新三取代嘧啶苯磺酰脲衍生物除草活性的QSAR

QSAR of the herbicidal activity of new trisubstituted pyrimidine phenylsulfonylurea derivatives

利用MOPAC-AM1方法计算了23种新三取代嘧啶苯磺酰脲衍生物的28个量化参数.基于这些参数,运用最佳变量子集回归方法建立新三取代嘧啶苯磺酰脲衍生物对油菜和反枝苋的除草活性(Bf和Af)的三参数(Hf,Q17,Q21)回归方程,相应复相关系数(R2)、去一法(LOO)交互检验复相关系数(Q2)分别为0.876,0.808和0.864,0.799,表明所建立的定量结构/活性关系(QSAR)模型具有良好的稳定性和预测能力.根据许禄规则和Andrea规则得到网络结构为3∶3∶1,Bf-BP和Af-BP模型的相关系数分别为0.995和0.997,显示了非常好的非线性关系.

28 quantum chemistry parameters of 23 new trisubstituted pyrimidine phenylsulfonylurea derivatives were calculated by using MOPAC-AM1 method. Based on these parameters, the three-parameter （Hf,Q17,Q21） regression equations between 2 herbicidal activity （Bf of Brassicanapus and Af of Amaranthus retroflexus L.） were established, and the structural descriptors of above molecules were established by Leaps-and-Bounds regression analysis. The correlation coefficient （R2）, leave-one-out （LOO） cross validation （Q2） of established Bf and Af models were 0.876, 0.808 and 0.864, 0.799, respectively, showing that the QSAR models have both favorable estimation stability and good prediction capability. By Xu Lu′s rule and Andrea′s rule, simulated with the final optimum BP neural network structure with 3∶3∶1 is obtained, with the correlation coefficient of Bf-BP model and Af-BP model being 0.995 and 0.997, respectively, implying that the relationship between Bf, Af and the three structural parameters have nice nonlinear correlation.