Prediction of discontinuous B-cell epitopes using logistic regression and structural information

Abstract

Computational prediction of discontinuous B-cell epitopes remains challenging, but it is an important task in vaccine design. In this study, we developed a novel computational method to predict discontinuous epitope residues by combining the logistic regression model with two important structural features, B-factor and relative accessible surface area (RASA). We conducted five-fold cross-validation on a representative dataset composed of antigen structures bound with antibodies and independent testing on Epitome database, respectively. Experimental results indicate that besides the well-known RASA feature, B-factor can also be used to identify discontinuous epitopes. Furthermore, these two features are complementary and their combination can remarkably improve the prediction performance. Comparison with existing approaches shows that our method can achieve better performance in terms of average AUC value and sensitivity for predicting discontinuous B-cell epitopes. © 2011 Liu R, et al.