Hi,
I am trying to use DESeq2 but I having troubles determining the reduced
formula for the nbinomLRT. I will like to compare the results with the
DEGs I got by nbinomGLMTest using DESeq.
With DESeq I did the following:
co=read.table("C_LPS_PBS_1h_DS.txt",header=T)
ind=c(1,2,4,5,1,3,4,5)
trt=c(rep("A",4),rep("E",4))
a.con=cbind(ind,trt)
adat=newCountDataSet(co,a.con)
adat=estimateSizeFactors(adat)
adat=estimateDispersions(adat,method="blind")
vst=varianceStabilizingTransformation(adat)
adat=estimateDispersions(adat,method="pooled-CR",
modelFormula=count~ind+trt)
a.vsd=getVarianceStabilizedData(adat)
fit0=suppressWarnings(fitNbinomGLMs(adat, count ~ 1))
fit1=suppressWarnings(fitNbinomGLMs(adat, count ~ ind))
fit1a=suppressWarnings(fitNbinomGLMs(adat, count ~ trt))
pvals.g=nbinomGLMTest(fit1,fit0)
pvals.t=nbinomGLMTest(fit1a,fit0)
table(pvals.t<0.01)
table(pvals.g<0.01)
And started with DESeq2:
co=read.table("C_LPS_PBS_1h_DS.txt",header=T)
ind=c(1,2,4,5,1,3,4,5)
trt=c(rep("A",4),rep("E",4))
a.con=cbind(ind,trt)
a.con2 <- colnames (co)
a.con2 <- colnames (co)
rownames(a.con) <-
c("A1","A2","A4","A5","E1","E3","E4","E5")
a.con2 <- data.frame (a.con)
ddsFull <- DESeqDataSetFromMatrix(co, a.con2, design = ~ ind + trt )
design(ddsFull) <- formula(~ ind + trt)
dds <- estimateSizeFactors(dds)
dds <- estimateDispersions(dds)
a.vsd=getVarianceStabilizedData(dds)
After this I am confused because I don't know how to write the reduced
formula that will allow me to compare the null model ~1 with the trt.
Thanks
[[alternative HTML version deleted]]