The existence of commonly observed coexpression edges (Supplementary Data) is a naive method to evaluate the non-randomness of each coexpressed gene pair. However, this method causes several problems in the statistical evaluation of the coexpression quality as follows; (i) Since the network edges in COXPRESdb are restricted to the top three coexpressed gene relationships from every guide gene for simplicity, they may reflect only a part of the gene coexpression relationships. (ii) Each platform includes different number of genes, and thus statistical significant of top k gene can be different (k = 3 was employed as the network edge). (iii) The all possible reference platforms should be compared to select the best one.
