Skip to main content

Table 3 Feature list and variable importance

From: Predicting the recurrence of noncoding regulatory mutations in cancer

Category

Mean decrease accuracy (%)

p < 5 × 10−6

Recurrence ≥ 4

Target Gene Features

HumanNet.CGS_L1

5.26

4.85

HumanNet.CGS_L2

6.78

6.42

InteractNet.CGS_L1

10.86

9.52

InteractNet.CGS_L2

10.16

10.84

RegNet.DEG_score

10.88

10.79

DEG_score

9.72

9.51

Binding TF Features

HumanNet.CGS_L1

4.91

2.92

HumanNet.CGS_L2

4.57

3.23

InteractNet.CGS_L1

2.67

2.66

InteractNet.CGS_L2

4.99

7.40

RegNet.DEG_score

6.68

3.61

DEG_score

4.77

6.13

TFBS Scores

diff_Log_Pval_FIMO

2.59

2.53

avg_Log_Pval_FIMO

8.04

7.48

Gain_or_Loss

1.57

0.37

Cancer Cell Epigenetic Signals

DnaseSig

17.34

16.94

H3K27ac

8.95

7.62

H3K27me3

11.94

10.23

H3K36me3

7.50

5.23

H3K4me3

9.45

7.99

H3K9me3

10.19

9.02

Genetic Signals

Distance.to.GWAS

15.82

11.00

Early.to.late_Rate

7.09

7.00

PhastCons

1.26

−0.18

Original Cell Epigenetic Signals

DnaseSig

17.90

17.05

H3K27ac

9.45

9.59

H3K27me3

5.91

7.26

H3K36me3

15.69

7.98

H3K4me1

11.50

9.69

H3K4me2

7.97

8.78

H3K4me3

7.88

7.02

H3K79me2

6.33

4.98

H3K9ac

7.79

8.79

H3K9me3

4.83

13.12

H4K20me1

8.38

7.44