Homolog-level DGE analyses - what to consider?

Question

[ponganta]

Hello,

I'm currently working on tissue-specific RNAseq data of a non-model polyploid, which is however closely related to a model species. My goal is to identify tissue-specific DEGs via pairwise comparisons.

I approached this task by:

1. de novo assembly of transcripts
2. annotation
3. quantification (Salmon)
4. Count aggregation on a homolog level (tximport)
5. Determining tissue-specific DEGs (using DESeq2::DESeq and DESeq2::lfcShrink)

My question is:

Does anyone have remarks on what is to be considered when dealing with homolog-level DGE analysis? Any pitfalls to watch out for? Transcriptome assembly and annotation are, as far as I can tell, of fairly high quality (transrate score: 0.4, mean bitscore: 640).

I'd be thankful for any insights!