I am using below design for DESeq2 analysis, how I can change below PCA CODE to make PCA based on Treatment and compartment?

dds$group <- as.factor(paste(dds$Batch,dds$Compartment,dds5$Treatment,sep="."))
design(dds) <- ~ group

PCA plot code need to change? what should I put in intgroup?

rlddda<-rlogTransformation(dds,blind=TRUE, fitType='local')
PCA<-plotPCA(rlddds, intgroup = "Treatment", returnData = TRUE)
write.table(PCA, "DESeq2-pca-table-rld", row.names=TRUE, sep="\t");
pdf("DESeq2-pca-5Family-table-rld.pdf");
plotPCA(rld5Family, intgroup = "Treatment", returnData = FALSE)
dev.off();

Many thanks nabiyogesh

Hi, you literally just need to do:

plotPCA(rld5Family,
  intgroup = c('Treatment', 'Compartment'),
  returnData = FALSE)

That will colour the points in the bi-plot based on every possible combination of Treatment and Compartment.

You may have to change your design formula, though, as you're currently using a merged 'group' design of Batch, Compartment, and Treatment. If you do also have a batch effect in your data, you may want to remove that for downstream applications via limma::removeBatchEffect() applied to the rlog or variance-stabilised expression levels.

For more advanced analyses via PCA, try out my own Bioconductor package (with very helpful edits by Aaron), PCAtools: PCAtools: everything Principal Component Analysis.

Kevin