Yet, the identification of modules and standardised interaction routes remains a non-trivial problem.
The identification of modules in complex networks has attracted much attention of the scientific community in the last years, and social networks constitute a prominent example.
While community detection in unipartite networks has been extensively studied in the past decade, identification of modules or communities in bipartite networks is still in its early stage.
This extension allows the identification of module-based prognostic markers, rather than gene-based markers.
Considering that the subnetwork marker extracted from protein interaction databases are more reproducible than individual gene markers [15], [16], we assume that the identification of module-module interactions is more robust than that of gene-gene interactions.
We introduce a novel approach for the identification of modules based on the persistence of isolated gene groups within an evolving graph process.
For the identification of modules within a dendrogram, we consider both global cuts, including the one suggested by the gap statistic [17], and local cuts.
This approach for the detection of module isolation, which we refer to as the Miso method, permits the identification of modules with differing internal cohesion and determines the statistical significance of each candidate module.
Moreover, the identification of modules as hills of the community landscape is a new approach, including traditional threshold-based algorithms and the novel local maxima based algorithms (as our ProportionaHill, GradientHill and TotalHill methods).
Because the detection and identification of modules is a notoriously difficult task [7], a different approach focuses on the identification of conserved network motifs [8] [10], that is, sub-networks of small size (typically two to five nodes) that are significantly more abundant in a network compared to a control network that had its edges randomly rearranged.
