A statistical model of overlapping volume in ligand binding cavities

Understanding and predicting protein-ligand binding preferences is an essential aspect of research in many fields, especially drug design. To assist in this effort, this paper presents VASP-I (Volumetric Analysis of Surface Properties for Intersections), a statistical model for estimating the probability that a set of cavities exhibit the same conserved region, and may thus have the same binding preferences. We applied this method to analyze ligand binding cavities of sequentially nonredundant structural representatives of the serine protease and enolase superfamilies. On these datasets VASP-I correctly distinguished sets of cavities with identical binding preferences from other sets with varying binding preferences. These results indicate that it can be possible to predict binding cavities that exhibit different binding preferences, even when the biochemical mechanism is unknown.