Prediction of Carcinogenicity of Chemical Substances by Combining Support Vector Machine with Ensemble Learning and Decision Tree

Abstract

Among chemicals existing in environments, very few are identified as carcinogens. But from the viewpoint of the animal protection, it becomes a social, urgent problem to develop a technology to predict the carcinogenicity of the chemicals from the structures. Up to now, a lot of researches have been done, but it is extremely difficult to predict the carcinogenicity of non-congeners with various chemical structures. In our previous study, relationships between carcinogenic ranks for many non-congeners and descriptors generated from their three dimensional molecular structures were analyzed using a support vector machine (SVM) technique, and it was concluded that SVM showed a satisfactory predictability for screening test using animals within short time, but its predictability power was inferior to an artificial neural network (ANN). In this study, SVM was combined with a decision tree method and an ensemble learning method to improve its predictability power, and it was shown that the combination led to a rather improvement in predicting the carcinogenicity of non-congeners from their structures.