Microsoft Word - 00-07-041.word

We analyze the structuture of a large metabolic network, that of the energy and biosynthesis metabolism of Escherichia coli. This network is a paradigmatic case for the large genetic and metabolic networks that functional genomics efforts are beginning to elucidate. To analyze the structure of networks involving hundreds or thousands of components by simple visual inspection is impossible, and a quantitative framework is needed to analyze them. We propose a graph theoretical description of the E. coli metabolic network, a description that we hope will prove useful for other genetic networks. We find that this network is a small world graph, a type of graph observed in a variety of seemingly unrelated areas, such as friendship networks in sociology, the structure of electrical power grids, and the nervous system of C. elegans. Moreover, its connectivity follows a power law, another unusual but by no means rare statistical distribution. This architecture may serve to minimize transition times between metabolic states, and also reflect the evolutionary history of metabolism. The information necessary to characterize the genetic and metabolic networks driving all functions of a living cell is being put within our reach by various genome projects. With the availability of this information, however, a problem so far unknown to molecular biologists will arise: how to adequately represent and describe the structure of large genetic networks. While it is trivial to understand the structure of metabolic pathways, transcriptional cascades, or signalling pathways that consist of a small number of genes, networks consisting of anywhere from hundreds to tens of thousands of components are less easily described. If networks encountered in areas ranging from neurobiology to sociology are any guide, we can assume that the structure of genetic and metabolic networks will be a complex mix of ordered and random elements. However, a body of theory suitable to analyze large networks exists only for networks at either end of this spectrum, networks that are perfectly orderly or completely random. Thus, a quantitative framework suitable to describe the structure of large biological networks remains to be developed. A description of genetic and metabolic networks is, however, only a first step towards a second, deeper level of inquiry, which asks for an explanation of the structure of such networks. Needles to say, any explanation of a network's organization, whether it invokes historical accidents or functional optimality principles requires a prior understanding of what that organization is. …