Similarity versus docking in 3D virtual screening

The development of similarity methods for fast flexible ligand superpositio n has recently received considerable attention. These efforts have brought similarity methods to a level of performance comparable to the well establi shed protein-ligand docking methods for binding mode assessment and molecul ar database screening. However, the strengths and intrinsic limitations of both methodologies have been also stressed out extensively. As the number o f resolved ligand-bound protein structures increases, combining ligand-base d and receptor-based approaches emerges as a consensus strategy to maximall y exploit the structural information available and improve the results obta ined with either of the methods alone.