[BioC] limma design problem with a two color dataset

[Gordon K Smyth]

Dear Yong Li,

This experiment has been designed in four blocks, each one representing a 
biological replicate of all five genotypes.  The best and safest way to 
analyse it is to treat each of the four blocks as batches, removing any 
block effect.  You can do this by:

   block <- factor(rep(1:4,each=4))
   design.block <- model.matrix(~block)

   targets$Cy3 <- removeExt(targets$Cy3)
   targets$Cy5 <- removeExt(targets$Cy5)
   design.mutants <- modelMatrix(targets, ref="Col0")

   design <- cbind(design.block,design.mutants)

This design will also allow for any probe-specific dye effects.

Best wishes
Gordon

> Date: Mon, 21 Mar 2011 12:06:12 +0100
> From: Yong Li <yong.li at zbsa.uni-freiburg.de>
> To: bioconductor at r-project.org
> Subject: [BioC] limma design problem with a two color dataset
>
> Dear list,
>
> I am analyzing a two color microarray dataset using limma. The dataset
> is from ArrayExpress database (accession number E-MEXP-1949). The
> samples include 4 mutants (rcd11, rcd13, rcd14 and sro11) and wild-type
> Col0. For the 4 mutants there are 4 biological replicates each. For Col0
> there are also 4 biological replicates but for each biological replicate
> there are 4 technical replicates. The targets I made is as the following:
>
> > targets
>                    FileName      Cy3      Cy5
> 1         122.rcd1.gpr.proc rcd11.r1  Col0.r1
> 2          124.354.gpr.proc sro11.r1  Col0.r1
> 3          125.466.gpr.proc rcd14.r1  Col0.r1
> 4          126.500.gpr.proc rcd13.r1  Col0.r1
> 5         133.rcd1.gpr.proc rcd11.r2  Col0.r2
> 6          134.354.gpr.proc sro11.r2  Col0.r2
> 7          135.466.gpr.proc rcd14.r2  Col0.r2
> 8          136.500.gpr.proc rcd13.r2  Col0.r2
> 9   140.90506.0001.gpr.proc  Col0.r3 rcd11.r3
> 10 141.090506.0001.gpr.proc  Col0.r3 rcd14.r3
> 11 142.090506.0001.gpr.proc  Col0.r3 rcd13.r3
> 12 143.090506.0001.gpr.proc  Col0.r3 sro11.r3
> 13 152.110506.0001.gpr.proc  Col0.r4 rcd14.r4
> 14 153.110506.0001.gpr.proc  Col0.r4 rcd13.r4
> 15 154.110506.0001.gpr.proc  Col0.r4 rcd11.r4
> 16 155.110506.0001.gpr.proc  Col0.r4 sro11.r4
>
> I am interested in finding DE genes between different mutants (rcd11...)
> and wild-type Col0. Reading in the data and normalization have been done
> but I was not able to make a design matrix. I have read the whole limma
> users guide, especially the part dealing with technical replicates but
> could not find something that is similar to this.
>
> Thanks in advance for your help!
>
> Yong Li

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