Comparing molecular dynamics-derived pharmacophore models with docking: A study on CDK-2 inhibitors

Abstract

We compared the performance of molecular dynamics (MD)-derived pharmacophore modeling approaches, Common Hit Approach (CHA), and the Molecular dYnamics SHAred PharmacophorE (MYSHAPE) approach, with semi-flexible constrained/unconstrained docking. The aim of this work is to enrich the hit-list of a virtual screening on CDK-2 known inhibitors as a case study. Cyclin-dependent kinases (CDKs) deregulation is associated with cancer growth. CDKs are an attractive target for anticancer agents. MD-derived pharmacophore models have been obtained with LigandScout 4.2.1. Docking analysis has been performed through Glide 7.6. The results highlighted the MYSHAPE approach has a better performance when multiple target-ligand complexes are available (ROC5% = 0.99). Moreover, the use of short molecular dynamics simulations improves the performance of the screening (ROC5% = 0.98–0.99) with respect to docking (ROC5% = 0.89–0.94). Outcomes of this work indicate that these approaches are highly suitable for prospective screening by a non-expert, and could be useful for the identification of novel CDK-2 inhibitors.