论文部分内容阅读
基于定量结构-活性相关(QSAR)研究氯苯、氯酚类化合物的性质具有重要意义。采用分子电性距离矢量(MEDV)表征氯苯、氯酚类有机毒物的分子结构,运用多元线性回归建立定量结构毒性相关(QSTR)模型,同时采用逐步回归结合统计检测筛选模型变量,建立了22个氯苯、氯酚类有机污染物结构与对戈卑鱼(Guppy)半数致死浓度lg1/(LC_(50))回归方程。另外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。建模的复相关系数(R_(cum))、留一法(LOO)交互校验复相关系数(R_(CV))和外部样本校验复相关系数(Q_(ext))分别为0.976 2、0.944 6和0.973 1。表明用MEDV表征氯苯、氯酚类有机物分子结构信息较好,所建QSTR模型的稳定性和预测能力良好。
It is important to study the properties of chlorobenzene and chlorophenols based on quantitative structure-activity correlation (QSAR). The molecular structure of chlorobenzene and chlorophenols was characterized by molecular electrical distance vector (MEDV), and the quantitative structure-related toxicity (QSTR) model was established by multiple linear regression. The model variables were screened by stepwise regression and statistical analysis. Structure of chlorobenzene and chlorophenols organic pollutants and regression equation of lg1 / (LC_ (50)) to Guppy. In addition, the internal and external double verification methods are used to deeply analyze and test the stability of the model. (R_ (cum)), ROO (R_ (CV)) and Q_ (ext) of LOO were 0.976 2, 0.944 6 and 0.973 1. It is indicated that the molecular structure information of chlorobenzene and chlorophenols is better with MEDV, and the stability and prediction ability of QSTR model are good.