【摘要】 依据经济合作与发展组织(OECD)关于定量结构-活性关系(QSAR)模型构建和使用导则,将780个有机化合物,以4:1的比例随机划分为训练集(624个化合物)和验证集(156个化合物),通过多元线性回归(MLR)方法构建了一个包含12个描述符的有机化合物鱼类生物富集因子(BCF)的QSAR模型。QSAR模型的调整决定系数R2ad j=0.809,去一法交叉验证系数Q2LOO=0.803,外部验证系数Q2EXT=0.732,表明模型具有较好的拟合优度、稳健性和预测能力。采用欧几里德距离方法表征模型应用域,通过威廉姆斯图分析模型离群点,并对模型进行机理解释。所构建的模型,可以用于预测应用域内有机化学品的生物富集因子。更多还原

【Abstract】 According to the principles of development and validation for quantitative structure-activity relationship(QSAR) proposed by Organization for Economic Cooperation and Development(OECD), a QSAR model of bioconcentration factors in fish was built. 780 organic compounds were randomly divided into training set(624 compounds) and validation set(156 compounds) at a ratio of 4:1. The method that developed the model with 12 molecular descriptors was multiple linear regressions(MLR). The adjusted coefficient of determination(R2 adj), the leave-one-out cross validated coefficient(Q2 LOO) and external explained variance(Q2 EXT) was 0.809, 0.803 and 0.732,respectively. It is indicated that the model had satisfied goodness-of-fit, robustness and predictive ability. The application domain of the model was characterized by the Euclidean distance, outliers were analyzed by Williams andmechanism was interpreted subsequently. The obtained model was able to predict the BCF of organic chemicals within the application domain.更多还原