Deleted:Picking differentially expressed genes using consensus

Question

[firestar]

Differential gene expression can obviously be run using various programs and workflows (I call methods from here on). And each method may not necessarily give the same list of DE genes. Would it make sense to pick DE genes as an intersect from various methods and use them a final list of DE genes? Or since methods differ in algorithm or differ in assumptions etc, this is not a good idea? I am looking for arguments 'for' or 'against' a consensus calling of DE genes?

And similar question applies to the list of GO terms obtained from a gene enrichment or gene set analysis.