[BioC] GEOquery and Limma

[Sean Davis]

On Tue, Oct 18, 2011 at 4:07 AM, Ovokeraye Achinike-Oduaran
<ovokeraye at gmail.com> wrote:
> Hi all,
>
> When running differential expression analysis on Limma with GDS or GSE
> data retrieved with GEOquery, what is the right input option: do.log2
> = TRUE or do.log2=FALSE? I read in one of the documentations
> (http://svitsrv25.epfl.ch/R-doc/library/GEOquery/html/GDS2MA.html)
> that to do the conversions:
>
> GDS2MA(GDS,do.log2=FALSE,GPL=NULL,AnnotGPL=TRUE)
> GDS2eSet(GDS,do.log2=FALSE,GPL=NULL,AnnotGPL=TRUE
>
> The results are obviously very different from the do.log2=TRUE option.
> But the results for logFC from do.log2=FALSE for GDS157, for example,
> looks a bit strange. Is there something I'm reading wrongly perhaps?

Hi, Avoks.  do.log2 when TRUE simply applies a log2 transformation to
the data when constructing an MAList or ExpressionSet from the GDS
data.  It is up to YOU to decide when to perform log2 transformation
of GEO data; there is not a "rule" to follow.  For GDS157, the data
are in GEO as linear intensities.  I believe limma assumes that the
data have been logged, so doing a log transformation, so log2
transformation is probably appropriate.

Sean