[BioC] MAS5 and GCOC difference to get CEL files

Adaikalavan Ramasamy ramasamy at cancer.org.uk
Mon Sep 13 12:13:38 CEST 2004


I assume that you used 
 i)  12 CEL files to obtain MAS 5.0 via R and
 ii) 4 CEL files via GCOS (GeneSpring Operating Software ?) which I
assume also gives MAS 5.0

IMHO, you cannot compare these 12 expression measures to the 4 files. 
It would be interesting to compare the MAS 5.0 expression from the
combined 16 CEL files via R and GCOS. This would tell you to which
extent the softwares differ but this page could be of help
http://stat-www.berkeley.edu/users/bolstad/MAS5diff/Mas5difference.html

Personally, I would not combine any expression measure performed with
two different softwares because any significant differences could be
either real or simply due to software differences.

As for the differing file sizes, could it be due to the new GCOS binary
format. I not familiar with GCOS at all.

Regards, Adai


On Mon, 2004-09-13 at 10:41, Delphine Fleury wrote:
> Dear all,
> I am analyzing affymetrix data ATH1 from the same biological experiment 
> (material growth, harvesting and RNA extraction done at the same time in 
> the same conditions) but hybridized at 2 different dates (the 2d set 
> done 1 year later). So , 12 chips were preprocessed with MAS5 getting 
> CEL files and 4 chips with GCOS.
> At first, I see that the size of the CEL files are about 11 MB for the 
> 12 chips and about 5 MB for the 4 last chips. Secondly, the mean signal 
> of the 4 last chips is lower than those from the 12 chips (cf boxplots 
> before normalization, and AffyPLM). 
> Is it possible that the lower signal is related to the preprocessing 
> method from GCOS ?
> Is there someone who observe the same things ?
> Could it be a problem to analyse all chips together ?
> Thanks
> Delphine
>



More information about the Bioconductor mailing list