QSAR study of selective HIV integrase inhibitors based on Boosting regression

摘要：

The QSAR study of 67 selective HIV integrase inhibitors is based on Boosting regression combined with MLR. Seven descriptors have been calculated by E—Dragon software as independent variables. The IC50 values of compounds are used as dependent variables. Seven models have been constructed by MLR and their variables are selected by particle swarm optimization. By testing the generalization of each model, the values of the internal and external validation of the RDF, Topological and 3D-MoRSE descriptor model are more than 0.5, so the Boosting-MLR model has been constructed by these descriptors. The best Boosting-MLR model of RDF descriptor improved the correlation coefficients (R2) prediction to 0.827, up from 0.794 when only MLR was applied. By the model validation, Boosting-MLR has shown the Stability generalization.

关键词： quantitative structure-activity relationship boosting regression HIV integrase inhibitors

作者: Changhong Zhou Yuntao Zhang Zhengjun Cheng

作者单位: Institute of Applied Chemistry, China West Normal University, Nanchong Sichuan 637002, China

会议类型: 国际会议

会议名称: The 2010 International Conference on Computer Application and System Modeling(2010计算机应用与系统建模国际会议 ICCASM 2010)

会议地点: 太原

会议语种:英文

页码: 251-255

在线出版日期: 2010-10-22（万方平台首次上网日期，不代表论文的发表时间）