含氮杂环化合物理化性质和生物活性的QSPR/QSAR分析  

QSPR/QSAR of Physicochemical Property and Bioactivity of Nitrogen-Containing Heterocyclic Compounds

作　　者：陈艳[1] 岳玮[1] 王彬[1] 

机构地区：[1]徐州工程学院化学化工学院,江苏徐州221111

出　　处：《武汉大学学报（理学版）》2014年第1期52-56,共5页Journal of Wuhan University:Natural Science Edition

基　　金：国家自然科学基金(21272095);徐州工程学院培育项目(XKY2011102)资助项目

摘　　要：在分子拓扑理论的基础上,计算了含氮杂环化合物的3类拓扑指数:分子连接性指数、形状指数和分子电性距离矢量.利用多元线性回归的方法建立了含氮杂环化合物的疏水性参数(lgP)和对淡水纤毛虫毒性(-lgc)的QSPR/QSAR(定量结构-性质/活性相关)模型,相关系数分别为0.998 1和0.980 6.模型经Jackknife法和逐一剔除法(LOO)交互检验证明具有良好的稳健性和预测能力,为了进一步提高相关系数0.980 6模型的相关性,以模型中的3个参数为人工神经网络输入层,设定3∶3∶1的网络结构,构建人工神经网络(BP)算法模型,相关系数R提升为0.993 3.该模型能很好地预测含氮杂环化合物的理化性质和生物活性.Based on the chemical topological theory, Kier＇s molecular connectivity indexes （Xi）, molecular shape indexes（Kin） and molecular electronegativity-distance vector （Mk） were calculated. The QSPR/QSAR（quantitative structure-property/activity relationship） models of hydrophobic property parameter （lgP） and toxicity （- lgc） to freshwater ciliate of nitrogen-containing heterocyclic compounds were constructed by multiple linear regression with the correlation coefficient（R） of 0. 998 1 and 0. 980 6, respectively. The models exhibited excellent stability and pre- dictability evaluated by Jackknife method and leave-one-out （LOO） cross-validation procedure. In order to improve the correlation of the later model, three structural parameters were used as the input neurons of artificial neural network, and a 3 ： 3 ： 1 network architecture was employed. A satisfied model could be constructed with the back-propagation algorithm,and the correlation coefficient R was 0. 993 3. It is concluded that the models can be used to predict the physicochemical properties and bioactivities of nitrogen-containing heterocyclic compounds.

关 键 词：含氮杂环化合物 拓扑指数 疏水性参数 生物毒性 QSPR QSAR(定量结构一性质 活性相关) 

分 类 号：O613.61[理学—无机化学]