Skip to main content

Table 3 Performance comparison with individual and combined feature encoding schemes for pS 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

71.5

79

62.3

72.1

75.4

0.42

0.74

SVM

64.31

79.9

45.0

64.2

71.2

0.26

0.62

Structure

RF

87.34

86.5

80.6

90.2

88.3

0.74

0.94

SVM

70.12

96.4

37.7

65.6

78.1

0.43

0.67

Sequence

RF

70.3

80.1

58.2

70.3

74.9

0.39

0.73

SVM

77.65

99.1

51.2

77.5

83.1

0.59

0.75

Functional features

RF

62.87

93.9

24.6

60.6

73.6

0.26

0.59

SVM

62.58

93.4

24.6

60.5

73.4

0.25

0.59

Functional annotation

RF

62.75

90.7

28.2

60.9

72.9

0.24

0.60

SVM

62.50

93.6

24.1

60.4

73.4

0.25

0.58

Combined

RF

89.16

89.4

88.9

90.8

90.1

0.78

0.95

SVM

88.50

79.9

99.1

99.1

88.5

0.79

0.89

  1. Performance metrics for best results are highlighted in bold