[BioC] lumi: target ids versus probe ids?

[Mark Dunning]

Hi Michal,

I would definitely recommend working with the probe-level data rather
than the gene-level data.

I have seen many examples of genes with multiple probes where the
results from different probes can look completely different: one probe
can be unexpressed all the time, whereas the other can be differentially
expressed. The reason could be that one probe targets multiple isoforms
of the gene, whereas the other could be targetting one very specific and
rare example. 

I don't think averaging the two signals is at all useful in such
situations.

Hope this helps,

Mark

On Thu, 2008-05-22 at 11:30 -0400, Michal Blazejczyk wrote:
> Hello,
> 
> I have a question regarding best practices of the analysis of Illumina
> data using lumi, more specifically - regarding targets vs probes.
> In noticed that when I import probe-level data using lumiR(),
> and them normalize them, I still end up with probe-level data.
> Are there any "summarization" methods for Illumina data?  I mean,
> when I have expression values for multiple probes from the same
> target, how should I treat them?  The "gene profile" format in
> BeadStudio simply takes a mean of all probes that come from the
> same target.  Is this the right thing to do?  Should it be done
> before or after normalisation?  Or should I ignore target ids
> when I have probe-level data?
> 
> Best regards,
> Michal Blazejczyk
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor