Skip to main content

Table 4 Comparison of geneset differential expression analysis method performances, based on the simulation model proposed by M Ackerman

From: Functional Analysis: Evaluation of Response Intensities - Tailoring ANOVA for Lists of Expression Subsets

Ā Ā 

Cut-Off

Ā 

a2.fixed

faeri.

fixed.null

faeri.

fixed.perms

GSA.

mean*

GSA.

absmean*

GSA.

maxmean*

Globaltest.

asymptotic

Globaltest.

gamma

Globaltest.

permutations

Gsea.

pval

Gsea.

fdr

Samgs.

pval

Samgs.

qval

Ā Ā 

0.001

Set3

0

0

0

0

0

0

0

0

0

0

0

0

1

H 0

Ā 

0.01

Set3

1

0

0

0

0

0

0

0

0

0

0

0

2

Ā Ā 

0.1

Set3

3

3

3

12

0

0

2

5

3

2

1

2

27

Ā Ā 

0.001

Set2

100

96

97

15

25

12

54

96

95

69

20

91

94

Ā 

100% DE

0.01

Set2

100

99

99

15

25

12

95

99

97

99

77

98

99

Uni

Ā 

0.1

Set2

100

100

100

78

72

64

99

99

99

100

100

99

100

Ā Ā 

0.001

Set5

72

40

36

0

2

0

3

64

49

14

2

39

56

Ā 

50% DE

0.01

Set5

89

75

78

0

2

0

42

89

84

42

25

81

84

Ā Ā 

0.1

Set5

100

94

94

7

17

3

93

99

98

93

84

96

99

Ā Ā 

0.001

Set1

86

62

60

1

2

2

1

10

8

2

0

7

13

Ā 

100% DE

0.01

Set1

89

69

70

1

2

2

20

35

25

16

3

23

41

Uni+Cor

Ā 

0.1

Set1

95

80

80

11

23

11

58

59

58

46

47

56

73

Ā Ā 

0.001

Set4

55

43

39

0

0

0

0

15

7

4

3

4

11

Ā 

50% DE

0.01

Set4

65

52

54

0

0

0

18

30

24

13

14

23

43

Ā Ā 

0.1

Set4

81

65

65

1

13

1

56

60

57

43

47

56

73

Ā Ā 

0.001

Set7

0

100

100

1

67

40

100

100

100

1

0

100

100

Ā 

100% DE

0.01

Set7

0

100

100

1

67

40

100

100

100

10

1

100

100

Bidir

Ā 

0.1

Set7

6

100

100

22

98

82

100

100

100

69

32

100

100

Ā Ā 

0.001

Set9

0

84

82

0

7

2

32

95

93

0

0

86

93

Ā 

50% DE

0.01

Set9

0

95

95

0

7

2

93

100

100

1

0

100

100

Ā Ā 

0.1

Set9

9

100

100

16

48

15

100

100

100

27

8

100

100

Ā Ā 

0.001

Set6

0

84

84

0

11

8

5

28

17

6

1

15

25

Ā 

100% DE

0.01

Set6

0

88

89

0

11

8

39

50

43

17

8

43

61

Bidir+Cor

Ā 

0.1

Set6

2

93

93

35

49

54

79

79

78

57

42

78

88

Ā Ā 

0.001

Set8

0

63

62

0

2

1

3

30

14

0

0

13

23

Ā 

50% DE

0.01

Set8

0

75

76

0

2

1

36

50

42

3

0

40

58

Ā Ā 

0.1

Set8

3

86

86

28

27

12

75

78

75

31

15

76

85

  1. Each set of measurements presented was generated 100 times. The measurements in the table show, for each method, the number of detections of the different sets of measurements defined. The upper part of the table presents the results for unidirectional genesets only (all members over expressed) and the lower part concerns groups with a simulated mixed answer (over and under expression). The data related to H0 shows the number of genesets detected by chance (false positives).