论文部分内容阅读
以2D-autocorrelation描述符为结构参数,采用PSO和逐步回归的方法进行变量筛选,再结合SVM等机器学习算法对28种苯丙烯盐类化合物对EBV-EA病毒的抑制性活性进行定量构效关系(QSAR)研究.研究结果表明,PSO-v-SVM模型具有最优的模型稳健性和预测效果.由PSO选入的构成该模型的5个2D-autocorrelation描述符为ATS5v,ATS6e,ATS8e,ATS3p,GATS5p;该模型对训练集的拟合和留一法交叉验证结果的相关系数R~2和q_(cv)~2分别为0.986和0.930,对测试集预测结果的相关系数R~2_(ext)达0.955.对5个变量的理化意义的分析表明,极化率、Van der Waals体积和电负性对苯丙烯盐类化合物的抑制性活性影响分别约占57.13%、15.90%和26.97%.
Using 2D-autocorrelation descriptors as structural parameters, PSO and stepwise regression were used to select variables. Combined with machine learning algorithms such as SVM, the quantitative structure-activity relationship of 28 styphnic salt compounds against EBV-EA virus was determined. (QSAR) .The results show that PSO-v-SVM model has the best model robustness and prediction effect.The five 2D-autocorrelation descriptors selected by PSO are ATS5v, ATS6e, ATS8e, ATS3p , And GATS5p. The correlation coefficients R ~ 2 and q_ (cv) ~ 2 of the model to the training set and the left one-way cross validation are 0.986 and 0.930, respectively. The correlation coefficient R ~ 2_ (ext ) Was 0.955.Analysis of the physico-chemical significance of the five variables showed that the influence of polarizability, Van der Waals volume and electronegativity on the inhibitory activity of the benzenesulfonate compounds were about 57.13%, 15.90% and 26.97%, respectively.