Linear B-cell Epitope Prediction Research Based On PCA And SVM

B cell epitope is apart of antigen surface that is recognized by B cell surface receptor, and combine with antibody specificity of the area. B cell epitope is a group of sites which has its own characteristics. It distributes on surface of antigens by arranged way of the linear or conformation sex, thus causes the mutually binding of antigen and antibody, mediating humoral immune response. Epitope predicted has played a vital role in vaccine development and disease diagnosis. In recent years, the researchers began epitope prediction by computer, verify the follow-up biological experiment. The method that combines computer with experiment not only can save a lot of cost, but also guarantee the accuracy of the results, thus improving the work efficiency.The research of this paper is to construct the standard data sets of the linear B cell epitope. Firstly, extracted features based on the five characteristics of antigens on the surface of amino acids and AAindex database provides 527 amino acids of the physical and chemical properties. This paper applies a principal component analysis(PCA) dimensionality reduction linear B cell epitope prediction Algorithm. The core idea of above algorithm is affinity peptide prediction for different physical and chemical properties of amino acids. In condition of ignored the specificity, the forecasting method for integral can affect the performance of the predictor. Secondly, overall antigen amino acids are classified using support vector machine(SVM) method. In this paper, first using PCA dimension reduction, then applying support vector machine(SVM) algorithm. This kind of strategy can avoid this problem of ignored the specificity and effectively improve the performance of B cell epitope prediction algorithm, and then predict the table, and use the sensitivity, specificity, accuracy, and markov correlation coefficient of four evaluation parameters set up a comprehensive evaluation system.