Studies Of QSPR Of Chlorinated Alkyl Chloroacetate Based On Molecular Electrical Distance Vector

Molecular electronegativity distance vector is a new vector molecules using Miao Shuzi constructing distance feature and atomic electronegativity of all atom molecular environment among. By simulating the interaction between the non hydrogen atoms, molecular electronegativity distance vector can indirectly characterize the molecular structure of organic molecules. Molecular descriptor has good resolution on the structures of various compounds, good correlation with the molecules of physical and chemical properties or biological activity.Therefore, this paper defines the atomic properties and atomic types, on 132 kinds of chloroacetic acid chlorinated alkyl esters of molecular electronegativity distance vector calculation, obtening 10 molecular descriptors.Retention index of gas chromatography is one of the most important chromatographic parameters identification of compounds. We applied several methods, partial least squares method, artificial neural network, genetic algorithm optimization of artificial neural network method, to analyze the Quantitative Structure-Pharmacokinetics Relationship for 132 kinds of chloroacetic acid chlorinated alkyl ester molecular electronegativity distance vector and the same fixed solution with different temperature gas chromatography index, and established the partial least squares linear regression model, the nonlinear model of artificial neural network and genetic algorithm to optimize artificial neural network model,the establishment of molecular electronegativity distance vector descriptors and gas chromatography retention index by three kinds of model of QSPR relationship.The 132 kinds of chloroacetic acid chlorinated alkyl ester is divided into training set and the prediction set. The partial least square method, artificial neural network, genetic algorithm optimization of artificial neural network method are all used to establish the model to train on the training set. Then we use leaving the group method and external validation method to obtain the prediction value of the prediction set. External validation method is used to examine the model’s ability to predict and leaving a group of model checks the generalization performance estimation ability and stability.Among all the three models established in this passage, the partial least square method belongs to linear model; artificial neural network model and genetic algorithm optimization of artificial neural network model belong to non-linear model.The three models study on comprehensive of chloroacetic acid chlorinated alkyl ester molecular electronegativity distance vectordescriptors and gas chromatographic retention indices of QSPR and satisfactory results areobtained.