Kalliokoski, Tuomo and Vulpetti, Anna (2011) Large-scale evaluation of CavBase for analyzing the polypharmacology of kinase inhibitors. Molecular Informatics, 30 (11/12). pp. 923-925. ISSN 1868-1751

Abstract

Binding site similarity analysis has been suggested to be useful in the prediction of selectivity profiles of kinase inhibitors. One of the few readily available methods for binding site analysis is CavBase, which is part of Relibase+ from Cambridge Crystallographic Data Centre. CavBase uses a simple scheme to encode the surface properties of all cavities in the protein structure to a set of descriptors called pseudocenters. Cavities are detected by a modified version of LigSite algorithm. Although CavBase has been available for some time, there are no published studies that investigate the impact of the various parameters of the method to the actual prediction accuracy. In this study, several different searching strategies of CavBase were tested on a kinase data set of 16 inhibitors. Different query definitions, as well as the accuracy of the all three scoring functions implemented in the software, were investigated. The data set was constructed by mapping the biological activity data from Ambit kinase panel to the structural information from the Protein Database (PDB). Data set consisted of 938 crystal structures for 131 kinases. An inhibitor was considered to be either active or inactive against a given kinase with threshold Kd of 10µM. Overall, CavBase demonstrated reasonable ranking accuracy with an average Receiver Operating Characteristic Area Under Curve (ROC AUC) value of 0.73. The default strategy of defining the query pocket by selecting pseudocenters using the distance of 4Å from inhibitor atoms and the use of original scoring function produced the best results. The study showed that it is possible to identify, on the basis of binding site similarities, those kinases that are likely to bind the same ligand, thus providing a useful tool for the rational design of targeted polypharmacology in kinase field. The low throughput of CavBase limits the applicability of the method for larger data sets (such as the whole PDB).

Item Type:	Article
Keywords:	CavBase, polypharmacology, kinase, inhibitor, binding site
Date Deposited:	13 Oct 2015 13:15
Last Modified:	13 Oct 2015 13:15
URI:	https://oak.novartis.com/id/eprint/5583