Skip to main content

Table 4 Basic network properties and the limiting effects of using induced subgraphs for novel link prediction, for each investigated species

From: Efficient link prediction in the protein–protein interaction network using topological information in a generative adversarial network machine learning model

Species

Number of proteins (nodes)

Number of interactions (links)

Network sparsity with self-loops included (%)

Induced subgraph node coverage on N90 (%)

Induced subgraph node coverage on N100 (%)

Known missing links that could not be predicted in N100 from N90 (%)

Homo sapiens

3555

16,586

0.26

67.2

69.8

12.3

Saccharomyces cerevisiae

2052

16,635

0.79

77.8

79.3

6.5

Mus musculus

3868

16,548

0.22

64.0

66.5

14.7

Rattus norvegicus

4053

22,249

0.27

68.5

71.2

10.1

Sus scrofa

3731

14,637

0.21

68.9

71.5

15.3

Mean

3452

17,331

0.28

69.3

71.7

11.8