Global Alignment of Protein-Protein Interaction Networks for Analyzing Evolutionary Changes of Network Frameworks

Recent technological advances have yielded protein-protein interaction (PPI) networks across multiple species. To interpret these networks, comparison methods to characterize interactions across species have gained importance. However, the methodologies of most such studies have been limited to the relationships of protein complexes or pathways between different species because methodologies for network comparisons are limited. In this study, we introduce a novel comparison problem of biological networks, focusing on the framework structure of the networks, and compare these structures to find changes and conservations in the smaller networks. For this purpose, we define an alignment score between two networks based on the validity of the framework structure in each species and on the conservation of homologous genes in the networks. We propose an algorithm to find a high-score alignment based on k-means clustering. Experiments using D. melanogaster and C. elegans PPI networks show that our algorithm identified the network conservation on genes belonging to cancer-related pathways.