Skip to main content

Table 1 Performance comparison among different tools on the test set. Performance of different tools are presented with five metrics in percentage: accuracy (acc), sensitivity (sens), specificity (spec), F1 score (F1) and area under the receiver operating characteristic curve (AUROC)

From: AMPlify: attentive deep learning model for discovery of novel antimicrobial peptides effective against WHO priority pathogens

Tool

Model

Acc

Sens

Spec

F1

AUROC

iAMPpred

originala

74.01

87.90

60.12

77.18

80.70

iAMP-2L

originala

77.96

88.26

67.66

80.02

–

AMP Scanner Vr.2

originala

78.50

90.66

66.35

80.83

88.33

re-trained, 10 epochsb

90.66

91.14

90.18

90.70

97.40

re-trained, early stoppedc

91.20

90.42

91.98

91.13

97.03

AMPlify

single sub-model 1

92.40

90.90

93.89

92.28

97.54

single sub-model 2

91.98

91.02

92.93

91.90

97.40

single sub-model 3

92.51

92.69

92.34

92.53

97.82

single sub-model 4

92.10

90.90

93.29

92.00

97.27

single sub-model 5

92.57

92.57

92.57

92.57

97.98

ensemble

93.71

92.93

94.49

93.66

98.37

  1. aModels presented in the referenced papers are available through online servers
  2. bThe best hyperparameter as stated in the referenced paper
  3. cThe optimal number of training epochs determined by early stopping is 16