Table 2 Performance of individual features.

From: Prioritization Of Nonsynonymous Single Nucleotide Variants For Exome Sequencing Studies Via Integrative Learning On Multiple Genomic Data

 

Neutral

Disease

Combined

MRR(%)

AUC(%)

MRR(%)

AUC(%)

MRR(%)

AUC(%)

SIFT

21.56

78.44

45.14

54.86

33.29

66.71

PolyPhen2

15.48

84.52

47.56

52.44

29.76

70.24

LRT

23.10

76.90

47.87

52.13

35.89

64.11

MutationTaster

27.43

72.57

53.06

46.94

40.42

59.58

MutationAccessor

17.59

82.41

52.13

47.87

32.99

67.01

GERP

23.73

76.27

47.19

52.81

35.62

64.38

Phylop

23.22

76.78

46.44

53.56

34.99

65.01

Siphy

21.14

78.86

46.77

53.23

34.15

65.85

MSRV

18.94

81.06

52.33

47.67

33.74

66.26

SInBad

17.16

82.84

53.00

47.00

35.33

64.67

CADD

15.29

84.71

54.49

45.51

35.16

64.84

Expression

34.07

65.93

37.13

62.87

35.63

64.37

GO

37.03

62.97

36.66

63.34

36.80

63.20

KEGG

23.38

76.62

35.46

64.54

29.60

70.40

miRNA

35.31

64.69

36.17

63.83

35.75

64.25

Pfam

31.04

68.96

31.43

68.57

31.24

68.76

Sequence

40.90

59.10

41.95

58.05

41.42

58.58

PPI

19.78

80.22

25.55

74.45

22.94

77.06

TSFC

33.81

66.19

35.25

64.75

34.54

65.46

All (snvForest)

3.37

96.60

15.97

84.00

9.76

90.23

  1. Variant features are effective in distinguishing disease variants from neutral ones but are ineffectiveness in discriminating between disease variants. Gene features are medium effective in distinguishing disease variants from both neutral and disease controls. The integration of all the features yields much higher performance than any individual feature.
Back to article page