Structure-based drug discovery (SBDD) has become an indispensable technique for hit identification at the early stage of drug discovery. However, the accuracy of current scoring functions is not high enough to confer success to every targets thus remains to be improved. Precedently, we developed binary pose filter (PF) using knowledge derived from the protein-ligand interface of single X-ray structure of specific target. This novel approach had been validated as an effective way to improve ligand enrichment. Continuing from it, we attempt to incorporate protein flexibility from multiple crystal structures of the same target to build PF ensemble (PFE). To this end, we constructed a comprehensive data set to meet the requirement of ensemble modeling and validation, which contains 10 diverse targets, 118 well-prepared X-ray structures of protein-ligand complexes and large benchmarking actives/decoys sets. Notably, we designed our unique workflow of two-layer classifiers based on the concept of ensemble modeling and applied it to building PFEs for all targets. Via extensive benchmarking studies, we demonstrated that (1) PFE coupled with Chemgauss4 improves significantly the early enrichment of Chemgauss4 alone; (2) It shows greater consistency in boosting early enrichment and displays larger overall enrichment than the prior PF. In addition, we analyzed pairwise topological similarity among cognate ligands used to construct PFE and found that it is the higher chemical diversity enriched led to its improved performance. Taken together, the current outcome indicates that the incorporation of protein flexibility via ensemble modeling is able to enhance the screening competence of structure-based scoring functions.

Ensemble machine learning to improve scoring functions

254th National Meeting (2017)

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