Skip to main content

Table 2 Performance comparison of KNN and SVM-L on unseen sequences from the six datasets, λ1=0.5

From: Optimal selection of molecular descriptors for antimicrobial peptides classification: an evolutionary feature weighting approach

Dataset

ML

Sens(%)

Spec(%)

Prec(%)

Bal Acc(%)

Acc(%)

MCC

AUC

DAMPD_AMP

KNN

72.16

9 4 . 1 7

6 8 . 6 3

83.17

9 0 . 8 7

0 . 6 5 0

0.832

 

SVM-L

7 7 . 3 2 a

91.62

61.98

8 4 . 4 7

89.47

0.631

0 . 8 4 5

APD3_AMP

KNN

70.82

9 2 . 1 1

6 5 . 1 0

81.47

8 8 . 4 5

0 . 6 0 9

0.815

 

SVM-L

8 9 . 2 4

82.87

51.98

8 6 . 0 5

83.97

0.597

0 . 8 6 1

DAMPD_ANTIBACTERIAL

KNN

8 0 . 0

90.91

60.27

8 5 . 4 5

89.30

0.634

0 . 8 5 5

 

SVML

74.55

9 3 . 1 0

6 5 . 0 8

83.82

9 0 . 3 7

0 . 6 4 0

0.838

APD3_ANTIBACTERIAL

KNN

65.97

9 3 . 9 1

6 8 . 3 5

79.94

89.26

0.607

0.799

 

SVM-L

8 1 . 9 4

91.55

65.92

8 6 . 7 5

8 9 . 9 5

0 . 6 7 6

0 . 8 6 7

DAMPD_BACTEROCIN

KNN

8 0

87.50

50.00

8 3 . 7 5

86.49

0.561

0 . 8 3 8

 

SVM-L

60

9 6 . 8 8

7 5 . 0 0

78.44

9 1 . 8 9

0 . 6 2 6

0.784

APD3_BACTEROCIN

KNN

75.86

9 4 . 2 3

70.97

85.05

91.35

0.682

0.850

 

SVM-L

9 3 . 1 0

92.95

7 1 . 0 5

9 3 . 0 3

9 2 . 9 7

0 . 7 7 4

0 . 9 3 0

  1. *Each value is the performance on the testing dataset by the classifier built by the machine learning algorithm (second column) on the dataset after applying the best compromise solution for λ1=0.5 (first column)
  2. aBold font indicates the best value per measure for every dataset