Structure based virtual screening: Fast and slow

Abstract For many decades virtual screening methods have provided a convenient and cost effective in silico solution in the early stages of drug discovery. In particular, molecular docking uses structural information to approximate protein–ligand recognition, providing valuable information for large chemical libraries at a fast pace with multiple success stories to validate the approach. Nevertheless, fast turnaround of results required assumptions and approximations which compromise the accuracy of these algorithms. On the other side of the spectrum, physical‐based molecular simulations offer more precise and realistic models of protein–ligand binding at the cost of being slower and requiring more expensive computing infrastructure. Both fast and slow approaches are useful and solve different aspects of the same problem. Here, we aim to review these approaches focusing on their capabilities, context of usage and limitations, presenting multiple examples along the way. This article is categorized under: Molecular and Statistical Mechanics > Molecular Mechanics Software > Molecular Modeling Structure and Mechanism > Computational Biochemistry and Biophysics