Three-dimensional shape-based searching of conformationally flexible compounds

Steric shape plays a crucial role in receptor-ligand binding. This paper describes a three-phase database searching strategy for rapidly finding compounds similar in shape to a given shape query. The query is constructed from a receptor surface model which is derived from one or more structures. The first phase of the search screens the database, using shape-based indices as screens, for compounds which are potentially similar in shape. The second phase aligns the candidate structures and evaluates shape similarity based upon a grid-volume comparison and optional volume overlay, optimization. The third phase flexibly fits candidate structures to the surface model and evaluates steric strain and electrostatic complementary. Conformational flexibility is first addressed by representing compounds in the shape database as a set of diverse conformers and later by the flexible fitting process.