论文部分内容阅读
采用三维全息原子场作用矢量(3D-HoVAIF)对115个酰肼类抗结核病药物进行定量构效关系(QSAR)研究.运用多元线性回归(MLR)和偏最小二乘回归(PLS)建模,同时采用内部及外部双重验证的办法对所得模型稳定性能进行了深入分析和检验.MLR建模和PLS建模的复相关系数(R_(cum)~2)、留一法交互校验复相关系数(Q_(LOO)~2)以及外部样本校验复相关系数(Q_(ext)~2)分别为0.766、0.663、0.748和0.733、0.614、0.715.结果表明,3D-HoVAIF能较好表征抗结核药物分子结构信息,所建QSAR模型具有良好稳定性和预测能力,为抗结核病药物的研发提供一定的理论基础.
Quantitative structure-activity relationship (QSAR) studies of 115 hydrazide anti-tuberculosis drugs were conducted using 3D-HoVAIF vector.Multiple linear regression (MLR) and partial least squares regression (PLS) At the same time, the internal and external double verification methods were used to analyze and test the stability of the model.Multiple correlation coefficients (R cum (~ 2)) between MLR modeling and PLS modeling, (Q_ (LOO) ~ 2) and the external sample calibration Q_ (ext) ~ 2 were 0.766,0.663,0.748 and 0.733,0.614,0.715 respectively.The results showed that 3D-HoVAIF could better characterize anti-TB The drug molecular structure information, the QSAR model has good stability and predictive ability, and provides a theoretical basis for the development of anti-tuberculosis drugs.