Welcome to use BJTEpitope for T-cell epitope prediction for the MHC class I allele HLA-A*0201
Accurate prediction of T-cell epitopes plays an important role in T cell-mediated vaccine design. In this paper, we firstly constructed a prediction model for T-cell epitope using Naïve Bayes method. The training dataset contains 44 epitopes and 45 non-epitopes. The leave-one-out cross-validation accuracy (LOOCV) is 96.63%. Then, an independent test dataset was used to evaluate the performance of the model. The test set contains 44 epitopes and 44 non-epitopes respectively. The prediction accuracy and Matthews¡¯s correlation coefficient (MCC) are 86.36% (76/88) and 0.7303 respectively, which are both the highest values among all following prediction servers SVRMHC, EpiJen, MHCPred, SVMHC, SYFPEITHI, and BIMAS using the same test set. Finally, we developed an easy-to-use program BJTEpitope for T-cell epitope prediction using the model. Following is the relationship between the number of features and average prediction accuracy.