Search

JAK2 has been considered as an important target for the development of anti-cancer agents. In this study, considering the flexibility of its binding site, an integrated strategy combining Bayesian categorization modeling and ensemble docking was established. Four representative crystal structures were selected for ensemble docking by the hierarchical clustering of 34 crystal structures according to the volume overlaps of each structure. A retrospective virtual screening was performed to validate this integrated strategy. As the preliminary filtration, the Bayesian model enhanced the ratio of actives by reducing the large amount of decoys. After docking the remaining compounds, the comparison between the ensemble and individual results showed that the enrichment of ensemble docking improved significantly. The results of analysis on conformational changes of two top ranked active inhibitors when docking into different proteins indicated that compounds with flexible conformations well fitted the different binding site shapes were more likely to be potential JAK2 inhibitors. This high efficient strategy will facilitate virtual screening for novel JAK2 inhibitors and could be even applied in drug discovery against other targets. [ABSTRACT FROM AUTHOR]