[BioC] 2x2 factorial with dye effect, limma design

[Gordon K Smyth]

Dear Thomas,

limma provides the function makeContrasts() to generate the contrast matrix automatically.  If you
are not certain that your manually created matrix is correct, why not use the automatic way?

Best wishes
Gordon

On Tue Dec 5 17:52:53 CET 2006 Thomas Degenkolbe wrote:
> Hi
>
> I'm trying to set up a contrast matrix for my microarray experiment and
> I'm not sure whether I did it in the right way. The experiment has a 2x2
> factorial design (S or T; c or d) and samples (Sc, Sd, Tc, Td) were
> hybridized in a loop design. I want to include the dye effect in the
> design and calculate the contrasts for:
> d vs. c;
> dT-cT;
> dS-cS and
> the interaction [(dT-cT)-(dS-cS)].
>
> Following the example for a single factor exp. in the user guide with
> dye effect I tried this:
>
>> design <- modelMatrix(targets, ref="Sc")
> Found unique target names:
>  Sc Sd Tc Td
>> design <- cbind(Dye=1, design)
>> design
>          Dye Sd Tc Td
> Slide03   1  1  0  0
> Slide04   1  0  1 -1
> Slide05   1 -1  1  0
> Slide06   1  0  0  1
> Slide09   1  0 -1  0
> Slide10   1  1  0 -1
> Slide19   1 -1  1  0
> Slide20   1  1  0  0
> Slide21   1  0 -1  1
> Slide22   1  0  0 -1
> Slide23   1 -1  0  0
> Slide25   1  0 -1  1
> Slide27   1 -1  0  0
> Slide28   1  0 -1  1
>
>> contrast.matrix <- cbind(
> +   "d-c"=c(0,0.5,-0.5,0.5),
> +   "dT-cT"=c(0,0,-1,1),
> +   "dS-cS"=c(0,1,0,0),
> +   "Interaction"=c(0,-1,-1,1))
>> rownames(contrast.matrix)<-colnames(design)
>> contrast.matrix
>      d-c dT-cT dS-cS Interaction
> Dye  0.0     0     0     0
> Sd   0.5     0     1    -1
> Tc  -0.5    -1     0    -1
> Td   0.5     1     0     1
>
>
> Is this correct?
>
>
> Thank you very much in advance,
> Greetings,
> Thomas Degenkolbe
>
> Max-Planck-Institute of Molecular Plant Physiology
> AG Hincha
> Central Infrastructure Group Transcript Profiling
> D-14424 Potsdam
>
>