Table 1

The feature selection results of the filter methods

IG

CHI

ReliefF

CFS


Structural features

200

300

200

300

200

300

200

300


A-philicity

2

5

2

5

5

4.7

7.9

8.8

B-DNA twist

0

0

0

0

2

2.7

6.2

4.7

Bendability

2

3

2

3

6

6.7

8.4

7.8

Stacking energy

3

3

3

3

7

7.7

5.7

9.4

DNA-bending stiffness

19.5

17.3

20

17.3

13

12

17.6

9.4

DNA denaturation

11

14.7

10.5

14.7

9

9.7

18.9

10.1

Duplex free energy

21.5

20

22

20

15

15.7

11

8.5

Duplex disrupt energy

23.5

17.7

23

17.7

17.5

16.3

5.7

5

Nucleosome position

2

2.3

2

2.3

3

4

2.6

7.5

Propeller twist

6

8

6

7.7

5.5

5.3

5.3

9.1

Protein deformation

4

4

4

4

6.5

5.7

3.1

7.5

Protein-DNA twist

2

1.3

2

1.3

3.5

4

4.4

4.7

Z-DNA

3.5

3.7

3.5

4

7

5.7

3.1

7.5


The distribution of each feature in the dimension subset selected by different filter methods. The percentage of each feature in a selected subset is calculated as follows: (the number of dimensions related to the feature/the total number of dimensions in the selected dimension subset)*100%.

Gan et al. BMC Bioinformatics 2012 13:4   doi:10.1186/1471-2105-13-4

Open Data