Skip to main content

Table 1 Prediction accuracies of our method for three datasets and comparison with other reported results.

From: Prediction of protein structural classes for low-homology sequences based on predicted secondary structure

Dataset

Reference

Prediction accuracy (%)

 
  

α

β

α + β

α/β

Overall

25PDB

[6]

69.1

61.6

60.1

38.3

57.1

 

[22]

60.6

60.7

44.3

67.9

58.6

 

[5]

NA

NA

NA

NA

59.9

 

[15]

77.4

66.4

45.4

61.3

62.7

 

[17]

64.3

65.0

61.7

65.0

64.0

 

[16]

93.7

81.3

71.7

73.1

80.3

 

This paper

92.8

83.3

70.1

85.8

82.9

640

[21]

73.9

61.0

33.9

81.9

62.3

 

[16]

90.6

81.8

66.7

85.9

80.8

 

This paper

89.1

85.1

71.4

88.1

83.1

1189

[7]

NA

NA

NA

NA

53.8

 

[6]

57.0

62.9

25.3

64.6

53.9

 

[23]

NA

NA

NA

NA

54.7

 

[14]

48.9

59.5

26.6

81.7

56.9

 

[5]

NA

NA

NA

NA

58.9

 

[17]

62.3

67.7

63.1

66.5

65.2

 

[21]

75.8

75.2

31.8

82.6

67.6

 

[16]

87.4

84.7

53.1

84.7

78.3

 

This paper

89.2

86.7

65.6

82.6

81.3

  1. The accuracies are evaluated by jackknife test and measured by the percentage of correctly predicted Proteins. The best results are highlighted in bold face.