Skip to main content

Table 3 The effect of IP and RNA features on prediction performance.

From: Prediction of RNA-binding amino acids from protein and RNA sequences

Features

Sn (%)

Sp (%)

Pr (%)

Ac (%)

NP (%)

Fm (%)

CC

sIP

41.3

70.9

13.2

68.1

56.1

52.2

0.08

prev_tIP

57.0

83.5

27.0

80.9

70.3

67.8

0.29

tIP

57.4

83.8

27.5

81.3

70.6

68.2

0.30

sIP + R

57.3

83.1

26.7

80.6

70.2

67.8

0.29

prev_tIP + R

61.5

85.1

30.7

82.8

73.3

71.4

0.34

tIP + R

61.4

85.2

30.8

82.9

73.3

71.4

0.35

  1. The non-redundant dataset with a window size=15 was encoded into the feature vectors with different interaction propensities separately or with the RNA feature (R) in combination. The prediction performance was evaluated by 5-fold cross validation. sIP: the interaction propensity of single amino acids. prev_tIP: the previous version of the interaction propensity of amino acid triplets. tIP: the modified interaction propensity of amino acid triplets. Sn: sensitivity. Sp: specificity. Pr: precision rate. Ac: accuracy. NP: net prediction. Fm: F-measure. CC: correlation coefficient.