[BioC] Limma time series analysis question

[Gordon K Smyth]

Dear January,

You design seems ammenable to a pretty standard analysis, which would go 
like:

   experiment <- factor(experiment)
   time <- factor(time)
   design <- model.matrix(~time+experiment)
   fit <- eBayes(lmFit(y,design))
   summary(decideTests(fit))

   T2vsT1: topTable(fit,coef=2)
   T3vsT1: topTable(fit,coef=3) etc

Here experiment takes on values A, B, C, and time times on values T1, T2 
etc.

Best wishes
Gordon


> Date: Thu, 28 Oct 2010 12:56:08 +0200
> From: January Weiner <january.weiner at mpiib-berlin.mpg.de>
> To: BioC <bioconductor at stat.math.ethz.ch>
> Subject: [BioC] Limma time series analysis question
>
> Dear all,
>
> I am wondering what would be the optimal approach for limma to the
> following setup:
>
> I have an experiment repeated three times (A, B, C). Each repeat is
> measured at five different time points (T1...T5) in three replicates
> (R1...R3):
>
> A T1 R1
> A T1 R2
> A T1 R3
> A T2 R1
> ...
>
> C T5 R1
> C T5 R2
> C T5 R3
>
> Thus, A T1 is matched with A T2, T3, T4 and T5, but not with the B
> series. So it is a mixed design.
>
> I want to see genes that are differentially expressed relative to T1.
> How do I do it correctly in limma?
>
> Best regards,
> j.
>
> -- 
> -------- Dr. January Weiner 3 --------------------------------------
> Max Planck Institute for Infection Biology
> Charit?platz 1
> D-10117 Berlin, Germany
> Web?? : www.mpiib-berlin.mpg.de
> Tel? ?? : +49-30-28460514

______________________________________________________________________
The information in this email is confidential and intend...{{dropped:4}}