfully predicted the HSs of the PPIs of dimers