[BioC] Scaling affy CEL files

[James W. MacDonald]

Hi Garry,

Garry Moss wrote:
> Dear Bioconductor users,
> 
> I am trying to analyse expression experiment carried in two different
> Affymetrix facilities. The arrays from the two facilities show a
> significant difference in expression medians, and this difference
> persists when I use normalization such as RMA (with all the arrays
> togather). I would like to scale the CEL files and save the scaled
> data in the same format so I can use the new objects with the
> probe-level normalization methods offered by bioconductor,
> normalizing all the arrays together (i.e not just simple data frame
> of the expression values, but an object accepted by exspresso() or
> other functions).

This is almost surely a bad idea. Most of the current methods for 
computing expression values take advantage of the fact that the probe 
pattern within a probeset is similar for samples that were processed 
together. This really breaks down when you have samples processed in 
different facilities, and a simple scaling of the raw data will not 
alleviate the problem.

If the experiments are duplicated in some sense (e.g., you have similar 
treated and controls from both facilities, and the goal is to compare 
treated vs control) you should be able to compute expression values 
separately for samples from each facility, and then compare using a 
linear model that includes a batch effect.

If you don't have some sort of duplication (e.g., the controls are from 
facility 1 and you want to compare to treated from facility 2), then the 
biological and technical variability will be aliased, and there is no 
way to separate the two.

Best,

Jim


> 
> 
> Any help will be greatly appreciated,
> 
> Thanks, Garry
> 
> 
> ___________________________________________________________ Want
> ideas for reducing your carbon footprint? Visit Yahoo! For Good
> http://uk.promotions.yahoo.com/forgood/environment.html [[alternative
> HTML version deleted]]
> 
> _______________________________________________ Bioconductor mailing
> list Bioconductor at stat.math.ethz.ch 
> https://stat.ethz.ch/mailman/listinfo/bioconductor Search the
> archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor

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