Figure 4

Results: pattern correlations. Comparison of the expression pattern correlations. (A) Shown are the distributions of correlation coefficients ρ between host and target gene expression patterns (blue) of Cluster NO I and correlation coefficients ρ between the same host genes and sampled target genes (red). The medians are illustrated by blue and red lines, respectively. Δ _m indicates the difference between the two medians. A missing relation between host and target gene expression would result in Δ _m = 0. The distributions of Δ _m taken over all significant clusters of the three datasets are shown in the two histograms for TargetScan (B) and our consensus model (C). Missing distances of Δ _m = 0 in both histograms indicate that all significant clusters deviate from the null model (sampled data). Both histograms show distributions with two maxima, indicating that positive (green) and negative (orange) correlations are approximately equally distributed over all analyzed clusters.