Permutation testing and agreement of Relevance Networks constructed based on Pearson or Bayesian correlations.

<p>(A) Empirical (blue) and permutation-based (red) distributions of Pearson correlations from each half of the data, comparing all pairs of miRNAs (solid lines) and comparing all pairs of high-entropy miRNAs (dashed lines). (B) Empirical (blue) and permutation-based (red) distributions of Bayesian correlations from each half of the data, comparing all pairs of miRNAs (solid lines) and comparing all pairs of high-entropy miRNAs (dashed lines). (C) Estimated false discovery rates (red, based on permutations) and empirical false discovery rates (blue, taking other half of the data as gold standard) at varying Pearson correlation thresholds. (D) Estimated false discovery rates (red, based on permutations) and empirical false discovery rates (blue, taking other half of the data as gold standard) at varying Bayesian correlation thresholds. (E) Difference between Pearson and Bayesian estimated and empirical false discovery rates.