[BioC] limma question

Gordon Smyth smyth at wehi.edu.au
Wed Mar 10 23:06:51 MET 2004


At 06:33 PM 10/03/2004, Julia Engelmann wrote:
>Hello all,
>
>I am using limma on Affymetrix data and I read and followed the User's Guide
>so far,  but one thing is still not clear to me:
>Are the M-Values returned by the toptable- command estimated log2-fold
>changes?

The M-values are on the same scale as the expression values that you give 
lm.series(). If exprs(Set) contains log2-intensities, then coef=1 will 
contain log2-intensities for your first condition and coef=2 will contain 
log2-fold changes between your two conditions. If exprs(Set) contains 
loge-intensities then coef=2 is loge-fold changes. If exprs(Set) contains 
absolute intensities (not recommended) then coef=2 is the change is average 
value of the intensities.

Gordon

>  I know typing "?toptable"  in R says "M: estimate of the effect or
>the contrast", but I don't understand what that means.
>If they aren't, where can I get fold changes for my data?
>
>I have 4 untreated and 4 treated experiments, my R code is:
>design<- cbind(c(1,1,1,1,1,1,1,1),c(0,0,0,0,1,1,1,1))
>fit   <- lm.series(exprs(Set),design)
>eb    <- ebayes(fit)
>result <-toptable(coef=2,number=300,genelist=data,fit=fit,adjust="fdr")
>
>Thanks a lot for your help,
>Julia



More information about the Bioconductor mailing list