[BioC] network analyses on genes or probesets

[Marc Carlson]

Hi Markus,

One thing you could if you want to go from many probesets to single
genes, is to use the nsFilter() from the genefilter package to retain
the "high variance" probesets.  This will not "combine" probesets, but
instead can be used to jettison all but one probeset (for each gene)
that seems likely to maybe be measuring something.  It also has a
parameter that will let you only get back things that actually have an
entrez gene ID.

Here is a link in case this sounds like something you might want to
pursue this strategy:

http://www.bioconductor.org/packages/release/bioc/html/genefilter.html


  Marc





Markus Schmidberger wrote:
> Hi all,
>
> I am using affymetrix microarray data. In the raw data I have probes.
> After preprocessing (background correction, normalization and
> summarization) I have probesets. Now I can use the annotation packages
> (e.g. library(hgu133a.db)) to get the assignment between probesets and
> genes.
>
> Are there any functions, methods (median, mean,...), literature how to
> combine probesets to genes?
>
> I want to run some network analyses on gene level. Therefore I need
> some kind of 'gene intensity'.
>
> Thanks
> Markus
>