Statistical analysis of co-expression properties of sets of genes in the mouse brain

We propose a quantitative method to estimate the statistical properties of sets of genes for which expression data are available and co-registered to a reference atlas of the brain. It is based on graph-theoretic properties of co-expression coefficients between pairs of genes. We apply this method to mouse genes from the Allen Gene Expression Atlas. Co-expression patterns of a list of several hundreds of genes related to addiction are analyzed, using ISH data produced for the mouse brain at the Allen Institute. It appears that large subsets of this set of genes are much more highly co-expressed than expected by chance.