Skip to main content

Table 3 Prediction accuracies of testing samples using different K.

From: Predicting the network of substrate-enzyme-product triads by combining compound similarity and functional domain composition

Representation of compound

K

Prediction accuracy for each class (%)

  

Networking triads (SN)

Non-networking triads (SP)

SMILES

3

92.67

92.03

 

5

89.79

92.92

Graph

3

95.34

99.48

 

5

94.18

99.48