[BioC] Re: RMA vs VSN

w.huber at dkfz-heidelberg.de w.huber at dkfz-heidelberg.de
Wed Jun 23 10:49:28 CEST 2004

Roger Vallejo wrote:
> This must be of interest for those preprocessing data from
> affymetrix chips. We have compared RMA vs VSN performing an
> lme-ANOVA. If you are wondering what to use RMA or VSN? or what
> are the potential pitfalls or benefits from using either
> normalization or background data correction
> approach. Then, please read below and make your own conclusions.

Hi Roger,

I don't get your point. As Rafael has already pointed out, vsn is a
normalization method, while RMA is a combination of PM/MM synthesis,
normalization, and probe set summary.

In the vsn vignette I give an example how vsn can be used instead of
quantile normalization as a normalization method within RMA (through the
espresso function). The example is simple-minded, since it ignores the
MMs! This is better than subtracting (PM-MM), but there are smarter ways
to use the MMs, e.g. in GCRMA.  When I made the example, it was
interesting to see that a parametric normalization method like vsn could
in many cases outperform a non-parametric method method like quantile
normalization. But affy preprocessing is a combination of several steps,
and especially the background correction (PM/MM synthesis) seems to
benefit a lot from fine-tuning - see Rafa's and coworkers' recent work.

Also, when you try to estimate fold-changes from weakly expressed genes,
you will always have to deal with the variance-bias trade-off. The
relative measurement error in weak probe signals is large, and thus
log-ratios between them can fluctuate strongly. Different algorithms make
different choices about how to deal with them. What is best for you
depends on the question of interest and the experimental design. E.g. when
you have a very large number of replicates you don't want shrinkage, while
with a small number of replicates, shrinkage is often beneficial.

Best wishes,

Wolfgang Huber
Division of Molecular Genome Analysis
German Cancer Research Center
Heidelberg, Germany
Phone: +49 6221 424709
Fax:   +49 6221 42524709
Http:  www.dkfz.de/abt0840/whuber

More information about the Bioconductor mailing list