Processing of small molecule databases for automated docking

Virtual screening involves the mining of small molecule databases from various sources. The small molecule databases used in virtual screening are typically processed, from simple 2D representations, to maximize their information content and to optimize them for input to the particular virtual screening technology being used. Processing interprets or adds molecular information related to connectivity, stereochemistry, protonation, tautomers and conformation. For virtual screening with an automated docking protocol, a technique that relies on specific intermolecular atom-atom contacts for ranking molecules, it is expected that the pre-processing protocol can affect the results of the docking experiment. The possible effects of processing on docking results have not been extensively studied, and this topic has only recently emerged as a significant aspect of the docking-based virtual screening process. One recent report highlights significant effects of different processing procedures on docking enrichment, while another outlines a general ligand preparation strategy. Here we survey and comment on recent practice in the field.