[BioC] Limma analysis questions

[Martin Janssen]

Dear list,

Just need some place to check my code for the analysis of my array
experiment using Limma. As I'm unsure and have a feeling I'm missing
something.

My experimental setup:
2 groups A and B of 6 samples in a common refrence design all in dye
swap using 2 color agilent slides.

My targets file looks like this:
Name Filename Cy3 Cy5 SampleCy3 SampleCy5
Name	FileName	Cy3	Cy5	SampleCy3	SampleCy5
A1	A1.txt	RF	A	REF	A1
A1ds	A1ds.txt	A	RF	A1	REF
A2	A2.txt	RF	A	REF	A2
A2ds	A2ds.txt	A	RF	A2	REF
A3	A3.txt	RF	A	REF	A3
A3ds	A3ds.txt	A	RF	A3	REF
A4	A4.txt	RF	A	REF	A4
A4ds	A4ds.txt	A	RF	A4	REF
A5	A5.txt	RF	A	REF	A5
A5ds	A5ds.txt	A	RF	A5	REF
A6	A6.txt	RF	A	REF	A6
A6ds	A6ds.txt	A	RF	A6	REF
B1	B1.txt	RF	B	REF	B1
B1ds	B1ds.txt	B	RF	B1	REF
B2	B2.txt	RF	B	REF	B2
B2ds	B2ds.txt	B	RF	B2	REF
B3	B3.txt	RF	B	REF	B3
B3ds	B3ds.txt	B	RF	B3	REF
B4	B4.txt	RF	B	REF	B4
B4ds	B4ds.txt	B	RF	B4	REF
B5	B5.txt	RF	B	REF	B5
B5ds	B5ds.txt	B	RF	B5	REF
B6	B6.txt	RF	B	REF	B5
B6ds	B6ds.txt	B	RF	B5	REF

Samples: A1-A6 B1-B6
Common refrence: REF RF
ds: dyeswap

Files contain nomalized R G values for 45220 probes:
reporterId	R	G
AA000001	2028.38	3370.22

Loading RG values:
data.import.RG <-
read.maimages(targets,source="generic",columns=list(R="R",G="G"),other.columns=NULL,annotation=list(ID="reporterId"),verbose=TRUE,sep="\t",quote="",names=targets$Name)

## apply the design of the experiment
design <- modelMatrix(targets, ref="RF")
design
## make MA data of the RG data, using normalized data so method=none
MAdata <- normalizeBetweenArrays(data.import.RG, method="none")
fit <- lmFit(MAdata, design)
## fit contrasts AtoB
fitAB <- contrasts.fit(fit,c(1,-1))
fiteB_AB <- eBayes(fitAB)
toptable = topTable(fiteB_AB, number=nrow(MAdata), adjust="fdr")
write.table(toptable, "toptableAvsB.txt", sep="\t")
## heatmap
selected <- p.adjust(fiteB_AB$p.value) <0.05
dim(selected)
MA.selected <- MAdata$M[selected, ]
dim(MA.selected)
hmcol <- maPalette(low="red",high="green",mid="yellow",k=21)
heatmap(MA.selected,col = hmcol)


So now my questions:
1. is this the right way to do the analysis? I have some feeling that
I'm missing something.

2. If I make a heatmap this way I still see dyeswaps, is this ok or am
I missing a step?

3. is it possible to make a heatmap of the toptable logFC data?

Any pointers are welcome.

Regards, Martin