A pharmacophore hypothesis can be created by recognizing key protein-ligand interactions in a co-crystal structure or homology model without relying on inference as in ligand-only methods. While such hypotheses have been successfully applied to virtual screening campaigns, protein-ligand complexes are flexible under physiological conditions. Hypotheses derived from a static view of the complex may not reflect the ensemble of key protein-ligand interactions and their geometries. We aimed to incorporate the effect of conformational flexibility on pharmacophore hypotheses perceived from an ensemble of a protein-ligand complex generated with molecular dynamics (MD) simulations. To do so, we define a feature vector microstate (FVM) as a cluster of pharmacophores from the conformational ensemble of a protein-ligand complex bearing the same features but different geometries. We improved on existing approaches as follows: (1) extract representative geometries of features in an FVM; (2) compute weights to relate the relative importance of FVMs; (3) use the representative pharmacophores and the weight for different FVMs to identify active compounds (experimental affinity &lt;10 μM) while reducing the matching of inactive compounds (experimental affinity &gt;10 μM) in a virtual screen. As a benchmark, we used forty diverse targets along with curated active and decoy compounds from the DUD-e dataset. To assess how our approach improves outcomes of virtual screens, we compared to existing approaches in Phase and LigandScout. The results of this work demonstrate that our MD-enabled approach can improve enrichment of virtual screens, as well as enable the discovery of informative pharmacophores by incorporating conformational flexibility of a protein-ligand complex.

Improving virtual screens with a new pharmacophore perception-scoring tool integrating MD of protein-ligand complexes

255th National Meeting (2018)

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