Skip to main content

Table 2 Cross-validation (CV) accuracies of feature sets examined

From: Differential analysis of combinatorial protein complexes with CompleXChange

feature set

set size

CV accuracy [%]

sign. dereg. complexes (q<0.05)

978

96.4±1.5

random complexes

978

86.6±2.2

random complexes (filtered)

978

94.2±0.7

sign. dereg. complexes (q<0.01)

429

96.6±1.1

random complexes

429

84.3±3.4

random complexes (filtered)

429

93.5±1.2

sign. dereg. complexes (q<0.001)

31

97.0±0.4

random complexes

31

66.8±9.9

random complexes (filtered)

31

88.4±3.9

sign. dereg. CORUM complexes

77

93.6±2.6

sign. dereg. hu.MAP complexes

16

85.5±3.9

DE complexes

1841

96.3±1.5

DE TFs

77

96.2±1.4

all TFs

601

94.9±2.0

  1. For the randomized complexes CV accuracy is reported as the mean across all permutations and its standard deviation. For all other evaluated sets CV accuracy depicts the mean and variance for the 100 iterations of CV