Measuring the expression conservation of components



To measure conservation in a sensitive way, we defined a co-expression interaction in a single organism using a loose Pearson correlation cutoff of 0.25.For every metagene m in every species s, we then collected the set of meta-genes where the correlation between the genes in the organism exceeded this cutoff.We then looked up which neighbors were also neighbors in the multiple species network.Let Igs, be the set of metagenes that gene g interacts with in species s.Let Im be the set of metagenes metagene m is connected to in the multiple species network.We computed a set-theoretic conservation measure for metagene m as:

Where Gms is the set of genes belonging to metagene m in organism s.Note that since the metagene mapping is nearly 1-to-1, Gms usually contains a single gene.The ECI measures the percent of a metageneís neighborhood that is present in the multiple species network.High ECIs correspond to those metagenes that have genes with co-expresssion neighborhoods enriched for interactions present in the multiple species network.For a component, we used the average ECI of its metagenes as a measure of the componentís degree of conservation.