Skip to main content

Table 4 Number of false negatives for each method

From: Protein complex prediction via verifying and reconstructing the topology of domain-domain interactions

 

Number of false negatives of our method

Number of false negatives of existing methods

α:Number of "net" false negatives caused by the assumption *

β:Number of true positives of existing methods

Ratio (α/β)

Clustering coefficient verified with DDI (A)

683

514

102

249

41.0%

MCL verified with DDI (A)

726

554

76

209

36.4%

MCODE verified with DDI (A)

739

636

58

127

45.7%

Clustering coefficient verified with DDI (A + B)

636

514

76

249

30.5%

MCL verified with DDI (A + B)

722

554

85

209

40.7%

MCODE verified with DDI (A + B)

714

636

42

127

33.1%

Clustering coefficient verified with DDI (A + C)

590

514

51

249

20.5%

MCL verified with DDI (A + C)

707

554

78

209

37.3%

MCODE verified with DDI (A + C)

689

636

27

127

21.3%

  1. *Number of "net" false negatives indicates known protein complexes which were predicted by existing methods but were missed by our method, since our method works as a filter for existing methods. Because some known complexes were missed by our method for the lack of DDI annotations, we calculated number of net false negatives caused by our assumption, which indicates the number of the net false negatives that have DDI annotations (See text).