QSAR studies on imidazothienopyrazines as IKK-β inhibitors: from 2D to 3D

This paper presents the investigation of quantitative structure-activity relationship (QSAR) for imidazothienopyrazines from 2D to 3D, for assistance of developing new IKK-β inhibitors. In 2D-QSAR, we built linear and nonlinear models by heuristic method (HM) and Support Vector Machine (SVM) methods, respectively. The results (R2HM(training set) = 0.8775, R2SVM(training set) = 0.9356, R2HM(test set) = 0.6452, R2SVM(test set) = 0.8736) showed that the nonlinear model is more precise to predict the activities of imidazothienopyrazines. In 3D-QSAR, Phase methodology was employed, resulting in a successful model for prediction of imidazothienopyrazines' bioactivities, and this model (R2(training set) = 0.9717, R2(test set) = 0.9209) is better than any one of 2D-QSAR models, linear or nonlinear. In addition, the discussion of 2D descriptors from 2D-QSAR and the 3D plots from Phase results provided suggestions and guidance to develop new compounds which may present better bioactivities against IKK-β. Copyright © 2009 John Wiley & Sons, Ltd.