[BioC] Limma model parameters design, combining a paired with a factorial design

[john herbert]

Dear James,
I am sorry for the delayed reply, I was excited to do the analyses (a
good sign from a science point of view). Thank you very much for your
help.

Although not independent as you say, I would guess the operation is
having a significant effect on gene expression, so I decided to do 2
separate analyses (comparing case vs. control for both before and
after the operation).

Kind regards,

John.

On Thu, Oct 13, 2011 at 6:57 PM, James W. MacDonald
<jmacdon at med.umich.edu> wrote:
> Hi John,
>
> On 10/13/2011 1:29 PM, john herbert wrote:
>>
>> Hello,
>> I have a complex design matrix to produce (for me anyway) for single
>> colour array data.
>>
>> I have 16 samples for 8 patients, which are paired (P1 to P4).
>>
>> Patients 1 and 3 are control patients and patients 2 and 4 are affected
>>
>> Before and after refer to before and after an operation.
>>
>> File1 P1 Control before
>> File2 P1 Control after
>> File3 P2 Affected before
>> File4 P2 Affected after
>> File5 P3 Control before
>> File6 P3 Control after
>> File7 P4 Affected before
>> File8 P4 Affected after
>>
>> If to look at the Limma manual, section "8.3 Paired Samples" looks
>> almost good for this design.
>>
>> However, does it need combining with a factorial design (section "8.7
>> Factorial Designs") as there is a before and after op?
>
> You don't have a factorial design. This is just a simple control vs affected
> with pairing.
>
> If the before and after were different people, this could be analyzed as a
> factorial design, but since they are the same person you cannot assume
> independence between the before and after samples.
>
> pairs <- factor(rep(1:4, each = 2))
> type <- factor(rep(1:2, each = 2, times = 2))
>
> design <- model.matrix(~type+pairs)
> fit <- lmFit(eset, design)
> fit2 <- eBayes(fit)
>
> topTable(fit2, coef = 2)
>
> Best,
>
> Jim
>
>
>
>>
>> How do I combine a design for paired samples with a factorial design?
>>
>>
>> > From the manual, below, the paired and the factorial examples looks
>> scary to combine. ?
>>
>> FileName SibShip Treatment
>> File1 1 C
>> File2 1 T
>> File3 2 C
>> File4 2 T
>> File5 3 C
>> File6 3 T
>>
>> A moderated paired t-test can be computed by allowing for sib-pair
>> e ects in the linear model:
>>>
>>> SibShip<- factor(targets$SibShip)
>>> Treat<- factor(targets$Treatment, levels=c("C","T"))
>>> design<- model.matrix(~SibShip+Treat)
>>> fit<- lmFit(eset, design)
>>> fit<- eBayes(fit)
>>> topTable(fit, coef="TreatT")
>>
>> frame:
>>
>> FileName Strain Treatment
>> File1 WT U
>> File2 WT S
>> File3 Mu U
>> File4 Mu S
>> File5 Mu S
>>
>>> TS<- paste(targets$Strain, targets$Treatment, sep=".")
>>> TS
>>
>> [1] "WT.U" "WT.S" "Mu.U" "Mu.S" "Mu.S"
>>
>>> TS<- factor(TS, levels=c("WT.U","WT.S","Mu.U","Mu.S"))
>>> design<- model.matrix(~0+TS)
>>> colnames(design)<- levels(TS)
>>> fit<- lmFit(eset, design)
>>> cont.matrix<- makeContrasts(
>>
>>   WT.SvsU=WT.S-WT.U,
>>   Mu.SvsU=Mu.S-Mu.U,
>>   Diff=(Mu.S-Mu.U)-(WT.S-WT.U), levels=design)
>>>
>>> fit2<- contrasts.fit(fit, cont.matrix)
>>> fit2<- eBayes(fit2)
>>
>> We can use topTable() to look at lists of di erentially expressed
>> genes for each of three
>> contrasts, or else
>> 48>  results<- decideTests(fit2)
>>>
>>> vennDiagram(results)
>>
>> _______________________________________________
>> 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
>
> --
> 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
>