Skip to main content

Table 2 Prediction accuracies for the dataset containing only weak binders and non-binders

From: Predicting MHC class I epitopes in large datasets

  Allele Name Binders Non-Binders Total DynaPredPOS NetMHC SVMHC YKW
   (n) (n) (n) AUC AUC AUC AUC
1 A*0201 616 135 751 0.67 0.77 0.65 0.66
2 A*0202 286 87 373 0.53 0.70 0.41 0.54
3 A*0203 261 126 387 0.58 0.79 0.58 0.60
4 A*0206 264 74 338 0.56 0.73 0.57 0.57
5 A*0301 335 106 441 0.58 0.70 0.60 0.59
6 A*1101 374 91 465 0.56 0.69 0.62 0.63
7 A*3101 278 103 381 0.42 0.69 0.48 0.56
8 A*3301 129 72 201 0.62 0.52 0.39 0.63
9 A*6801 273 96 369 0.54 0.68 0.44 0.57
10 A*6802 227 123 350 0.52 0.74 0.50 0.50
11 B*1501 169 33 202 0.59 0.79 0.49 0.59
  Average AUC     0.56 0.71 0.52 0.59
  1. Results from Dataset I, in which only weak binders (50 nM to 500 nM binding affinity) and non-binders (500 nM to 1000 nM binding affinity) were included. Alleles included in Dataset F, which had fewer than 200 binders and non-binders in total in Dataset I, were no longer included in the analysis. The average AUC for each method is included at the bottom of each column.