[BioC] Re-mapped Affy CDF files

[Sean Davis]

On 1/10/06 4:00 PM, "James MacDonald" <jmacdon at med.umich.edu> wrote:

> Has anybody taken a close look at the re-mapped cdf files that the MBNI at the
> University of Michigan have produced?
> 
> http://brainarray.mhri.med.umich.edu/Brainarray/Database/CustomCDF/genomic_cur
> ated_CDF.asp
> 
> I find the idea of revisiting the mapping of probes to genes as well as
> combining the disparate probesets for a given gene into a single probeset
> appealing, but I think something more than 'Hey that looks like a good idea!'
> is needed before one could make the case for switching from Affy's cdf to
> these re-mapped ones.
> 
> Unfortunately, given the fact that there isn't a one-to-one correspondence
> between the Affy and MBNI probesets, doing a critical comparison of the
> results is not straighforward. For instance, using the Affy spike-in or
> GeneLogic dilution datasets won't work AFAIK, since it seems that both Affy
> and GeneLogic spiked in probe-specific cRNA rather than spiking in full length
> transcripts for particular genes.
> 
> I know of at least one instance where a client was comparing samples with a
> particular gene 'knocked in' to wild type, and we didn't see any difference
> when using the Affy CDF but the gene was at the top of the list of significant
> genes when using an MBNI CDF, so we have some small indication that the MBNI
> CDFs might be better.
> 
> So, does anybody have any experience with these CDFs that might give an
> indication of their usefulness? Barring that, does anybody have a good way
> that one could reasonably compare the two probe mappings?

I'm not sure what their build process is, but doesn't Ensembl do some
probe-based mappings?

Sean