Background: In the big data era, it is of interests to mine large volume of genome-wide expression profiles to identify biologically meaningful patterns.Methods: Therefore, in this study, a gene correlation network was constructed upon 61 samples from three human liver expression datasets by calculating gene-gene expression correlation coefficient.Weighted correlation network analysis (WGCNA) algorithm was then applied to the gene network to detect modules (or sub-networks) of highly correlated genes.To functionally characterize these network modules, a series of statistical analyses were demonstrated.