[BioC] gcRMA: problem compute.affinities.local on gngnf1musa

[Julien Roux]

Thanks Jim for this clear explanation.
I will contact the authors of the method to get their comments on using 
compute.affinities() instead of compute.affinities.local()
Regards,
Julien

James W. MacDonald a écrit :
> Hi Julien,
>
> Julien Roux wrote:
>> Hello list,
>>
>> In *Schuster et al. 2007 (http://genomebiology.com/2007/8/6/R125) a 
>> method is presented allowing to correct *MAS5 present/absent calls 
>> based on gcRMA transformed PM threshold values.
>> The R script is available in sup data at 
>> http://genomebiology.com/content/supplementary/gb-2007-8-6-r125-s1.txt
>>
>> When trying to run it on my data I am faced to an error:
>>  > data
>> AffyBatch object
>> size of arrays=854x854 features (11 kb)
>> cdf=gnGNF1Musa (36182 affyids)
>> number of samples=16
>> number of genes=36182
>> annotation=gngnf1musa
>> notes= >  ai <- compute.affinities.local(data, Array=NULL)
>> Adjusting for optical effect................Done.
>> Computing base-position profiles for probe affinitiesErreur dans 
>> complementSeq(seqs, start = 13, stop = 13) :
>> Character N does not code for a nucleic acid.
>>
>> I tried with other platforms (e.g. mgu74av2) and this works.
>> So I guess this is due to an error in the annotation package gngnf1musa.
>> I installed this package (custom array affymetrix) made available by 
>> Cei Abreu-Goodger 
>> (http://article.gmane.org/gmane.science.biology.informatics.conductor/13659) 
>> at ftp://ftp.sanger.ac.uk/pub/cei/gngnf1_R_packages.tar.gz
>> What can be wrong?
>>
>> Then I tried:
>>  > ai <- compute.affinities(cdfName(data))
>> Computing affinities.Done.
>>
>> This worked. However I am not sure it is doing the same thing 
>> (Adjusting for optical effect? Computing base-position profiles for 
>> probe affinities?)
>> What is the difference between compute.affinities and 
>> compute.affinities.local?
>
> From the source for these functions:
>
> ##this is almost the same as compute.affinities
> ##except that affinity.spline.coefs is not loaded with data()
> ##but computed using the data provided by the user
>
> So I have to assume that the probe package for your chip has some 
> probes in which Affy used the 'N' IUPAC symbol. Since gcrma doesn't 
> understand anything but ACTG, this causes the error.
>
> So when you use compute.affinities(), you load the pre-calculated 
> affinity data rather than using the sequences from your chip. Hence no 
> error.
>
> Best,
>
> Jim
>
>
>>
>> Thanks for your help to clarify this.
>> Julien
>>
>>  > sessionInfo()
>> R version 2.7.0 (2008-04-22)
>> i386-apple-darwin8.10.1
>>
>> locale:
>> C
>>
>> attached base packages:
>> [1] splines   tools     stats     graphics  grDevices utils     datasets
>> [8] methods   base   other attached packages:
>> [1] gngnf1musaprobe_1.8.1 gngnf1musacdf_1.14.0  gcrma_2.12.0        
>> [4] matchprobes_1.12.0    affy_1.18.0           preprocessCore_1.2.0
>> [7] affyio_1.8.0          Biobase_2.0.1      
>


-- 
Julien Roux, PhD student
http://www.unil.ch/dee/page22707.html
Department of Ecology and Evolution
Biophore, University of Lausanne, 1015 Lausanne, Switzerland
tel: +41 21 692 4221    fax: +41 21 692 4165