作　　者： ; ; ; ; ; ;

机构地区： 兰州大学化学化工学院

出　　处： 《分析化学》 2006年第12期1759-1762,共4页

摘　　要： 运用定量结构-性质关系(quantitative structure-property relationsh ip,QSPR)方法对人体必需的氨基酸比旋光度进行了预测,同时运用交互检验(LOO)方法对所建立的模型进行了检验。应用启发式算法对描述符进行筛选并建立线性回归模型,所建立的模型的相关系数(R2)为0.918,但分别用+1、-1代表左旋和右旋的分子手性后,重新建立多元线性回归模型,其相关系数(R2)变为0.970。本研究所建立的QSPR模型为预测手性化合物比旋光度提供了一种有效的新方法。 The quantitive structure-property relationship （QSPR） method was developed to predict the power rotation of the amino acid. In order to ensure the reliability of this model, the leave-one-out （ LOO ） cross validation was used to test the model. Molecular descriptors calculated from the structure alone were used to represent molecular structures. The heuristic method was utilized to select the descriptors and then construct the multi-linear regression model, leading to a good correlation coefficient （R^2） of 0. 918. In order to build more reliability model, the descriptor of molecular chirality was introduced. Then a new multi-linear regression model was constructed. The correlation coefficient （ R^2 ） of this model was 0. 970. This paper provides an easy, effective and highly accurate method for predicting the power rotation of the amino acid. It can be used to predict the power rotation of other chemicals with chirality.

关 键 词： 氨基酸 定量结构 性质关系 比旋光度 启发式算法 多元线性回归

领　　域： [理学] [理学]