[BioC] limma, contrasts and affymetrix query

[James W. MacDonald]

Hi Iain,

Iain Gallagher wrote:
> Dear list
> 
> I wonder if someone could help me with some conceptual issues
> regarding limma, contrasts and analysis of an affymetrix study.
> 
> I have 8 samples (in duplicate) and 3 factors. The factors are O2
> level (Norm, Hypo), Radiation (Yes/No) and Treatment (None, DMSO,
> Drug1, Drug2).
> 
> The drugs were delivered to the cells using DMSO as a carrier. I am
> trying to use contrasts to examine the effect of the drug between two
> O2 levels taking into account the effect of DMSO:
> 
> TS <- paste(phenoData(rmaDataFiltered)$O2,
> phenoData(rmaDataFiltered)$Rtx, phenoData(rmaDataFiltered)$Treatment,
> sep=".")
> 
> TS <- factor(TS, levels=c("Hypo.No.None", "Hypo.No.DMSO",
> "Hypo.No.Drug1", "Hypo.No.Drug2",  "Hypo.Yes.None", "Hypo.Yes.DMSO",
> "Hypo.Yes.Drug1", "Hypo.Yes.Drug2",  "Norm.No.None", "Norm.No.DMSO",
> "Norm.No.Drug1", "Norm.No.Drug2", "Norm.Yes.None", "Norm.Yes.DMSO",
> "Norm.Yes.Drug1", "Norm.Yes.Drug2"))
> 
> design <- model.matrix(~0+TS) colnames(design) <- levels(TS) fit <-
> lmFit(rmaDataFiltered, design)
> 
> cont.matrix <- makeContrasts(HypoDrug1 =
> (Hypo.No.Drug1-Hypo.No.None)-(Hypo.No.DMSO-Hypo.No.None), NormDrug1 =
> (Norm.No.Drug1-Norm.No.None)-(Norm.No.DMSO-Norm.No.None),
> levels=design)
> 
> For these contrasts my thinking is as follows; I want to examine the
> effect of the drug versus untreated cells taking into account the
> effect of DMSO. So first I want to see the effect of DMSO on the
> untreated cells:
> 
> (Hypo.No.DMSO-Hypo.No.None) and (Norm.No.DMSO-Norm.No.None)
> 
> Then I want to remove this effect from the effect of the drug on
> untreated cells (i.e. create the difference of a difference):
> 
> HypoDrug1 = (Hypo.No.Drug1-Hypo.No.None)-(Hypo.No.DMSO-Hypo.No.None)
> (effect of drug - effect of DMSO)

Note that algebraically this is the same as Hypo.No.Drug1 - Hypo.No.DMSO

> 
> and
> 
> NormDrug1 = (Norm.No.Drug1-Norm.No.None)-(Norm.No.DMSO-Norm.No.None)
> (effect of drug - effect of DMSO)

Ditto for this one (Norm.No.Drug1 - Norm.No.DMSO). So you don't need to 
involve the baseline control samples in these comparisons.

> 
> My question is am i going about this in the right way?

Remember that you are just doing algebra here, so it is always possible 
that a given contrast actually reduces to something much simpler.

Best,

Jim


> 
> Thanks.
> 
> Iain
> 
> 
> _______________________________________________ Bioconductor mailing
> list Bioconductor at stat.math.ethz.ch 
> https://stat.ethz.ch/mailman/listinfo/bioconductor Search the
> archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor

-- 
James W. MacDonald, M.S.
Biostatistician
Douglas Lab
University of Michigan
Department of Human Genetics
5912 Buhl
1241 E. Catherine St.
Ann Arbor MI 48109-5618
734-615-7826
**********************************************************
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues