[BioC] How to avoid losing a sample as reference running an individual differential expression analysis (LIMMA)

[Gordon K Smyth]

> Date: Tue, 10 Feb 2009 06:04:23 -0800 (PST)
> From: Dmitriy Verkhoturov <verkhoturovdm at yahoo.com>
> Subject: [BioC] How to avoid losing a sample as reference running an
> 	individual differential expression analysis (LIMMA)
> To: bioconductor at stat.math.ethz.ch
>
> Hello listmemebers,
>
>
> I have data from two-color microarray expression profiling experiments 
> where 3 whole brain (WB) samples were compared to 3 Mauthner Cells (MC) 
> in a loop design (-> MC #1 -> WB #1 -> MC #2 -> WB #2 -> MC #3 -> WB #3 
> -> MC #1 ->). In addition to phenotype analysis I would also like to run 
> an individual analysis making all pair-wise comparisons. I'm using the 
> LIMMA package in R to do this. The problem is that the contrast matrix 
> that you have to set up requires that one of the samples be designated 
> as a reference, but in doing so you lose the sample.

No you do not lose the reference sample in any way.  On the contrary, 
comparisons are automatically computed between the nominal reference and 
all the other treatments.  Please read the documentation a little more 
carefully.

Best wishes
Gordon

> My question is this, is it possible to run individual analysis with this 
> data set without losing one as a reference point?
>
> Many thanks in advance,
> Dmitriy