Skip to main content

Table 4 Performance evaluation on the genome-wide data sets for worm, fly cress, fish, and human.

From: Accurate splice site prediction using support vector machines

 

Worm

Fly

Cress

Fish

Human

 

Acc

Don

Acc

Don

Acc

Don

Acc

Don

Acc

Don

MC

          

auROC(%)

99.62 ± 0.03

99.55 ± 0.02

98.78 ± 0.10

99.12 ± 0.05

99.12 ± 0.03

99.44 ± 0.02

98.98 ± 0.03

99.19 ± 0.05

96.03 ± 0.09

97.78 ± 0.05

auPRC(%)

92.09 ± 0.28

89.98 ± 0.20

80.27 ± 0.76

78.47 ± 0.63

87.43 ± 0.28

88.23 ± 0.34

63.59 ± 0.72

62.91 ± 0.57

16.20 ± 0.22

24.98 ± 0.30

WD

          

auROC(%)

99.77 ± 0.02

99.82 ± 0.01

99.02 ± 0.09

99.49 ± 0.05

99.37 ± 0.02

99.66 ± 0.02

99.36 ± 0.04

99.60 ± 0.04

97.76 ± 0.06

98.59 ± 0.05

auPRC(%)

95.20 ± 0.29

95.34 ± 0.10

84.80 ± 0.35

86.42 ± 0.60

91.06 ± 0.15

92.21 ± 0.17

85.33 ± 0.38

85.80 ± 0.46

52.07 ± 0.33

54.62 ± 0.54

WDS

          

auROC(%)

99.80 ± 0.02

99.82 ± 0.01

99.12 ± 0.09

99.51 ± 0.05

99.43 ± 0.02

99.68 ± 0.02

99.38 ± 0.04

99.61 ± 0.04

97.86 ± 0.05

98.63 ± 0.05

auPRC(%)

95.89 ± 0.26

95.34 ± 0.10

86.67 ± 0.35

87.47 ± 0.54

92.16 ± 0.17

92.88 ± 0.15

86.58 ± 0.33

86.94 ± 0.44

54.42 ± 0.38

56.54 ± 0.57

  1. Displayed are auROC and auPRC scores for acceptor and donor recognition tasks as archived by the MC method and two support vector machine approaches, one with the weighted degree kernel (WD) and one with the weighted degree kernewith shifts (WDS).