[BioC] Finding peaks on tiling arrays/ moving average?

[Sean Davis]

On Tuesday 06 February 2007 09:02, J.delasHeras at ed.ac.uk wrote:
> Dear all,
>
> I am looking at some tiling arrays (actually, not quite tiling... they
> have a 50bp oligo every 100bp or so) for a given chromosome.
>
> They have been hybridised with the product of ChIP using antibodies
> against several chromatin proteins.
>
> I am interested in some particular genes, and as I know their location
> I can easily find them and see what the patter is around them.
> But I am also interested in just finding where these proteins bind
> along the chromosome.
>
> Does anybody know of a procedure (or even a package) that would be
> useful for this purpose? I imagine that using moving averages (or
> medians) with a window teh same length as my average ChIP fragment
> would be a way to start, but then I have to come up with a way to give
> probes scores etc... and I feel that I'd just be reinventing the wheel.
>
> Any ideas/pointers would be very much appreciated! Thanks!

I have a pretty simple package here:

http://watson.nci.nih.gov/twiki/bin/view/Main/MeltzerLabSoftware

The method is described here in brief:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=16939795&query_hl=8&itool=pubmed_docsum

There is standard documentation available in the package.  Also, the website 
has the equivalent of a step-by-step tutorial.  If you are using Nimblegen 
arrays and have GFF files available, there is some code to read those files 
directly.  

Sean