Skip to main content

Table 2 Proportion of p-values being exactly zero for E. coli and Yeast, Pearson correlation results

From: A comparative evaluation of data-merging and meta-analysis methods for reconstructing gene-gene interactions

 

E. coli

Yeast

Rank-Product

0 %

0 %

FR-Effects

0 %

0 %

Random-Effects

0 %

0 %

Fixed-Effects

0 %

0 %

Fisher

32.5 %

19.1 %

Stouffer

9.5 %

11.7 %

SVA

0 %

0 %

RMA-Combat

9.0 %

2.5 %

Combat

9.4 %

2.8 %

Scaling

9.1 %

2.7 %

RMA

15.6 %

13.4 %

No-Correction

85.6 %

98.7 %

Random-Guess

0 %

0 %

  1. The majority of DM methods assigns a zero p-value to some percentage of the predictions, while only the Fisher and Stouffer MA methods do so. These percentages are higher in E. coli than in Yeast, suggesting that in the first compendium the statistical associations are stronger or more detectable due to higher statistical power