Skip to main content

Table 6 Comparative results of SSF-DDI on the TWOSIDES dataset in transductive setting (%)

From: SSF-DDI: a deep learning method utilizing drug sequence and substructure features for drug–drug interaction prediction

Method

ACC

AUC

F1

Prec

Rec

AP

CNN-DDI

85.75

92.16

86.67

81.39

92.68

89.5

MR-GNN

85.39

91.93

86.46

80.57

93.28

89.32

SSI-DDI

82.21

89.27

83.11

79.1

87.56

86.19

GAT-DDI

67.32

75.16

63.7

71.54

57.62

72.5

GMPNN-CS

86.96

92.94

87.85

82.2

94.35

90.38

DGNN-DDI

85.29

91.92

86.12

81.51

91.28

89.41

SSF-DDI (ours)

87.3

93.09

88.17

82.48

94.37

90.47

  1. The best results are highlighted in bold