Automatic selection of near-native protein-ligand conformations using a hierarchical clustering and volunteer computing