[BioC] how to do quantile normalization under dye swape situation

[J.delasHeras at ed.ac.uk]

Quoting yanju <yanju at liacs.nl>:

> Hello All,
>
> my microarray datasets are like follows. And I want to normalize the data.
> Targets:
> Sample    FileName    Dye    Stage    Cy3    Cy5
> ZWS57    57.gpr    T5C3    high oblong    ref       high
> ZWS58    58.gpr    T3C5    high oblong    high    ref
> ZWS61    61.gpr    T5C3    high oblong    ref       high
> ZWS62    62.gpr    T3C5    high oblong    high    ref
>
> ref is the commen reference. First I think Within Array Normalization is
> not neccesary (am i right?). Then I want to do the Between Array
> Normalization using "quantile" method to insure the commen reference
> have the same distribution. But with dye swap, I can not use neither
> "Rquantile" nor "Gquantile". What should I do? Or other normalization
> suggestions?
>
> Regards,
> Yanju Zhang

Hi Yanju,

why do you think you don't need to normalise within arrays?
In a 2-colour array experiment, within-array normalisation is usually  
the one normalisation that's required... but maybe I'm missing  
something about your experiment?

Unless your data have an unusual distribution, probably print-tip  
loess normalisation (this is within array, of course) will be  
appropriate. It looks like you're using 'limma', and your targets file  
indicates the orientation of the hybs, which then will be reflected in  
the design matrix... the linear model fit will take care of teh dye  
swaps. You can also add a "Dye Effect" coefficient to be calculated,  
if you wish. The Limma Users Guide contains some nice examples about  
this.

Jose

-- 
Dr. Jose I. de las Heras                      Email: J.delasHeras at ed.ac.uk
The Wellcome Trust Centre for Cell Biology    Phone: +44 (0)131 6513374
Institute for Cell & Molecular Biology        Fax:   +44 (0)131 6507360
Swann Building, Mayfield Road
University of Edinburgh
Edinburgh EH9 3JR
UK