[BioC] Limma design matrix for a complicated experiment design

julin at aecom.yu.edu julin at aecom.yu.edu
Tue Sep 30 22:03:59 CEST 2008


Dear list,

I am writing to ask if anyone can help me to define a design matrix for
our experiment.

I am working with MoGene ST1.0 chips.Samples are from WT or HIV-transgenic
mouse bone marrow-derived macrophages that we grew in dishes and then
either exposed to TB or not for 24 hours. Each pair is from the same mouse
(i.e. 5202/5203, 5208/5209,5204/5205, 5206/5207, 5210/5211).

Totally 10 chips with experimental design as following:

Chip        Pair        TB        HIV
5202        1        NoTB        NoHIV
5203        1        TB        NoHIV
5204        2        NoTB        HIV
5205        2        TB        HIV
5206        3        NoTB        HIV
5207        3        TB        HIV
5208        4        NoTB        NoHIV
5209        4        TB        NoHIV
5210        5        NoTB        HIV
5211        5        TB        HIV

We want to know the main effect of TB/NoTBtreatment(independent of the HIV
status) and HIV/NoHIV status(independent of the TB/NoTB treatment).

I am trying to use Limma to do this. But I am not sure if it is
appropriate to treat each pair of correlated arrays as a block or not.I
want to use the R script as following to define a design matrix and fit a
linear model:

---------------
TS<-paste(targets$TB,targets$HIV,sep=".")
TS<-factor(TS,levels=c("NoTB.NoHIV","TB.NoHIV","NoTB.HIV","TB.HIV"))
design<-model.matrix(~0+TS)
block<-targets$Pair
dupCor<-duplicateCorrelation(intensity,design,block=block)
dupCor$consensus.correlation
fit<-lmFit(intensity,design,block=block,correlation=dupCor$consensus)
---------------

If so, my design matrix will simply be:

     NoTB.NoHIV   TB.NoHIV   NoTB.HIV   TB.HIV
1             1          0          0        0
2             0          1          0        0
3             0          0          1        0
4             0          0          0        1
5             0          0          1        0
6             0          0          0        1
7             0          1          0        0
8             0          1          0        0
9             0          0          1        0
10            0          0          0        1

Is it correct? Any suggestions from you are appreciated.

julin



More information about the Bioconductor mailing list