Obviously, if your inputs are different, then the results are going to be different as well. For DESeq2, there's over 35000 genes going into the DE analysis, whereas for edgeR, there's less than half that number.
Clc genomics workbench fold change edger code#
MaxUp=pmax(resSigUp$MeanMut1,resSigUp$MeanMut2)įinally, the edgeR code for analysis 2: library("edgeR")įor starters, the filtering schemes are different. Res1=results(ddsFinal,contrast=c(0,1,-1),pAdjustMethod="BH", cooksCutoff=TRUE, independentFiltering=TRUE,alpha=0.01) Sum( rowSums(counts(dds_raw)) =0) #23622ĭds=DESeq(dds, minReplicatesForReplace=6) Write.table(counts(dds_raw),file="PatientCellLine_Mut_DESeq2_RawTableFull",quote=FALSE,row.names=TRUE, col.names=TRUE,sep="\t") "PatientCellLine_ARN_SEG","PatientCellLine_ARN_SEV")
Nom=c("PatientCellLine_ARN_COM","PatientCellLine_ARN_GAN","PatientCellLine_ARN_GEN","PatientCellLine_ARN_LEV","PatientCellLine_ARN_PAR","PatientCellLine_ARN_SAU","PatientCellLine_ARN_GRA","PatientCellLine_ARN_KOR","PatientCellLine_ARN_POP","PatientCellLine_ARN_RIP", MaxUp=pmax(resSigUp$MeanPatientCellLine,resSigUp$MeanCtrl) Ind_up_below100=which(as.vector(MaxUp) In both comparisons, I compare all the deregulated genes (top panel), genes with padj<=0.001 (middle panel) and genes with at least an average of 100 reads in one of the conditions (bottom panel). I attached the comparison of edgeR and DESeq2 for:ġ/ cell lines of patient (12 samples) vs cell lines of controls (3 samples): Ģ/ cell lines of patient with one specific mutation (6 samples) vs cell lines of patient with another specific mutation (6 samples): In this experiment, there are 3 conditions : cell lines of patient with one specific mutation (6 samples), cell lines of patient with another specific mutation (6 samples) and cell lines of controls (3 samples). I read that one can expect a concordance of 70-80% between both tools. I have analysed an experiment of ribodepleted samples using both DESeq2 and edgeR robust.
0 kommentar(er)
