Skip to main content

Table 3 Overall performance of 11 methods

From: Gsw-fi: a GLM model incorporating shrinkage and double-weighted strategies for identifying cancer driver genes with functional impact

Methods

sd of driver gene number

CGC overlap (%)

Mut-driver overlap (%)

HiConf overlap (%)

Method consensus (%)

CGC rank

Mut-driver rank

HiConf rank

Method consensus rank

Average rank

Dendrix

38.78

29.22

25.04

21.49

39.83

8

8

7

7

7.50

MutSigCV

106.26

37.05

31.73

26.48

66.72

5

4

4

3

4.00

MEMo

3.02

17.07

16.07

15.61

17.71

9

15

9

10

10.75

DriverNet

44.02

39.41

33.51

32.37

36.76

3

3

3

8

4.25

e-Driver

42.73

36.07

29.57

25.00

61.80

6

5

5

4

5.00

iPAC

3662.54

11.35

5.31

3.55

7.2

11

11

11

11

11.00

MSEA

355.44

13.35

7.83

5.82

14.88

10

10

10

9

9.75

DriverML

251.56

48.34

41.48

34.00

73.99

2

2

2

1

1.75

OncodriveFML

69.71

33.93

26.18

19.17

53.55

7

7

8

5

6.75

rDriver

9.00

38.18

27.55

21.80

46.40

4

6

6

6

5.50

GSW-FI (\(\lambda =0\))

9.36

63.26

45.44

36.14

GSW-FI (\(\lambda =0.5\))

12.24

65.13

49.01

41.10

67.07

1

1

1

2

1.25

GSW-FI (\(\lambda =1\))

18.25

53.86

41.41

34.50

  1. \(\bullet\) “sd of driver gene number” is the standard deviation of the identified driver gene number across cancer types