Skip to main content

Table 2 Classification result of peptide-encoding schemes

From: EpicCapo: epitope prediction using combined information of amino acid pairwise contact potentials and HLA-peptide contact site information

Method # of features 10-fold cross validation on training dataset only Holdout method using training dataset and testing dataset
sens spec F1 ACC AUC sens spec F1 ACC AUC
EpicCapo 360 0.883 ± 0.005 0.792 ± 0.006 0.886 ± 0.003 0.841 ± 0.004 0.915 ± 0.001 0.883 0.744 0.831 0.815 0.882
EpicCapo(3 AAPPs*) 27 0.876 ± 0.005 0.821 ± 0.005 0.862 ± 0.003 0.848 ± 0.003 0.916 ± 0.001 0.855 0.777 0.828 0.817 0.878
DPPS 90 0.865 ± 0.005 0.760 ± 0.007 0.834 ± 0.004 0.816 ± 0.004 0.888 ± 0.001 0.868 0.697 0.807 0.785 0.878
FASGAI 54 0.847 ± 0.004 0.761 ± 0.004 0.825 ± 0.003 0.801 ± 0.003 0.882 ± 0.001 0.840 0.730 0.803 0.787 0.874
z-scale 45 0.847 ± 0.005 0.732 ± 0.005 0.815 ± 0.004 0.793 ± 0.004 0.873 ± 0.002 0.848 0.676 0.788 0.765 0.858
ISA/ECI 18 0.799 ± 0.005 0.652 ± 0.005 0.760 ± 0.003 0.731 ± 0.003 0.797 ± 0.001 0.829 0.643 0.766 0.739 0.796
Binary encoding 180 0.883 ± 0.005 0.721 ± 0.006 0.831 ± 0.003 0.807 ± 0.003 0.883 ± 0.002 0.887 0.705 0.820 0.799 0.879
  1. Means and standard deviations were calculated by 20 iterations of 10-fold cross validation.
  2. Underlined values represent the highest performance.
  3. sens = sensitivity; spec = specificity; F1 = F-score; ACC = accuracy; AUC = area under the curve.
  4. *These three top-ranked AAPPs were MICC010101, SIMK990101, and SIMK990105 (see Additional file1).