Skip to main content

Table 1 Performance (mean ± std.%) comparison among different methods on the simulated gene expression data

From: Adaptively capturing the heterogeneity of expression for cancer biomarker identification

 

Sensitivity

Specificity

AUC

ACC

n = 6

Rankprod

33.24 ± 1.35

89.49 ± 0.91

70.11 ± 2.24

67.79 ± 0.94

Limma

39.73 ± 3.07

95.01 ± 1.99

78.54 ± 3.18

72.9 ± 2.59

SAM

32.95 ± 0.07

82.36 ± 6.68

70.02 ± 5.14

65.4 ± 4

GRP0.5

29.92 ± 2.13

96.85 ± 1.07

78.48 ± 3.04

69.08 ± 1.61

GRP0.7

40.97 ± 0.05

94.06 ± 3.47

78.61 ± 2.67

71.73 ± 3.07

GRP0.9

42.99 ± 0.02

92.86 ± 1.03

77.98 ± 3.35

70.11 ± 3.62

aGRP

43.45 ± 4.3

93.16 ± 0.85

80.08 ± 2.98

73.63 ± 2.51

n = 10

Rankprod

56.96 ± 1.34

85.48 ± 0.31

73.22 ± 0.85

73.27 ± 0.57

Limma

57.04 ± 3.03

95.49 ± 1.28

88.32 ± 2.92

80.17 ± 1.77

SAM

51.08 ± 3.05

77.9 ± 5.75

70.73 ± 4.56

68.73 ± 3.45

GRP0.5

47.05 ± 3.59

95.34 ± 1.65

85.42 ± 2.87

76.7 ± 0.99

GRP0.7

51.35 ± 3.58

95.16 ± 1.68

85.85 ± 2.98

77.89 ± 1.21

GRP0.9

51.01 ± 4.09

96.35 ± 1.18

85.87 ± 1.66

77.81 ± 1.71

aGRP

56.47 ± 3.4

96.16 ± 1.06

87.36 ± 2.67

79.7 ± 1.64

n = 20

Rankprod

56.51 ± 1.29

85.4 ± 0.31

78.03 ± 0.92

73.84 ± 0.54

Limma

86.84 ± 1.01

95.30 ± 1.61

96.02 ± 0.43

91.06 ± 0.37

SAM

85.37 ± 0.1

92.45 ± 5.56

90.12 ± 3.73

86.46 ± 3.31

GRP0.5

80.5 ± 0.99

95.92 ± 0.92

94.00 ± 1.03

89.65 ± 0.87

GRP0.7

80.81 ± 1.58

96.28 ± 0.73

95.74 ± 0.85

89.97 ± 0.98

GRP0.9

80.69 ± 1.88

96.21 ± 1.02

94.43 ± 1.02

90.13 ± 0.85

aGRP

86.4 ± 1.7

95.70 ± 0.57

95.85 ± 0.5

91.75 ± 0.94

n = 50

Rankprod

69.93 ± 0.69

80.07 ± 1.08

83.43 ± 0.92

76.08 ± 0.58

Limma

98.94 ± 3.9

95.95 ± 0.73

99.76 ± 1.01

96.57 ± 0.44

SAM

92.97 ± 0

89.36 ± 2.85

88.35 ± 1.51

90.82 ± 1.71

GRP0.5

97.16 ± 0.90

95.82 ± 1.01

99.51 ± 0.27

96.37 ± 0.25

GRP0.7

98.39 ± 0.47

95.43 ± 0.73

99.73 ± 0.16

96.56 ± 0.34

GRP0.9

97.06 ± 1.09

95.36 ± 1.04

99.54 ± 0.15

96.08 ± 0.92

aGRP

98.96 ± 3.4

97.3 ± 0.85

99.85 ± 0.08

98.78 ± 0.51

  1. Best values are in bold