Skip to main content

Table 9 The encoding methods of input vectors in the fifteen selected SVMs.

From: Protein subcellular localization prediction for Gram-negative bacteria using amino acid subalphabets and a combination of multiple support vector machines

No.

Encoding methods of input vectors

1

1-gram with 2 partitioned parts

2

1-gram with 3 partitioned parts

3

1-gram with 4 partitioned parts

4

1-gram with 4 partitioned parts (apply feature selection to No. 3)

5

1-gram with 6 partitioned parts

6

2-gram without any gaps

7

2-gram without any gaps (apply feature selection to No. 6)

8

2-gram with one gap

9

3-gram with 6 merged groups

10

3-gram with 7 merged groups

11

3-gram with 8 merged groups

12

4-gram with 4 merged groups

13

4-gram with 4 merged groups

14

4-gram with 4 merged groups

15

4-gram with 4 merged groups (apply feature selection to No. 14)