CLASSIFICATION OF QUATERNARY STRUCTURE USING SUPPORT VECTOR MACHINES AND BAYES METHODS
基于支持向量机和贝叶斯方法的蛋白质四级结构分类研究

The quaternary structure was classified using support vector machine method and Bayes method. It was found that the result of using support vector machine is the best, using 10-fold cross-validation test, the overall accuracy, true positive rate, Mattew's correlation coefficient and false negative rate are 74.2%, 84.6%, 0.474, 38.9% respectively; the result of Bayes method is not so good as that of the support vector machine method, the false negative rate of using 10-fold cross-validation test is the smallest. Those results show that the primary sequences of homo-oligomeric proteins contain quaternary information. The feature vectors appear to capture essential information about the composition and hydrophobicity of the residues in the surface patches that are buried in the interfaces of associated subunits. And they also show that the support vector machines is a specially effective method.