Molecular surface point environments for virtual screening and the elucidation of binding patterns (MOLPRINT)

A novel method (MOLPRINT) for virtual screening and the elucidation of ligand-receptor binding patterns is introduced which is based on environments of points on the molecular surface. In combination with the Tanimoto coefficient and applied to virtual screening, it achieves retrieval rates which are comparable to 2D fingerprints. In combination with information-gain based feature selection and a naive Bayesian classifier, information from multiple molecules can be combined and classification performance can be improved. The descriptor uses points relative to the coordinates of the molecule which are uniformly binned, thus it is translationally and rotationally invariant. Due to its local nature the descriptor is conformationally tolerant. The identification of active structures with minimal 2D similarity is facilitated, commonly referred to as "scaffold hopping". Features which are selected by the information-gain based feature selection step can be projected back on the molecular surface. They are shown to be consistent with experimentally determined binding patterns.