Reaction centric layout for metabolic networks

Abstract

The challenge is to understand and visualize large networks with many elements while at the same time give the biologist enough of an overview of the whole network to see and understand relationships between subnets. Classical biological visualization systems like Cytoscape use standard graph layout algorithms based on the molecules (DNA, RNA, proteins, metabolites) involved in the processes. We propose a new approach that instead focuses on the reactions that transform molecules, using higher-level macro-glyphs that summarize a large number of molecules in a compact unit, thus forming very natural and automatic clusters. We also employ natural clustering approaches to other areas of typical metabolic networks. The result is a graph with about 50-60% of the original node and 20-30% of the original edge count, which simplifies efficient layout and interaction significantly. © 2009 Springer-Verlag.