Hi,

I have been analysing RNA-seq data using DESEQ2, but keep being asked the same question. Can you / why haven't you incorporated the spike-ins for normalisation.

From my perspective, this isn't necessary due to the tests DESEQ2 runs, which consider composition bias/sequencing depth, which spike-in normalisation doesn't. Is this right? If not, does anyone know a way spike-ins can be incorporated into DESEQ2 analysis?

For context, I have been using the DESeqDataSetFromHTSeqCount function as I have read count files produced from my alignment script.

Thanks in advance, C

Yes, use controlGenes, run estimateSizeFactors() before running DESeq() and that gives you spike-in based scaling factors.

If there are large changes in the distribution of abundance across samples, then the methods that assume focal changes (the default library size correction) will not work. So this is why such experiments require additional information on what features can be considered unchanging.