Kinase profiling is an efficient strategy for kinase inhibitor discovery, polypharmacological drug discovery and drug repositioning. A variety of methods have been established for kinase profiling, of which the combined mode of experimental and computational approaches is attractive, particularly in the early drug discovery. With the increasing number of kinase-inhibitor complex structures, structure-based kinase profiling is of great interest. To fully exploit the potential of structure-based kinase profiling in drug discovery and drug repositioning, we developed a versatile web server, termed ProfKin, for structure-based kinase profiling for small molecules of interest, which is established based on an in-house comprehensive structural database (KinLigDB) of manually curated kinase-ligand complex structures and associated information.

The kinase structural database KinLigDB currently contains 4,219 curated kinase-ligand complex structures of 297 human kinases (covering 106 kinase families); most of them are related to human diseases. About 75% of these kinases have at least two structures with different ligands, and 92 kinases have ≥10 complex structures. A lot of 396 binding sites were found and defined for the kinases in the database. Notably, 73 kinases have two or more different binding sites. The analyses of the binding modes identified eight types of ligands, including type I inhibitor (3167 entries), type II inhibitor (283 entries), type III inhibitor (23 entries), type IV inhibitor (9 entries), competitive inhibitor (554 entries), covalent inhibitor (31 entries), activator (49 entries), and allosteric ligand (103 entries). A total of 2805 kinase-ligand complexes were annotated with the experimental binding data.

The structure-based kinase profiling approach behind the ProfKin web server works via integrating molecular docking and interaction fingerprinting methods. It is thought that molecular docking usually enables the prediction of correct binding modes for small molecules but sometimes fails to prioritize the docking poses, partly due to the limitations of the used scoring functions. The interaction fingerprinting method is exactly complementary to molecular docking, which probably prioritizes the docking poses by comparing them with the reference kinase−ligand complex structures guided by the generated key interaction features. The integration of these two methods will probably yield improved prediction results.