[BioC] vsn and RMA-affy ...?

[Wolfgang Huber]

Hi Dario,

> i was following the discussion about "vsn on log" array data.
> does it apply also for the RMA-preprocessed affy data?
> since the expression values are log-scaled, is it incorrect to  
> perform vsn normalization on them?

Yes, it is incorrect. It is also not necessary, because RMA does a
background correction that is (for the data I've seen) in fact quite
close to "variance stabilization".

Please see the vignette of vsn for an example of how to use vsn as a
normalization method for Affymetrix data:

http://www.bioconductor.org/repository/devel/vignette/vsn.pdf
(Section 3 on page 8)

library("affy")
library("vsn")

result = expresso(myAffyBatch, bg.correct = FALSE,
  normalize.method = "vsn",
  pmcorrect.method = "pmonly",
  summary.method = "medianpolish")


Cheers
 Wolfgang
-- 
------------------------------------------------------------------
Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber