[BioC] Separate channel analysis on connected design

[Naomi Altman]

I prefer it.  Theoretically speaking, you should get a bit more power 
from separate channel analysis.

--Naomi


At 04:53 AM 3/14/2011, Luis A. Alcaraz wrote:
>Hi,
>
>I'm trying to analyse data from a microarray experiment which consist in
>three mutants compared to wild type in a connected design (the experimental
>design is like in a 2x2 factorial design) with limma.
>When I do the analysis like in a factorial design (page 45 of the manual) I
>get a number of differentially expressed genes considerably lower than when
>I do separate channel analysis (page 50). From a biological point of view,
>for me it makes sense the new genes that I found with separate channel
>analysis, but I don't know if this kind of analysis (separate channel
>analysis with connected design) is right from a statistical point of
>view. What do you think?
>Thanks!
>
>Luis.
>
>         [[alternative HTML version deleted]]
>
>_______________________________________________
>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