Analysis of the Bipartite Networks of Domain Compositions and Metabolic Reactions

It is widely accepted that complexity of biological systems arises from combinations of common subunits. In this work we investigate the combinatorial patterns of protein domains in the metabolic networks and find several general rules in the patterns of domain combinations and their evolution. First, the reactions catalyzed by a domain subunit carrying specialized or accessory functions are often subsumed to the reactions catalyzed by a domain subunit carrying generic operations. Second, some reactions contain multiple domains in their enzymes because they require multiple chemical operations carried by distinct domains. Third, pleiotropy (multi-functionality) of enzymes either results from the similarity of the catalyzed reactions or is achieved by merging domains with distinct functions. Fourth, comparison of domain compositions and metabolic reactions between human and Escherichia coli suggests that requirements for novel reactions, redundancy and pleiotropy are the dominant driving factors for domain evolution. The methods and results provide a framework to study the combinatorial complexity of a biological system.