[BioC] experience with rankproduct vs. limma - combining results

[Hildebrand, Georg]

Hi bioc-community,

just for testing issue i have compared the output(diff. expr. genes) of limma
and rank product. So I found, that with equal filtering parameters (fdr,
foldchange), the intersection of genes that are assumed to be diff. expressed
between both methods is between 60-70%.

So although the approach of both methods is different, could it be imaginable
to combine the output of both packages to increase the size of a gene set (so
far regardless of stat. questions)? 
Has someone experience with that issue or experience with witch method
provides more validity?

best regards ,
Georg