有机污染物理化性质与毒理学性质的预测

本文采用分子电性距离矢量（MEDV）分别表征了多种有机污染物的分子结构,应用多元线性回归（MLR）技术,同时采用逐步回归结合统计检测对模型变量进行筛选,分别建立了有机污染物的沸点和急性毒性与MEDV的定量结构-性质关系（QSPR）模型。继用留一法和外部样本对模型稳定性能进行检验。结果表明：所建QSPR模型有良好的稳定性和预测能力,该研究结果为预测有机污染物的理化性质和毒理学性质提供了的理论依据。

Prediction of physical chemical property and genotoxicity of organic contaminants

The molecular electro-negativity distance vector（MEDV） was used to describe the chemical structure of diverse organic contaminants, with the help of multiple linear regression（MLR） technique and the stepwise multiple regression（SMR） method to filter variables, QSPR models were established with high correlation coefficients, they are the relationship between MEDV and boiling point, biodegradability, genotoxicity.In order to test the stability and predictability of the QSPR models, the cross validation with leave-one-out of procedure and the predictability for outer samples are carried out, the results of which showed that the established QSPR models possess good stability and predictability.The theory evidence were provided for predilting physicochemical property and toxicology property of organic pollutant.