Figure 3

Histograms of log odds ratios in the simulation studies. All edges (panel (a) with N = 200 for both data sets and panel (c) for unbalanced samples (N = 200/50)) and true positive edges (i.e. truly differentially connected between disease states, panel (b) for N = 200 and (d) for unbalanced samples). Red shaded results in panel (b) and (d) denote truly differentially connected edges that are not detected at cutoff of logOR = -40 (i.e. false negatives).