Prediction of B-cell Linear Epitopes with a Combination of Support Vector Machine Classification and Amino Acid Propensity Identification
Table 2
Comparison of the performances of LEPS, BepiPred, ABCPred, BCPred, and FBCPred systems.
Systems
SENa
SPEa
ACCa
PPVa
MCCa
PC dataset
LEPS
12.78
88.33
61.66
45.12
3.65
BepiPred
48.23
59.72
55.33
38.19
7.49
65.46
40.26
48.89
36.21
5.13
BCPred
50.92
59.35
52.83
36.07
4.43
FBCPred
51.03
52.55
52.20
35.26
3.17
AntiJen dataset
LEPS
26.72
84.48
73.81
28.85
10.10
BepiPred
51.79
57.61
55.52
22.02
6.04
ABCPred0.8
67.33
40.40
44.70
21.83
5.46
BCPred
58.84
54.87
53.92
23.34
8.93
FBCPred
60.31
51.21
51.45
22.33
6.73
HIV dataset
LEPS
48.33
74.84
63.45
71.44
22.76
BepiPred
50.16
60.85
56.72
61.22
9.72
ABCPred0.7
87.97
14.65
56.59
56.33
5.64
BCPred
80.18
54.57
66.57
65.55
29.80
FBCPred
73.20
58.20
67.13
65.56
27.81
AHP datasetc
LEPS
26.97
84.22
72.52
32.07
10.36
BepiPred
51.48
57.91
55.57
25.06
6.32
ABCPred0.8
68.28
39.06
45.58
24.51
5.45
BCPred
59.45
54.80
54.50
26.32
9.73
FBCPred
60.40
51.66
52.31
25.38
7.60
aSEN: sensitivity; SPE: specificity; PPV: positive prediction value; ACC: accuracy; MCC: Matthews’ correlation coefficient, unit, %.
bThe subscripts of ABCPred denote threshold values according to the highest accuracy.
cThis dataset is a merge of the other 3 datasets.
