Understanding Protein Structure Prediction Using SVM_DT

A novel approach to rule generation for understanding protein secondary structure prediction by integrating merits of both support vector machine and decision tree into a new algorithm called SVM_DT, which is better than that of decision tree and is similar to that of SVM.

Abstract

The explanation of a decision made is important for the acceptance of machine learning technology, especially for such applications as bioinformatics. Support vector machines (SVM) have shown strong generalization ability in a number of application areas, including protein structure prediction. However, it is a black box model. On the other hand, a decision tree has good comprehensibility. In this paper, a novel approach to rule generation for understanding protein secondary structure prediction by integrating merits of both support vector machine and decision tree is presented. This approach combines SVM with decision tree into a new algorithm called SVM_DT. The results of the experiments of protein secondary structure prediction on RS126 data sets show that the comprehensibility of SVM_DT is much better than that of SVM. Moreover, the generalization ability of SVM_DT is better than that of decision tree and is similar to that of SVM. Hence, SVM_DT can be used not only for prediction, but also for guiding biological experiments.