Studies on Quantitative Structure-Toxicity Relationship of Alcohols, Phenols, Ethers, Ketones and Esters

In order to find out the quantitative relationship between toxicity and structures of alcohols, phenols, ethers, ketones, esters, some structures of compounds were characterized by the values of molecular vertexes and their interaction. Two models of the quantitative structure-toxicity relationship (QSTR) were established by the methods multiple linear regression (MLR) and stepwise regression (SMR). A comparison of two models indicated that the model 2 (M2) showed better simulation results and the multiple correlation coefficient (R) was 0.952 and the value of standard deviation (SD) was 0.325. Jacknife method used for testing its stability indicated that the regression model 2 had an acceptable stability and a good predictive ability. In addition, the model was tested by the cross-validation with the leave-one-out (LOO) procedure. And the multiple correlation coefficient in cross-validation (RCV) was 0.927 and the value of standard deviation (SDCV) was 0.396, which showed the stability and predictive ability of the model were desirable.