[BioC] Using a weight for Illumina data? (limma)

[Benno Puetz]

As I have a lot of genes which show essentially no expression, I thought
of using the Detection (or rather 1-detection) as weight in lmFit. As I
have not found any mention of this yet, I wanted to ask for comments
whether that would be reasonable.

When subsequently using eBayes(), this would probably reduce the 
influence of the low variance in those genes?

-- 
Benno Pütz
Statistical Genetics
Max-Planck-Institut of Psychiatry             Tel.: +49-89-30622-222
Kraepelinstr. 10                              Fax : +49-89-30622-601
80804 Munich, Germany