会议论文

Application of support vector regression to QSAR study of neonicotinoid insecticides

中文

2011

76-81

华中三省(河南、湖北、湖南)昆虫学会2011年学术年会

华中三省(河南、湖北、湖南)昆虫学会2011年学术年会论文集

2011-11-12

河南新乡

中国昆虫学会<&wdkj&>河南省昆虫学会

本校为第一机构

为深入认识新烟碱类杀虫剂活性与其结构的关系,从32种新烟碱类化合物的疏水性参数等24个描述符出发,以基于支持向量回归(SVR)的"多轮末尾淘汰法"对描述符进行汰选,并分析所保留的6个描述符重要性,得出因子重要性排序如下:疏水性参数＞分子范德华体积＞分子表体积＞二面角2＞吡啶环净电荷＞轨道能级差;结合单因子效应得出各保留因子影响活性趋势.以保留的描述符构建定量构效关系(QSAR)模型,其拟合MSE为0.080,R2为0.959,留一法MSE为0.300,R2为0.848,明显优于参比模型.基于SVR非线性筛选描述符的QSAR的研究方法具有广泛应用前景.

　　In order to understand the relationship between activity and structure of Neonicotinoid insecticides, 24 descriptors including Hydrophobic parameter for 32 Neonicotinoid insecticides were initially adopted, they were further selected by applying "multi-round optimization" based on support vector regression (SVR), and 6 important descriptors were retained finally.Further analysis demonstrated that the order of the importance of the retained descriptors is : Hydrophobic parameter > Molecular van der Waals volume > Molecular size volume > Dihe...