The accurate QSPR models to predict the bioconcentration factors of nonionic organic compounds based on the heuristic method and support vector machine

The heuristic method (HM) and support vector machine (SVM) were used to build the linear and nonlinear quantitive structure–property relationship (QSPR) models for the prediction of the fish bioconcentration factors (BCF) for 122 diverse nonionic organic chemicals using the three descriptors calculated from the molecular structure alone and selected by HM. Both the linear and nonlinear model can give very satisfactory prediction results: the square of correlation coefficient R2 was 0.929 and 0.953, the root mean square (RMS) error was 0.404 and 0.331, respectively for the whole dataset. The prediction result of the SVM model is better than that obtained by heuristic method, which proved SVM was a useful tool in the prediction of the BCF. At the same time, the HM model showed the influencing degree of different molecular descriptors on bioconcentration factors and then could improve the understanding for the bioconcentration mechanism of organic pollutants from molecular level.