I have RNAseq read counts for genes in the plant plastid genome only, i.e. for 79 genes. Would you think it is appropriate to use DESeq2 to normalise the counts for library size when the number of genes is so small? I am comparing mutant with wild type, the two genotypes have massive differences in plastid gene expression.

We used custom code to obtain the counts for plastid genes only, and therefore I can't include all genes in the normalisation step and then filter for plastid genes.

```

The DESeq method works off the assumption that most genes are unchanging. If all your genes are expected to change, then this won't work.

Ideally, you have some genes in the mix you expect not to change, you can tell DEseq what genes to normalize with.

We used custom code to obtain the counts for plastid genes only

This reads to me that counts for all genes are available, hence I would run normalization on all genes, and then export the size factors to your subset of genes. I don't see what would be wrong with that. It is probably better than trying to work with just 79 genes for normalization, especially because you expect massive DE events.