A Data Science Approach to Understanding Water Networks Around Biomolecules: The Case of Tri-Alanine in Liquid Water

Herein, we use recently developed data science algorithms to illustrate the complexity of the water network surrounding the hydrated peptide tri-alanine extracted from molecular dynamics simulations. We estimate the dimensionality of water variables and show that it is sensitive to the underlying secondary structure of the peptide. We show that water wires threading the peptide encode important information on the secondary structure. Interestingly, the free-energy landscape as revealed by the water wires is very rough for alpha-configurations and rather smooth for beta-configurations. The structured nature of the free-energy landscape is washed out if one uses more standard collective variables such as the number of hydrogen bonds around the peptide. Our results provide fresh insights into the molecular ingredients behind the coupling of protein and solvent degrees of freedom relevant for many biophysical and chemical processes.