using composite transcriptome

[dannynielsen0]

Hello folks,
I am using salmon to quantify reads from RNAseq data that we generated from two closely related, non-model, species. Fortunately (and perhaps unfortunately!) we have good reference genomes for both species, and for which we have manually annotated detoxification related genes (these are of most interest to us). I have previously used hisat2 and htseq to align reads to either of the two genomes and then quantify. This has worked great, but the resulting count data differ depending on which species the reads are aligned to - no surprise here.

So my question/goal is: Salmon would allow me to use a composite transcriptome - made using cd-hit to cluster the transcriptomes from both species. I would like to utilize the decoy-aware mapping based approach, but am trying to decide how best to do this. As this requires mapping against a hard-masked genome from the organism.

I am interested if anyone has thoughts or experience working with similar data. What to do if you want to investigate differential expression between two closely related species, both of which you have well-annotated genomes?

Thanks in advance for any thoughts!