Fig. 3: Multiplex network modularity of rare disease groups.

a Pipeline for disease module significance assessment. The size of the largest connected component (LCC) for genes associated with rare genetic disease groups collected from Fig. 2 were used to determine network relevance. b The heatmap shows the modularity of all rare disease groups across all network layers as measured by the respective module size significance (p < 0.05: *, p < 0.01: **, p < 0.001: ***; p < 0.0001: ****, Benjamini–Hochberg corrected empirical p-values determined by node randomization, see Methods). In the tissue-specific network layers, only selected disease groups display pronounced modularity, often recapitulating known mechanisms and tissue specificities of particular rare diseases, but also revealing novel relationships. Network layers containing relationships that are relevant across biological levels of organization, such as protein–protein interaction, phenotypic and functional similarity networks, also display modularity across a wide range of disease groups.