RNAseq analysis with deseq1, deseq2, and edgeR. Is the same number of replicates needed in each group?

Question

[arturo.marin]

Hi,

usually I using a pipeline that I programmed thanks to the book RNAseq by sample. For the differential expression analysis step this book provide 3 scripts, for deseq1, deseq2 and edgeR methods. All the data that I had analyzed until now have the same number of samples in the two groups. Now probably I will have data with different number of samples in each groups, so it is possible do the analysis with deseq1, deseq2 and edgeR or I need the same number of replicates in the groups?

[arturo.marin]

Ok, thank you so much Istvan Albert.

[Istvan Albert]

I want to also point out that deseq1 is considered obsolete at this time. The latest Bioconductor won't install it, and if you want to install all three into the same environment you'd probably get into trouble.

If you want to use deseq1 make a new, separate conda environment for deseq1 alone, then switch to that environment when running it.

Answer 1

[Istvan Albert]

The number of replicates do not have to match.

In the examples that you had the experimental design was specified as 3x3 meaning 3 vs 3, but you can just as well have other combinations, 3 by 6 etc

You can't have fewer than 2 replicates for deseq2 and edger though.