Skip to main content

Table 10 Comparative performance of the novel feature set and traditional feature set using SVM

From: Identification of protein functions using a machine-learning approach based on sequence-derived properties

Protein class

Novel feature set (33 features)

Traditional feature set (451 features)

 

Training Accuracy

Test accuracy

Sensitivity

Specificity

AUC

Training accuracy

Test accuracy

Sensitivity

Specificity

AUC

Transport

75.0273

73.31

36.2

100

0.681

73.2636

72.19

33.6

100

0.668

Transcription

87.9723

88.15

99.3

77.2

0.882

92.5625

97.36

98.3

96.4

0.974

Translation

97.0864

97.34

62.5

100

0.813

98.8642

98.67

81.3

100

0.906

Gluconate utilisation

96.8288

96.22

71.4

100

0.857

98.7315

98.11

85.7

100

0.929

Amino acid biosynthesis

74.8627

77.83

42.6

100

0.713

73.5272

77.43

41.5

100

0.708

Fatty acid metabolism

92.4123

90.22

45.7

100

0.728

90.5586

87.77

32.1

100

0.66

Acetylcholine receptor inhibitor

98.448

99.06

80

100

0.9

100

99.53

100

99.5

0.998

G-protein coupled receptor

78.6998

80.87

48.3

100

0.741

76.1838

77.35

38.8

100

0.694

Guanine nucleotide-releasing factor

97.4359

97.92

77.1

99.6

0.883

98.8681

98.75

82.9

100

0.914

Fibre protein

96.789

95.89

0

100

0.5

99.8471

99.31

83.3

100

0.917

Transmembrane

85.2931

85.67

58.3

100

0.791

79.8937

80.58

43.5

100

0.717

  1. AUC: Area under the curve.