Fig. 5

Comparison of the results between CPA in four semantic annotation categories, gene overlap counts, and module preservation. For the preservation analysis, we used the Z-summary metric, a composite preservation measure that identifies modules that are preserved in other networks based on their connectivity and density statistics. Significant module pairs for the gene wise overlap were computed the same way as for the annotation overlap using CPA (Methods). The figure shows the blood modules that were significantly associated with the caudate nucleus network. Each different method is depicted with a green diamond. The oval shapes with thick black lines represent modules significantly correlated with the disease phenotype. Only one module is shared by all three approaches (lightcyan1), which is not significantly correlated with the disease phenotype. The overall overlap between the methods is small