Hi,

I have 31 samples from three different breed groups where I am studying the effect of diet on breeds. I constructed multi-factor designs for DESeq2 and EdgeR as follows:

DESeq2:

EdgeR:

Could anyone please tell me what made such a big difference in results from two different programs? I would appreciate your help.

Thank you very much!

I have 31 samples from three different breed groups where I am studying the effect of diet on breeds. I constructed multi-factor designs for DESeq2 and EdgeR as follows:

countdata<-read.csv("htseq-merged_table.csv", row.names=1)

countdata<-countdata[1: (nrow(countdata)-5),]

sampleinfo<-read.csv("SampleTable_mod.csv")

sample<-sampleinfo$Sample

breed<-sampleinfo$Breed

diet<-sampleinfo$Diet

sampletable<-data.frame(Sample=sample, breed=breed, diet=diet)

countdata<-countdata[1: (nrow(countdata)-5),]

sampleinfo<-read.csv("SampleTable_mod.csv")

sample<-sampleinfo$Sample

breed<-sampleinfo$Breed

diet<-sampleinfo$Diet

sampletable<-data.frame(Sample=sample, breed=breed, diet=diet)

colnames(countdata)<-sampletable$Sample

ddsMat<-DESeqDataSetFromMatrix(countData=countdata, colData = sampletable, design = ~breed+diet+breed:diet )

dds<-DESeq(ddsMat)

resultsNames(dds)

#[1] "Intercept" "breedFIN" "breedTEX" "breedFXT" "dietFLU"

#[6] "dietCON" "breedFIN.dietFLU" "breedTEX.dietFLU" "breedFXT.dietFLU" "breedFIN.dietCON"

#[11] "breedTEX.dietCON" "breedFXT.dietCON"

FINTEXFlu<-results(dds, contrast=list("breedFIN.dietFLU","breedTEX.dietFLU"))

FINTEXFluSig<-subset(FINTEXFlu, padj < 0.1) #(36)

FINTEXFluSig<-FINTEXFluSig[abs(FINTEXFluSig$log2FoldChange) >=1,]

nrow(FINTEXFluSig)

25

ddsMat<-DESeqDataSetFromMatrix(countData=countdata, colData = sampletable, design = ~breed+diet+breed:diet )

dds<-DESeq(ddsMat)

resultsNames(dds)

#[1] "Intercept" "breedFIN" "breedTEX" "breedFXT" "dietFLU"

#[6] "dietCON" "breedFIN.dietFLU" "breedTEX.dietFLU" "breedFXT.dietFLU" "breedFIN.dietCON"

#[11] "breedTEX.dietCON" "breedFXT.dietCON"

FINTEXFlu<-results(dds, contrast=list("breedFIN.dietFLU","breedTEX.dietFLU"))

FINTEXFluSig<-subset(FINTEXFlu, padj < 0.1) #(36)

FINTEXFluSig<-FINTEXFluSig[abs(FINTEXFluSig$log2FoldChange) >=1,]

nrow(FINTEXFluSig)

25

EdgeR:

group<-factor(paste(breed, diet, sep="."))

ED<-DGEList(counts=countdata, group=group)

Y<-calcNormFactors(ED)

design<-model.matrix(~0+group)

colnames(design)<-levels(group)

Y<-estimateGLMCommonDisp(Y, verbose=TRUE)

Y<-estimateGLMTrendedDisp(Y, design)

Y<-estimateGLMTagwiseDisp(Y, design)

fit<-glmFit(Y, design)

mycontrast<-makeContrasts(FINF.TEXF=FIN.FLU-TEX.FLU, FINF.FXTF=FIN.FLU-FXT.FLU, TEXF.FXTF=TEX.FLU-FXT.FLU, levels=design)

FINTEXF<-(glmLRT(fit, contrast=mycontrast[,"FINF.TEXF"]))

summary(Y<-decideTestsDGE(FINTEXF, p=0.1)) # (-98,+480)

ttFINTEXF<-topTags(FINTEXF, n=sum(98+480))

sigFINTEXF<-ttFINTEXF[abs(ttFINTEXF$table$logFC) >=1.0,]

nrow(sigFINTEXF)

519

ED<-DGEList(counts=countdata, group=group)

Y<-calcNormFactors(ED)

design<-model.matrix(~0+group)

colnames(design)<-levels(group)

Y<-estimateGLMCommonDisp(Y, verbose=TRUE)

Y<-estimateGLMTrendedDisp(Y, design)

Y<-estimateGLMTagwiseDisp(Y, design)

fit<-glmFit(Y, design)

mycontrast<-makeContrasts(FINF.TEXF=FIN.FLU-TEX.FLU, FINF.FXTF=FIN.FLU-FXT.FLU, TEXF.FXTF=TEX.FLU-FXT.FLU, levels=design)

FINTEXF<-(glmLRT(fit, contrast=mycontrast[,"FINF.TEXF"]))

summary(Y<-decideTestsDGE(FINTEXF, p=0.1)) # (-98,+480)

ttFINTEXF<-topTags(FINTEXF, n=sum(98+480))

sigFINTEXF<-ttFINTEXF[abs(ttFINTEXF$table$logFC) >=1.0,]

nrow(sigFINTEXF)

519

Thank you very much!

## Comment