分子距边矢量研究链烃与醛酮的定量构效关系

Study On Quantitative Structure-Property Relationship of Chain Hydrocarbons,Aldehydes and Alkanones by Molecular Distance-edge Vector

按碳原子及键合特性分类定义并计算了链烃包括烷、烯、炔、双烯、烯炔烃的分子距离 边数矢量(MDE),将153个链烃的MDE矢量与相应的沸点相关联,得到了良好的线性模型.复相关系数R=0.9976,均方根误差RMS=4.69K.同时还进行了随机检验和交互检验,其复相关系数和均方根误差分别为R=0.9975、RMS=4.72K和R=0.9972、RMS=5.13K.结果表明模型具有良好的稳定性和预测能力.进一步对杂原子用染色因子进行标识,提出了一种适用于含杂原子体系分子结构描述的MDE矢量,并以72种脂肪醛和脂肪酮的沸点为例,进行了定量结构 性质关系 (QSPR)研究,同样得到了满意的结果:分子定量建模、随机抽样建模和定量预测的复相关系数及均方根误差分别为R=0.9989、0.9990、0.9980及RMS=3.93K、4.12K、5.52K.表明MDE矢量与链烃及醛酮的沸点均具有良好的相关性,肯定了该矢量的合理性和有效性,可望在定量结构 性质/活性 (QSPR/QSAR)研究中获得广泛应用.

An extented molecular distance edge (MDE,μ) vector of chain hydrocarbons including alkanes,alkens,alkynes,dienes,and alkenynes are defined and calculated. Quantitative structure property relationship (M1) on boiling point of chain hydrocarbons is proposed by multiple linear regression method with good results: root mean square error RMS=4.69K and correlation coefficient R=0.9976. Random sampling modeling and cross validation (CV) procedure are also performed and satisfactory results have been obtained with average correlation coefficient being R=0.9975,0.9971 and average RMS=4.72K,4.83K respectively. In order to generalize MDE for other organic compounds,a novel molecular distance edge (MDE) vector,which can be applied to discribe the molecular structure of both fatty aldehydes and fatty alkanones,is developed by using dying factors to express the carboxyl ( > CO) based on a molecular distance edge (MDE) vector of alkanes.and the QSPR model between the new MDE vector and boiling point (Tb) is also developed for 72 aldehyde and alkanone compounds with good results:R=0.9989 and RMS=3.93K.Random modeling procedure is also performed with average correlation coefficient being R=0.9990 and average root mean square error RMS=4.12K; and quantitative prediction is done with R=0.9980 and RMS=5.52K. All these results show that the MDE vector is a useful vector in QSPR/QSAR studies.