The　ultimate　goal　of　genomics　research　is　to　describe　the　network　of　molecules　and　interactions　that　govern　all　biological　functions　and　disease　processes　in　cells.　Nonlinear　interactions　among　genes　in　terms　of　their　logic　relationships　play　a　key　role　for　deciphering　the　networks　of　molecules　that　underlie　cellular　function.　We　present　a　method　based　on　a　graph　coloring　scheme　and　information　theory　to　identify　the　gene　expression　network　with　lower　and　higher　order　logic　interactions　of　genes.　The　analysis　of　oncogenes　and　suppressor　genes　from　a　colon　cancer　mRNA　microarray　dataset　identifies　a　gene　expression　network　with　directionality　and　weights　that　reflects　intracellular　communication　pathways.　The　success　of　the　proposed　method　in　mining　hidden,　complicated　gene　interactions　and　reliably　interpreting　experimental　results　suggests　that　the　proposed　method　is　a　useful　tool　for　understanding　cancer　systems.　Extension　of　this　method　holds　the　potential　to　be　fruitful　for　understanding　other　complex,　nonsymmetric　systems.　(C)　2007　Elsevier　Inc.　All　rights　reserved.