When beginning a virtual high-throughput screening (vHTS) campaign against a given target, scientists have to make several decisions with regard to the starting material they will use. In particular, they have to choose one or more starting structural models to use in docking. Furthermore, said models have to be altered such that the setup of the protein is suitable for the underlying software being used for vHTS.Consequently, this chapter is split into two main sections. The first tries to convey informationaboutthepitfallsinproteincrystalstructures,inordertomakeaninformed decision about structure selection. The second deals with the common issues one faces when trying to use one or more structures in the context of virtual screening.
Selecting a Protein Structure for Virtual ScreeningThis section focuses on how to select a structure for virtual screening. The discussion is specific to structures determined by X-ray crystallography. There are, of course, other structure elucidation methods (via NMR or via homology modeling). It should be noted, however, that structures determined by these methods are also subject to imprecision and inaccuracy. NMR yields multiple models, so the user has to select the one or more appropriate NMR conformations to use in vHTS. Use of homology models for docking-based vHTS should probably be done only as a last resort, as the side chain placement in the binding site in such models is likely to be less accurate than in the models generated using experimental information.
4.2.1Why Are There Errors in Crystal Structures?When presented with a structural file, ones initial temptation is to treat the contained coordinates as a direct 3D image of a protein structure; the reality, however, is that the Virtual Screening. Edited by C. Sotriffer