Lu X, Tao S, Hu H, Dawson R W
Department of Urban and Environmental Sciences, Peking University, Beijing , People's Republic of China.
Chemosphere. 2000 Nov;41(10):1675-88. doi: 10.1016/s0045-6535(00)00050-3.
A bioconcentration factor (BCF) estimation model for a wide range of nonionic organic compounds was developed on the basis of molecular connectivity indices and polarity correction factors. The nonlinear topological modeling using polarity correction factors resulted in the best BCF estimation quality for all of the 239 compounds studied, with a mean absolute estimation error of 0.478 log units. Residual analysis indicated that the estimation errors came from many sources including BCF measurement, test species, and selection of descriptors. Statistical robustness of the developed model was validated by modified jackknifed tests where random deletion of a set of compounds and specific deletion of a class of compounds were both performed. Comparison between the MCI-based (molecular connectivity indices) model and a Kow-based (octanol/water partition coefficient) model revealed that the BCF estimation based on topological parameters was as good as that achieved by Kow.
基于分子连接性指数和极性校正因子，开发了一种针对多种非离子有机化合物的生物浓缩因子（BCF）估算模型。使用极性校正因子的非线性拓扑建模，在所研究的239种化合物中，对BCF的估算质量最佳，平均绝对估算误差为0.478对数单位。残差分析表明，估算误差来自多个来源，包括BCF测量、试验物种和描述符的选择。通过修改后的留一法检验验证了所开发模型的统计稳健性，该检验既进行了一组化合物的随机删除，也进行了一类化合物的特定删除。基于分子连接性指数（MCI）的模型与基于辛醇/水分配系数（Kow）的模型之间的比较表明，基于拓扑参数的BCF估算与基于Kow的估算效果相当。
