Flux-Based vs. Topology-Based Similarity of Metabolic Genes

We present an effectively computable measure of functional gene similarity that is based on metabolic gene activity across a variety of growth media. We applied this measure to 750 genes comprising the metabolic network of the budding yeast. Comparing the in silico computed functional similarities to those obtained by using experimental expression data, we show that our computational method captures similarities beyond those that are obtained by the topological analysis of metabolic networks, thus revealing — at least in part — dynamic characteristics of gene function. We also suggest that network centrality partially explains functional centrality (i.e. the number of functionally highly similar genes) by reporting a significant correlation between the two. Finally, we find that functional similarities between topologically distant genes occur between genes with different GO annotations.