The accurate prediction of biomolecular recognition is fundamental to modern drug discovery.  Despite recent advances, the computational evaluation of the interactions between a protein and drug candidates is still problematic.  Specifically, challenges exist in both determining accurate binding affinities and reproducing experimentally determined binding poses.  Here, we present a novel algorithm, CrystalDock, which eschews traditional binding free-energy calculations, instead leveraging existing structural information to �dock� molecular fragments.  Given a binding pocket of interest, CrystalDock identifies localized regions of the pocket, called microenvironments, and mines a structural database of known ligand-fragment/receptor-microenvironment pairs for matches.  Subsequent structural alignment places these fragments into the pocket.   Positioned fragments can be combined to form novel compounds, or can be used to optimize known inhibitors. CrystalDock was first validated against previously solved protein-ligand complexes.  Computational binding free-energy calculations were then used to demonstrate that suggested ligand modifications improved potency of known inhibitors.

CrystalDock: Leveraging known structural information for fragment-based drug design

245th National Meeting (2013)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.