基于DFT方法建立多溴代联苯醚代谢产物的正辛醇-水分配系数的预测模型

Prediction of n-octanol/water partitioning coefficients for selected metabolites of polybrominated diphenyl ethers by DFT method

基于量子化学方法建立了多溴代联苯醚（PBDEs）代谢产物的正辛醇-水分配系数（logKow）的预测模型,整个数据集包括19个羟基多溴代联苯醚（HO-PBDEs）和15个甲氧基多溴代联苯醚（MeO-PBDEs）,logKow的数值跨越3个数量级（4.63~7.67）范围.所有化合物应用密度泛函理论（DFT）进行结构优化,在最优构型的基础之上计算分子极化率等6个量子化学描述符,并采用多元线性回归（MLR）建立模型.结果表明,分子极化率和氢原子最正净电荷是影响化合物在正辛醇相和水相之间分配的主要因素.模型具有良好的统计学性能,相关系数的平方r2=0.941,均方根误差rms=0.198.模拟外部验证和交叉验证表明模型具有良好的稳健性和预测能力,可用于同系列化合物logKow的预测.

A quantum chemical method has been developed to estimate the n-octanol/water partitioning coefficient （logKow） of metabolites of polybrominated diphenyl ethers （PBDEs）. The data set covers 19 hydroxylated-PBDEs （HO-PBDEs） and 15 methoxylated PBDEs （MeO-PBDEs）, with logKow values varying from 4.63 to 7.67. The optimized ground-state geometries were calculated using density functional theory to quantify six quantum chemical descriptors. The resultant regression model through multiple linear regression had a correlation coefficient r2 = 0.941 and a root-mean-square error rmse = 0.198. The polarizability and the most negative atomic partial Mulliken charge on a hydrogen atom were the most important parameters affecting the partitioning of the PDBEs between n-octanol and water. Simulated external validation and leave-10%-out cross-validation demonstrate the statistical robustness and prediction power of the derived model, which could be applied to predict logKow values of HO-PBDE and MeO-PBDE congeners.