Functional meta-analysis of double connectivity in gene coexpression networks in mammals

Abstract

In functional genomics, the high-throughput methods such as microarrays 1) allow analysis of the relationships between genes considering them as elements of a network and 2) lead to biological interpretations thanks to Gene Ontology. But up to now it has not been possible to find relationships between the functions and the connectivity of the genes in coexpression networks. To achieve this aim, we have defined a double connectivity for each gene by the numbers of its significant negative and positive correlations with the other genes within a given biological condition, or group. Here, based on the analysis of 1,260 DNA microarrays, we show that this double connectivity clearly separates two types of genes, those with a predominantly strong negative connectivity, hub- genes, and those with a predominantly strong positive connectivity, hub+ genes. Interestingly, the hub+ genes concerned transcription factors more often than by chance and, similarly, for the hub- genes concerning miRNA predicted targets. Furthermore, a meta-analysis of GO annotations carried out on 67 groups in humans and rats shows that these two types of genes correspond to a functional biological duality. The hub- genes were mainly involved in basic functions common to all eukaryote cells, whereas the hub+ genes were mainly involved in specialized functions related to cell differentiation and communication. The separation and the biological role of these hub- and hub+ genes provide a powerful new tool for a better understanding of the control and regulation of the key genes involved in cellular differentiation and physiopathological conditions.