论文部分内容阅读
采用三维全息原子场作用矢量(3D-HoVAIF)研究89个四氢咪唑苯二氮卓酮(TIBO)类抗艾滋病药物的定量构效关系(QSAR)。分别运用偏最小二乘回归、人工神经网络建模,同时采用内部及外部双重验证的办法深入分析和检验模型的稳定性。PLS与ANN建模的复相关系数(R_(cum)~2)、留一法(leave-one-out,LOO)交互校验(cross-validation,CV)复相关系数(Q_(CV)~2)和外部样本校验复相关系数(Q_(ext)~2),分别为0.802、0.710、0.552和0.871、0.864、0.760。表明用3D-HoVAIF表征TIBO类抗艾滋病药物分子结构信息较好,建立QSAR模型的稳定性和预测能力良好,运用ANN建模优于PLS及前人报道的多元线性回归(multiple linear regression,MLR)。
Three-dimensional holographic atom field interaction (3D-HoVAIF) was used to study the quantitative structure-activity relationship (QSAR) of 89 anti-HIV drugs against tetrahydroimidazobenzophenazone (TIBO). Using partial least-squares regression and artificial neural network modeling respectively, the internal and external double verification methods are used to deeply analyze and test the stability of the model. The correlation coefficient (R_ (cum) ~ 2) and the leave-one-out (LOO) cross-validation (CV) correlation coefficient (Q_ (CV) ) And the external sample calibration complex correlation coefficient (Q ext ~ 2), respectively 0.802,0.710,0.552 and 0.871,0.864,0.760. It is indicated that the molecular structure of TIBO anti-AIDS drug is better characterized by 3D-HoVAIF, the stability and prediction ability of QSAR model is better. The ANN model is superior to PLS and multiple linear regression (MLR) .