含氧有机物结构与毒性(log1/C)关系研究  

Study on Quantitative Structure-toxicity Relationship of Oxygen-containing Organic Compounds

作　　者：兰作平[1,2] 廖立敏[3] 余瑜[1] 朱照静[2] 

机构地区：[1]重庆医科大学药学院,重庆400016 [2]重庆医药高等专科学校,重庆400030 [3]内江师范学院资源与环境科学学院,内江641112

出　　处：《世界科技研究与发展》2012年第1期39-42,共4页World Sci-Tech R&D

基　　金：重庆市卫生局医学科研资助项目(2009-2-134);四川省教育厅青年基金(09ZB036);重庆市教委资助项目(KJ112501)资助项目

摘　　要：为探寻含氧有机物结构与毒性(log1/C)的关系,应用分子顶点及顶点相互作用值对部分含氧有机物进行了结构表征。采用多元线性回归及逐步回归建立了2个定量结构-毒性关系模型。经过比较,发现模型(M2)具有最佳模拟结果,此时模型的复相关系数(R)为0.952,标准偏差(SD)为0.325。采用Jackknife法对模型进行了稳健性检验,结果表明,回归模型具有可接受的总体稳健性及良好的预测能力;另外,采用留一法(leave-one-out)对模型进行交互检验,复相关系数(R_(cv))为0.927,标准偏差(SD_(cv))为0.396,这也说明所建模型的稳定性与预测能力均较为理想。In order to explore the relationship between structures and toxicity of oxygen-containing organic compounds, values of molecular vertexes and their interaction were proposed to describe the structures of oxygen-containing organic compounds. Through multiple linear regression （MLR） and stepwise regression, two models of toxicity were obtained. By comparison, model （M2） showed better results of simulation. The correlation coefficient （R） of the model was 0. 952 and the standard deviation （SD） of the model was 0. 325. Jackknife method was used to test its stability, the results showed that the regression model with acceptable stability and good predictive ability. In addition, the model was evaluated by performing the cross-validation with the leave-one-out （LOO） procedure. The cross-verification multiple correlation coefficient （ Rcv ） and the standard deviation （ SDcv ） were 0. 927 and 0. 396, respectively. This also showed that the stability and predictive ability of the model was desirable.

关 键 词：含氧有机物 毒性 分子顶点 定量结构-性质相关 

分 类 号：O621.2[理学—有机化学]