[BioC] Affymetrix promoter arrays

[Jenny Drnevich]

Hi Ania,

I had only marginal success in using the oligo package to look at my 
promoter tiling data, and I never did get any responses about the arguments 
for makePDpackage. If you were able to easily make the package for the 
human promoter tiling array and read in your data, then you got as far as I 
did. I had the hardest time reading in my data because the number of probes 
in the bpmap file did not match the number of probes on the mouse promoter 
tiling array, and so the indexing didn't match. I ended up having to change 
some of the indexing manually, and I finally got my data read in correctly, 
I _think_. However, I wasn't able to do much more than a MA plot and 
density plots, on raw data and after background correction and 
normalization. My two arrays were very similar, but since the oligo package 
didn't seem to have any statistics, I then used Affy's TAS software. More 
on this later...

What did you do that showed that 2 chips were 'different'? Were the two 
different in the same way, and did they happen to be a paired set of chIP 
and control? Have you tried background correction and/or normalization? The 
rma background correction method works with promoter tiling data, and you 
can do just background correction using bg.correct.rma(). Likewise, you can 
do only the quantile normalization with normalize(). Both functions return 
an AffyBatch object. I think it is important to look at your raw data to 
see if there are any problems and whether background correction and/or 
normalization see appropriate. Other software packages that have 
statistical methods for tiling data don't have any quality control on the 
raw data, so using the oligo package is the place to start.

The documentation for Affy's TAS software is pretty sparse, and I spent 3 
afternoons on the phone with tech support. They were very helpful and 
answered lots of questions. It turns out that for their commercially 
available tiling arrays, the probes are only the R direction. The standard 
chIP protocol pulls down double-stranded DNA, so even if the TF binds to 
the + strand, the - strand at that location will also be enriched. Tech 
support did send me a TAS workflow that shows the analysis of human 
promoter arrays, but I don't think it's available on their website yet - 
you may have to call and ask for it. The downside is that while it does 
some statistical algorithms, picking the threshold to declare an 'enriched' 
region is done mostly by eye, and they have no way to easily give you a 
list of the enriched regions. You can take the file into their Integrated 
Genome Browser to overlay the enriched regions to the genome, but you have 
to manually see what genes are near the enriched region. Very tedious...

Another software program that you should be able to use is chipinspector. I 
didn't have replicates, so I couldn't use it. There is a presentation 
available on using chipinspector for human promoter tiling data: 
http://www.affymetrix.com/userForum/event/Symposium_Genomatix.uf  There was 
a similar presentation earlier this month, but it's not available yet to 
download.

Good luck,
Jenny


At 05:37 AM 2/27/2007, you wrote:
>Hello Jenny,
>
>I found your email on bioconductor mailing list. I've started human 
>promoter array analysis and I ended up with the same set of questions you 
>asked. I saw no answer to them on the list, have you figured them out perhaps?
>I would be very grateful for any help, as I have no clue now what to do 
>with my data.
>
>Till now I managed to make a package (with bpmap and cif file, I took only 
>PR files - again, are F files for other strand so should I use them for 
>promoters of genes located on the other strand?) and got raw intensity 
>values. I can see that 2 out of 6 chips look differently (so probably sth 
>is wrong with them). But I have no idea how to judge which are good 
>quality and what is bad. I'm stucked:(
>
>Thanks a lot!
>
>Ania Lorenc
>
>
>
>--
>  Anna Lorenc | Max Planck Institute for Evolutionary Anthropology
>  Deutscher Platz 6 | 04103 Leipzig | Germany
>  tel: +49 341 3550 503 | fax: +49 341 3550 555
>  www.eva.mpg.de/genetics/

Jenny Drnevich, Ph.D.

Functional Genomics Bioinformatics Specialist
W.M. Keck Center for Comparative and Functional Genomics
Roy J. Carver Biotechnology Center
University of Illinois, Urbana-Champaign

330 ERML
1201 W. Gregory Dr.
Urbana, IL 61801
USA

ph: 217-244-7355
fax: 217-265-5066
e-mail: drnevich at uiuc.edu