Cross-Disciplinary Detection and Analysis of Large Network Motifs

The detection of motifs has recently become an important part of network analysis across all disciplines. In order to detect these network motifs, software such as FANMOD and MAVisto has been created. Although relatively quick in detecting small motifs, this software is comparatively slower when identifying large motifs, such as those of size six or seven. As a result of this time constraint, the discovery and analysis of large motifs in networks of all disciplines is virtually nonexistent. Using FANMOD, motifs were detected in biological, social, and other networks, ranging from motifs with three nodes to those with eight nodes. Topological analysis revealed that similar networks have similar small motifs, but as size increases differences arise. Three-node motifs are the same for almost all undirected networks, and are commonly found in the larger motifs of these networks. Significance profiles of common motifs showed similar low-level structure in multiple undirected networks as well, while the analysis of directed networks revealed both similarities and dissimilarities between networks of different disciplines.