[BioC] Contrast matrix in limma

[Ryan C. Thompson]

Hi,

I don't believe that you can meaningfully test main effects using a 
model that includes an interaction term (as yours does). If I recall 
correctly, in order to test main effects, you would have to fit an 
additive model such as "~Diet + Disease", and any results from that 
model will not be valid for genes with a significant interaction.

-Ryan

On Fri 11 Jul 2014 09:12:27 AM PDT, kaushal Chaudhary [guest] wrote:
> Hello,
>
> In my micro array expriemt,  I have  four different types of  samples (nodiet.nodisease, nodiet.disease, diet.nodisease, and diet.disease).  I want to make the  following 5 comparisons.
> Diet.nodisease vs.  nodiet.nodisease (effect of diet  in no disease)
> Diet.disease vs. nodiet.disease  (effect of diet in disease)
> Nodiet.disease vs. Nodiet.nodiease (effect of disease in no diet)
> Diet.disease vs. Diet.nodisease (effect of disease in diet)
> Diet.disease vs. nodiet.nodisease
> I used  makecontrasts function  in limma in R.
> Contrast.matrix=makecontrasts(
>                                                   Diet.nodisease vs.  nodiet.nodisease= Diet.nodisease - nodiet.nodisease
>                                                    Diet.disease vs. nodiet.disease  = Diet.disease - nodiet.disease
>                                                    Nodiet.disease vs. Nodiet.nodiease= Nodiet.disease -Nodiet.nodiease
>                                                    Diet.disease vs. Diet.nodisease= Diet.disease - Diet.nodisease
>                                                   Diet.disease vs. nodiet.nodisease= Diet.disease -  nodiet.nodisease
>                                                    )
>
> I also want to get the main effect of diet and disease and interaction at  the same time. (diet, disease, diet*disease).  I suppose the code gives  2 diet effects and 2 diseases effect.   I would appreciate any advice.
>
> Thanks !!!
>
>
>
>   -- output of sessionInfo():
>
> None.
>
> --
> Sent via the guest posting facility at bioconductor.org.
>
> _______________________________________________
> 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