Gene networks from DNA microarray data: centrality and lethality

We construct a gene network based on expression data from DNA microarray experiments, by establishing a link between two genes whenever the Pearson’s correlation coefficient between their expression profiles is higher than a certain cutoff. The resulting connectivity distribution is compatible with a power-law decay with exponent γ ∼ 1, corrected by an exponential cutoff at large connectivity. The biological relevance of such network is demonstrated by showing that there is a strong statistical correlation between high connectivity number and lethality: in close analogy to what happens for protein interaction networks, essential genes are strongly overerpresented among the hubs of the network, that is the genes connected to many other genes. DNA microarray experiments are one of the most powerful tools for studying interactions between genes on the scale of the whole genome. It is widely believed that a huge amount of biologically relevant information is encoded in the results of such experiments, and that new analytical methods need to be developed to extract it. In this work we propose to analyse the expression data obtained in microarray experiments by constructing a network of coregulated genes: the genes are the nodes of the network, and a link is established between two nodes whenever they are similarly expressed across many experimental conditions. On one hand, we show that such network, like many other known networks of self-organizing origin, shows a connectivity distribution that decays with a power law corrected, for large values of the connectivity, by an exponential cutoff [1, 2]. On the other hand, we show that the network encodes biologically relevant information in