In Silico Study of Polypharmacology with Ligand-based Interaction Fingerprint

Abstract

The past years have witnessed the versatile applications of interaction fingerprint method, including three-dimensional structure analysis, docking-pose clustering and filtering, scoring function improvement and enhancing enrichment of virtual screening. However, it's still unclear whether it's possible to study the polypharmacology with such a strategy. We have explored this important question by assessing the performance of ligand-based interaction fingerprint (LIFt), a new approach providing insights into the target profiles for the selected small drug. According to our results, it's found that LIFt could recognize most of the native targets for the promiscuous kinase inhibitor staurosporine on the basis of experimental determined complex structures. In addition, with assistance of physics-based docking and sampling techniques, LIFt can predict the kinase-selectivity profile as well as the unexpected off-targets for the clinical drug or experimental candidates with appreciated accuracy. More encouragingly, a prospective prediction of new target for the established synthetic anti-tumor drug TN-16 was experimentally validated, which suggests the promise of LIFt in practical use of polypharmacology study.