AUC and accuracy of epitope predictions on the test set^*.

*Comparison of our final ensemble classifier (machine learning approach ML-advanced, see Figure 1) to the classifier by El-Manzalawy et al. [22][23]. Both were trained on the original training set (three times more non-binding than binding peptides) and on the balanced training set (equal number of binding and non-binding peptides) and finally applied on the test set. The results on the balanced training set can directly be compared to those in Table 1.