Skip to main content

Table 4 Performance comparison with individual and combined feature encoding schemes for pT site prediction on the independent dataset

From: Predicting phosphorylation sites using machine learning by integrating the sequence, structure, and functional information of proteins

Attributes

Methods

Accuracy (%)

Sensitivity (%)

Specificity (%)

Precision (%)

F-measure (%)

MCC

AUC

Physiochemical property

RF

77.54

92.1

47.1

78.5

84.7

0.45

0.76

SVM

67.05

77.7

55.9

64.9

70.7

0.34

0.66

Structure

RF

89.58

94.8

78.8

90.3

92.5

0.75

0.96

SVM

77.66

99

33

75.6

85.7

0.47

0.66

Sequence

RF

71.79

86

42.1

75.7

80.5

0.31

0.69

SVM

73.74

94.4

69.5

70.3

79.9

0.16

0.57

Functional features

RF

72.75

92.1

32.3

74.0

82.1

0.31

0.63

SVM

72.43

93.0

29.3

73.4

82.0

0.29

0.61

Functional annotation

RF

68.5

92.6

18.1

70.3

79.9

0.16

0.57

SVM

68.34

92.3

31.7

70.3

79.8

0.15

0.55

Combined

RF

90.28

97.8

74.4

88.9

93.2

0.77

0.97

SVM

73.96

100

19.4

72.2

83.9

0.37

0.59

  1. Performance metrics for best results are highlighted in bold