[BioC] Mapping using absolute coordinates

[Ludo Pagie]

Dear Arnoud,

I am working on a package (CGtools) that does exactly this for
Chip-chip data, or condensed/processed Chip-seq data (raw
chip-seq data is a bit too much to handle). The package is in
production and thus not available from CRAN or bioconductor
site. But for making alignment plots it is working ok. The
package makes an effort to handle overlapping genes correctly
and not to use feature data twice, etc. It's easy to make plots
on 5' or 3' gene borders only, or on all borders together. We
also make an attempt to include a confidence interval around the
average profile. With some work you can make scaled alignments
just as well but as you said these plots tend to introduce
biases. Making alignments on gene subsets of different gene
length is much more instructive.

If you want to use it, even though it is under development,
please let me know offline and I'll sent you a copy.

Ludo Pagie


On Fri, Feb 26, 2010 at 04:13:39PM -0500, Droit Arnaud wrote:
> Dear BioC,
> 
> I am working on ChIP-chip / ChIP-Seq data and I would like to do average
> gene mapping using absolute instead of relative coordinates.
> Does anyone have some idea or knows a package to do this ?
> 
> In more details, I want to do a plot of different
> histone modification enrichments (from WIG files) on a composite
> average gene using absolute coordinates like in PMID 17679090. In
> general people are using relative gene mapping (as first proposed by
> the Young lab in 2005 PMID 16122420) but this technique (easy to
> implement since each coding region is simply divided into equally
> sized bins) often introduce a bias for histone marks/proteins which
> localized at an absolute distance from the start (or the end) of a
> gene (and the bias is even worst when people group genes by their length).
> 
> Thanks
> 
> Arnaud.
> 
> _______________________________________________
> 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