Skip to main content

Table 3 Twenty-three sequence-derived features

From: A genetic algorithm-based weighted ensemble method for predicting transposon-derived piRNAs

Index

Feature

Dimension

Parameter

Annotation

F1

1-Spectrum Profile

4

No Parameters

Used in [20]

F2

2-Spectrum Profile

16

No Parameters

Used in [20]

F3

3-Spectrum Profile

64

No Parameters

Used in [20]

F4

4-Spectrum Profile

256

No Parameters

Used in [20]

F5

5-Spectrum Profile

1024

No Parameters

Used in [20]

F6

(3, m)-mismatch profile

64

m: the max mismatches

New features

F7

(4, m)-mismatch profile

256

m: the max mismatches

New features

F8

(5, m)-mismatch profile

1024

m: the max mismatches

New features

F9

(3, w)-subsequence profile

64

w: penalty for the non-contiguous matching

New features

F10

(4, w)-subsequence profile

256

w: penalty for the non-contiguous matching

New features

F11

(5, w)-subsequence profile

1024

w: penalty for the non-contiguous matching

New features

F12

1-RevcKmer

2

No Parameters

New features

F13

2-RevcKmer

10

No Parameters

New features

F14

3-RevcKmer

32

No Parameters

New features

F15

4-RevcKmer

136

No Parameters

New features

F16

5-RevcKmer

528

No Parameters

New features

F17

PCPseDNC

16 + λ

λ: the highest counted rank of the correlation

New features

F18

PCPseTNC

64 + λ

λ: the highest counted rank of the correlation

New features

F19

SCPseDNC

16 + 6 × λ

λ: the highest counted rank of the correlation

New features

F20

SCPseTNC

64 + 12 × λ

λ: the highest counted rank of the correlation

New features

F21

Sparse Profile

5 × d

d: the fixed length of sequences

New features

F22

PSSM

d

d: the fixed length of sequences

New features

F23

LSSTE

32

No parameters

Used in [21]