How to Perform Differential Expression Analysis Between Two Species' Transcriptomes?

Question

[PALA]

I would like to compare the differentially expressed genes in the same tissue between two species, which belong to the same family but different genera. I have RNA-seq data from this tissue at various developmental stages for both species, and I’ve assembled the transcriptomes using de novo assembly with Trinity (one of the species also has a reference genome). Previously, I’ve come across methods for cross-species differential expression analysis that involve identifying orthologous genes between the species. However, I’m unsure how to proceed, especially since differential expression analysis typically requires a reference transcriptome, whereas identifying orthologous genes often results in protein sequences. I would appreciate guidance on the specific steps or tools needed for this process. Thank you!

[Mark]

I'm sorry but why would you perform the experiment without having an idea of how to perform the analysis?

[Michael]

Good question, my impression is though that at least 50% of experiments in biology are conducted like that.

Answer 1

[Michael]

The result will be very noisy and the process complex. You need to consider if it is worth pursuing in the sense that anyone including you will believe the results. In any case, you will be comparing apples and oranges or worse. Not having a genome sequence for all species makes it much worse. You could try the following:

• Annotate the transcripts with Trinotate and/or translate them into protein sequences
• Use OrthoFinder or the like to find orthologues. At this point, you would like to focus on 1:1 orthologues. Because you have only transcriptomes, this isn't possible. So you will eventually end up comparing genes with different copy numbers.
• Ignoring these downsides and profiting from having sequenced multiple stages, run conventional DE analysis _within_ each species contrasting different stages. I hope you have replicates.
• Compare the list of genes that come out across species using the orthologs. You can also plot time-series curves for selected genes where you scale expression values.
• Beware of genes having cyclic expression _within_ each life cycle stage.