The use of independent studies to increase specificity in network determination.

SynTReN was used to build a synthetic dataset of 400 samples, these were randomly subdivided into two subsets of 200 each. The discretization-based co-expression networks were calculated for each and the shared edges used to give a third network. The 10% of the genes with the lowest variance were selected and the possible gene-pairs for those determined, all of these genes were not defined by ac, du or re relationships. The low-variance based gene-pairs detected are preferentially discarded by this procedure, suggesting that this is one reasonable technique for discarding false relationships.