MLR和人工神经网络研究取代芳烃定量构效关系  

QSAR study of substituted aromatic compounds using MLR and artificial neural network

作　　者：金飙[1] 陈忱[1] 金俏[2] 

机构地区：[1]沈阳化工大学环境与安全工程学院,辽宁沈阳110142 [2]北京石油化工学院化学工程学院,北京102600

出　　处：《计算机与应用化学》2014年第12期1503-1506,共4页Computers and Applied Chemistry

基　　金：国家自然科学基金项目(51408372)

摘　　要：采用量子化学密度泛函B3LYP方法,计算了17个取代芳烃类化合物的电子结构参数;筛选出影响发光菌毒性显著的5个变量,并建立其结构与毒性之间的MLR模型和ANN模型以资对比。所建MLR模型的相关系数R=0.839,标准误差Se=0.298;ANN模型的相关系数.R=0.978,标准误差Se=0.126。表明所建立的ANN模型具有更好的稳定性和预测能力。结果表明,取代芳烃类化合物对发光菌毒性与零点振动能ZPE和分子最高占有轨道能EH成反比例关系。For 17 substituted aromatic compounds,quantum chemistry calculation of electronic properties were carried out at density functional theory（DFT） B3LYP/ 6-311＋G＊ level by Gaussian09.5 important parameters were selected and the quantitative structure-activity relationship（QSAR） model was set up by multiple linear regressions（MLR） method.Furthermore,using artificial neural network（ANN）,the QSAR model was obtained in order to make contrast.For the artificial neural network method,the correlation coefficient R=0.839 and the standard error Se=0.298,while for the multiple linear regression analysis R=0.978 and Se=0.126.These shows that the QSAR models have both favorable estimation stability and good prediction capability.This indicates that the toxicity of these compounds to photobacterium phosphoreum and both ZPE and EH are in direct proportion.Successful QSARs were developed to predict the toxicity of substituted aromatic compounds to photobacterium phosphoreum.

关 键 词：取代芳烃 人工神经网络 定量结构毒性关系 密度泛函 

分 类 号：O641[理学—物理化学]