[BioC] Help on mapping10xba142

[James W. MacDonald]

Zhang, Jacob (Zhongfa) wrote:
> Dear All:
> 
> Thank you  all for your timely and valuable suggestions in trying to
> help out in installing the mapping10kxbar142 to R.   Unfortunately, none
> worked though.  The most promising method given by James attached below
> worked well until the   last step that need to build an R-package.  When
> in R I trye: 
> 
> 
>>makePDpackage("Mapping10K_Xba142.cdf","Mapping10K_Xba142.csv", 
> 
> "Mapping10K_Xba142_probe_fasta", type = "SNP")
> 
> R gave:
> 
> Assuming Affymetrix Chip
> The package will be called pdmapping10kxba142 
> SNP Array
> Reading Mapping10K_Xba142.cdf 
> Magic number is not 67. This is probably not a binary cdf file.

Ah, this is an incorrect assumption on my part. The cdffile is not 
binary, and a quick look indicates that Affy is still using their 
previous setup for the 10k chip, so it is incorrect to tell 
makePDpackage that the chip is a SNP chip (as odd as that may seem).

When Affy switched from the 10k to 100k SNP chips, they made some 
fundamental changes in the chip design. The 10k chips are very similar 
in many ways to the expression arrays, so the existing functionality for 
the expression arrays will work with these chips as well.

You can make a package for these chips in one of two ways:

library(makecdfenv)
make.cdf.package("Mapping10K_Xba142.cdf")

and then install using R CMD INSTALL. You would then use the affy 
package to read in the celfiles and compute expression values for each 
SNP. Unfortunately, that is the extent of functionality in the affy 
package for these chips, so any other analyses will require you to write 
your own functions.

Alternatively you can do:

library(makePlatformDesign)
makePDpackage("Mapping10K_Xba142.cdf", "Mapping10K_Xba142_probe_fasta")

and install using R CMD INSTALL. You would then use the oligo package to 
read in the celfiles and possibly compute expression values. At this 
point you would have to look at the functions in oligo and see if they 
are applicable to your chip. Since the oligo package is designed to work 
with most oligonucleotide chips it should ultimately contain 
functionality to analyze your chips. However, it is still under 
development so the functions may or may not be in existence as yet.

If you do use the oligo package, any feedback or bug reports you may 
have will be appreciated. It is probably not necessary to send them to 
the whole list. Instead, send to me, Rafael Irizarry, and Benilton 
Carvalho. You have my email address, and Rafa and Benilton's can be 
found by packageDescription("oligo").

Best,

Jim


> 
> 
> Any suggestions?
> 
> Thanks, 
> 
> Jacob Zhang

-- 
James W. MacDonald
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
7410 CCGC
Ann Arbor MI 48109
734-647-5623