Table 1 Valuable attributes from random-forest and decision tree classifiers.

From: Computational B-cell epitope identification and production of neutralizing murine antibodies against Atroxlysin-I

Attribute

Decreased Gini

Decreased Accuracy

Positive charged RHK

75,28

37.50*

Negative charged DE

79,12

52,71

Uncharged STNQ

101,78

63,27

Special CGP

100,9

43.43*

Hydrophobic AVILMFW

92,84

43.16*

gravy

198,59

60,82

Aliphatic index (5)

118,75

45.49*

% Atoms of C

234,22

76,8

% Atoms of H

212,17

58,34

% Atoms of N

233,88

59,07

% Atoms of O (12)

198,23

68,39

% Atoms of S (1)

166,32

67,65

% Arg (4)

72,98

39.25*

% His (2)

68.85*

53,73

% Lys (3)

36.46*

42.84*

% Asp

80,68

56,97

% Glu (6)

66.12*

52,07

% Serine

107,13

76,88

% Thr (11)

94,8

73,62

% Asn (9)

74,87

47.39*

% Gln

73,67

55,23

% Cys

77,22

45.50*

% Gly

98,59

52,81

% Pro(10)

58.36*

39.87*

% Ala

86,08

65,12

% Val

48.48*

38.31*

% Ile

68.87*

40.67*

% Leu

74,26

42.55*

% Met

87,91

61,82

% Phe

48.65*

39.62*

% Tyr

64.44*

56,46

% Trp(8)

57.03*

51,84

Isoelectric point (7)

183,09

54,9

  1. The attributes in bold represent the first nodes of the decision tree and (*) indicate the best attributes from random-forest.
Back to article page