EdgeR -- normalisation and paired test.

[raphael123]

Hi I am runing EdgeR to make an exact t-test (Iwould like to do a pairwise cmparaison but I am still not sure about that).
So here is to way I can compare time 0 versus time 8 for the positive responder on the experiment.
In one case I am normalising with all my data, then I filter all those who respond positively
In the other case I just take the positive responder for normalisation.

Of course the first case seems better but I would like to have your comments ...
Question : Would you take some data from another experiment to help data normalisation ?
Question : Would you give me a tips to use subject number to pair my sample over time ?

miRNA=read.table("merged_filtered_data.csv", header=TRUE,row.names=1)
roundmiRNA=round(miRNA)

info=read.table("samples_infos.csv", header=TRUE,row.names=1)
miRNAdesign=data.frame(row.names = colnames(info),
  week=t(info)[,"week"],
  subject=t(info)[,"sub"],
  response=t(info)[,"MARSD"]
)

is_responder=miRNAdesign$response==1

#first way to do it

y <- DGEList(counts=roundmiRNA,group=miRNAdesign$week)
y <- calcNormFactors(y)
y <- estimateCommonDisp(y, verbose=TRUE)
y <- estimateTagwiseDisp(y,trend="none")
et <- exactTest(y[,is_responder])
topTags(et)


#second way to do it

y <- DGEList(counts=roundmiRNA[,is_responder],group=miRNAdesign$week[is_responder])
y <- calcNormFactors(y)
y <- estimateCommonDisp(y, verbose=TRUE)
y <- estimateTagwiseDisp(y,trend="none")
et <- exactTest(y)
topTags(et)

[TiborNagy]

You can download data of other experiments from SRA.