Table 4

The table shows the class separation measured by 10-fold stratified cross-validation using a C4.5 decision tree learner on the training dataset (MTS). In each row a different combination of input parameters was used as indicated by X (present) in the last four columns. Separation baseline for this dataset is 50%
Nr
 %sep
 TN
 FP
 FN
 TP
 FPrate
 TPrate
 score
 antigen
 PTM
 var
a
 57.78
 1086
 759
 799
 1046
 0.41
 0.57
 X
 -
 -
 -
b
 54.58
 1242
 603
 1073
 772
 0.33
 0.42
 -
 X
 -
 -
c
 49.86
 920
 925
 925
 920
 0.50
 0.50
 -
 -
 X
 -
d
 60.08
 1382
 463
 1010
 835
 0.25
 0.45
 -
 -
 -
 X
e
 70.41
 1399
 446
 646
 1199
 0.24
 0.65
 X
 X
 X
 X
f
 63.47
 1607
 238
 1110
 735
 0.13
 0.40
 -
 X
 X
 X
g
 65.42
 1337
 508
 768
 1077
 0.28
 0.58
 X
 -
 X
 X
h
 65.28
 1259
 586
 695
 1150
 0.32
 0.62
 X
 X
 -
 X
i
 66.18
 1263
 582
 666
 1179
 0.32
 0.64
 X
 X
 X
 -
j
 56.02
 1298
 547
 1076
 769
 0.30
 0.42
 -
 X
 X
 -
k
 61.00
 1446
 399
 1041
 804
 0.22
 0.44
 -
 X
 -
 X
l
 63.82
 1520
 325
 1010
 835
 0.18
 0.45
 -
 -
 X
 X

Column abbreviations: %sep (% class separation), score (sum score), antigen (PCA19 derived antigenicity score), PTM (post-translational modifications), var (variability score), TPrate (True Positive rate) and FPrate (False Positive rate).

Sollner et al. Immunome Research 2008 4:1   doi:10.1186/1745-7580-4-1