Skip to main content

Table 5 Prediction accuracies for electrostatic energy and different feature selection methods.

From: The role of electrostatic energy in prediction of obligate protein-protein interactions

FS method

ZH-AA

ZH-AT

MW-AA

MW-AT

 

n

accuracy

n

accuracy

n

accuracy

n

accuracy

No FS

210

93.89%

171

95.42%

210

92.08%

171

96.30%

Chi2

8

97.69%

2

97.69%

10

91.09%

6

97.69%

GR

4

96.92%

8

96.92%

4

86.80%

6

97.69%

IG

8

97.69%

2

97.69%

8

88.78%

10

96.37%

mRMR

10

96.15%

10

97.69%

10

90.94%

10

96.10%

  1. Prediction accuracies of LDR for all subsets of features (ZH-AT, ZH-AA, MW-AT and MW-AA) using electrostatic energy values as the prediction properties and different feature selection (FS) methods; n indicates the number of features.