[BioC] Normalization with common reference

Wolfgang Huber w.huber at dkfz-heidelberg.de
Mon May 31 21:09:54 CEST 2004

Hi Emanuele,

marchiem at libero.it wrote:

> Hi Wolfgang,
> first of all thank you very much for your quick reply...
> we use gDNA as reference because our lab want to compare our data with other labs which use the same design. We have data express in in gDNA units...
> We normalize these data with Genespring using a globlal normalization method that calculate the scaling factor from the 50th percentile of genes.

So then why not use a global scaling to the median with bioconductor as 
well? There are options for that in the normalization procedures of the 
'marray' and 'limma' packages.

> ...but if I have a time course i don't think I can use these single channels methods to normalize, am I wrong? If I don't consider the reference channel I could confound a brighter array as a more expressed...how can I distinguish that?
> sorry if I am too trivial and thanks again!

I am not sure I understand. It is the purpose of normalization methods 
to correct for some arrays being more or less bright, and the ones that 
I mentioned do so. They rely on some assuptions:
- e.g. median normalization assumes that the median intensity for each 
array should be the same
- e.g. vsn requires that there is a set of >50% of genes on the array 
that are approximately not differentially expressed (in the non gDNA 
- e.g. quantile normalization assumes that the true distribution of 
intensities is the same on all arrays

Whether that holds for your data you need to check. I don't think it has 
anything to do with whether you're looking at a time series or something 

Best wishes
Wolfgang Huber
Division of Molecular Genome Analysis
German Cancer Research Center
Heidelberg, Germany
Phone: +49 6221 424709
Fax:   +49 6221 42524709
Http:  www.dkfz.de/abt0840/whuber

More information about the Bioconductor mailing list