Hit clustering can improve virtual fragment screening: CDK2 and PARP1 case studies

摘要: Virtual fragment screening could be a promising alternative to existing experimental techniques. However, reliable methods of in silico are yet established and validated. In order develop such an approach we first checked how successful the molecular docking can predicting binding affinities poses. Using our Lead Finder software RMSD energy prediction was observed 1.35 kcal/mol−1 on set 26 experimentally characterized inhibitors, predicted pose from one <1.5 A. Then, explored 68 fragments obtained 39 drug molecules for which co-crystal structures were available PDB. It appeared that participate oriented non-covalent interactions, as hydrogen bonds metal coordination, correctly docked 70-80% cases suggesting potential success rediscovering corresponding drugs by approach. Based these findings we’ve developed virtual technique involved structural filtration protein-ligand complexes specific interactions subsequent clustering minimize number preferable starting candidates. Application this method led 2 millimolar-scale PARP1 inhibitors with new scaffold.