[BioC] LIMMA - timecourse and direct comparison exp. design

[Gordon K Smyth]

> Date: Tue, 22 Feb 2005 17:01:42 -0800
> From: Federico Scossa <fscossa at pw.usda.gov>
> Subject: [BioC] LIMMA - timecourse and direct comparison exp. design
> To: bioconductor at stat.math.ethz.ch
>
> Hi all !
> I posted a message several days ago but I didn't get any answers. So I
> assume that the time effect in a direct comparison exp. design (across
> multiple timepoints) is non-estimable.
> Is there anyone able to confirm my assumption?

The time effect is not estimable using log-ratio analyses.  You can only estimate the time effect
using a single channel analysis, for example using the function lmscFit() in limma.

A section on single channel analysis was added to the limma User's Guide in version 1.8.19,
available from http://bioinf.wehi.edu.au/limma

Gordon

> I paste my old message again here:
>
> "my experiment is a 2-color direct comparison wt vs mutant, but repeated on
> 5 successive timepoints.
> each timepoint is structured this way:
>
> FileName	Cy3	Cy5
> 1		mu1	wt1
> 2		wt1	mu1
> 3		mu2	wt2
> 4		wt2	mu2
>
> (1 and 2 are to distinguish the 2 bioreps)
>
> I'm interested not only in the DE genes between wt and mutant at each
> timepoint, but I would like to extract information independently from the
> wt and mutant time-courses, even if I don't have direct comparison between
> successive timepoints. Is this possible? I think there is an example in
> limma's user guide (chapter 16, time-course experiments), but it is for
> one-color array (at least, I guess), and I don't really know how to adapt
> the script to my case."
>
> Any comment or suggestion is really appreciated.
> Thank you.
> Federico