[BioC] normalizing only 2 affy samples

[James W. MacDonald]

Hi Juliet,

On 4/9/2012 9:55 AM, Juliet Hannah wrote:
> All,
>
> Can anyone suggest a strategy to normalize just two affy samples?
>
> I do not seek to carry out any inferential procedures. I would just
> like to make a scatter plot
> of the expression values from both arrays just to see if the
> experiment worked (that is
> expression is being measured).

When you say 'normalize' do you really mean normalize, or are you using 
that term in the context of normalization and summarization, in order to 
get probeset-level expression values?

I'll assume for sake of argument that you mean normalization and 
summarization.

With only two arrays, it isn't clear what the best course of action 
should be. You could argue that mas5() is a better idea, as the model 
being fit is probably the simplest, and is more likely to have 
assumptions fulfilled. The downside to that approach is that mas5() 
really isn't very good.

The summarization method in rma() fits a much more complex model, and 
given only two samples, you could argue that the estimates for probe and 
chip effects won't be very stable.

So either method has inherent drawbacks with so few samples. I would 
tend to use rma() anyway, but that is my bias and is partially dictated 
by a long history of using rma(), and a desire for consistency. I 
actually doubt there will be that much difference in the end.

You might also consider using an MA plot rather than a scatter plot for 
visualization. It will tend to be more interpretable and easier to see 
what is going on.

Best,

Jim


>
> Thanks,
>
> Juliet
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> 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
University of Washington
Environmental and Occupational Health Sciences
4225 Roosevelt Way NE, # 100
Seattle WA 98105-6099