分子全息QSAR方法预测苯衍生物对蝌蚪的急性毒性

采用半静态培养法测定了一系列苯衍生物对蝌蚪的12h致死毒性.应用分子全息QSAR方法构建了化合物的亚结构与毒性之间的定量模型,探讨了分子全息长度、碎片尺寸以及区分常数对模型质量的影响,并检验了模型的稳健性.结果显示,研究的51种苯衍生物的半致死毒性(12h log1 /LC₅₀,mol/L)分布在2.07～4.56之间.最优HQSAR模型以6～7为碎片尺寸,83为全息长度,6个主成分,Leave-One-Out交叉验证的r²_cv是0.849,最终的非交叉验证的拟合模型r² 为0.942 ,表明模型稳健性能良好,可用于预测苯衍生物的毒性.

Hologram Quantitative Structure-Activity Relationship for Predicting Acute Toxicity of Benzene Derivatives to the Tadpoles (Rana japonica)

Acute lethal toxicity, the negative logarithm of 12 h acute median lethal molar concentration (expressed as 12h-log1/LC50, mol/L) of a series of benzene derivatives to Rana japonica tadpoles was determined. The relationship between the structure of benzene derivatives and their acute lethal toxicity was investigated by using hologram quantitative structure-activity relationship (HQSAR). The influence of hologram length, fragment size and distinction parameters on the quality of HQSAR model was studied. The robustness and predictive ability of the model were also validated by Leave-One-Out cross-validation procedure. The tested 51 compounds revealed a range of acute median lethal toxicity (12h-log1/LC50, mol/L) from 2.07 to 4.56. As a result, the best model was generated using a fragment size of 6-7 from a hologram length of 83 with 6 components. The HQSAR model showed a conventional correlation coefficient r2 of 0.942 and a Leave-One-Out cross-validation r2(cv) equal to 0.849, indicating the model possess high statistical quality in the prediction of acute toxicity of novel benzene analogs.