Evaluation of AutoDock and AutoDock Vina on the CASF-2013 Benchmark

Computer-aided protein–ligand binding predictions are a valuable help in drug discovery. Protein–ligand docking programs generally consist of two main components: a scoring function and a search algorithm. It is of interest to evaluate the intrinsic performance of scoring functions, independently of conformational exploration, to understand their strengths and weaknesses and suggest improvements. The comparative assessment of scoring functions (CASF) provides such an evaluation. Here we add the AutoDock and Vina scoring functions to the CASF-2013 benchmark. We find that these popular, free software docking programs are generally in the first half (AutoDock) and first quarter (Vina) among all methods tested in CASF-2013. Vina is the best of all methods in terms of docking power. We also find that ligand minimization has an important impact, reducing the performance difference between AutoDock and Vina.