Skip to main content

Table 1 Statistics of the prediction performance

From: Semi-supervised drug-protein interaction prediction from heterogeneous biological spaces

Data

Methods

AUC

Sensitivity(%)

Specificity(%)

* Enzyme

Combining weighted profile

92.2

6

99.9

 

LapRLS

95.0

53

99.9

 

NetLapRLS

98.3

75

99.9

* Ion channel

Combining weighted profile

90.7

17

99.7

 

LapRLS

96.1

36

99.8

 

NetLapRLS

98.6

72

99.9

* GPCR

Combining weighted profile

86.9

13

99.7

 

LapRLS

93.4

24

99.8

 

NetLapRLS

97.1

50

99.8

Nuclear receptor

Combining weighted profile

81.0

11

99.4

 

LapRLS

85.0

16

99.4

 

NetLapRLS

88.8

21

99.5

  1. The AUC is the area under the ROC curve, normalized to 100. The cutoff for sensitivity and specificity is set to select the number of the interactions in the test data.