Skip to main content

Table 2 Results of the feature and classifier selection for the considered seven structural classes.

From: Modular prediction of protein structural classes from sequences of twilight-zone identity with predicting sequences

Class

Kernel

C

Feature selection method

# of selected features

all-α

RBF (γ = 0.05)

10

Wrapper with SVM

117

all-β

RBF (γ = 0.1)

7

Wrapper with NB

53

α/β

Polynomial (exp = 2)

2

ReliefF

46

α+β

RBF (γ = 0.15)

4

CFS

163

Multi-domain

Polynomial (exp = 1.5)

0.5

Wrapper with SVM

105

Membrane

Polynomial (exp = 1.5)

10

Symmetrical Uncertainty

46

Small

Polynomial (exp = 2.5)

15

Wrapper with NB

18