Approach for multi species comparative transcriptomics and differential expression analyses

Question

[fish_enthusiast]

Hello everyone,

I want to do differential expression analyses on six species of closely related fish (with no reference genomes) in the same family, where there are three clades which contain a large and a small species:

Clade 1: Species A (large) vs Species B (small)

Clade 2: Species C (large) vs Species D (small)

Clade 3: Species E (large) vs Species F (small)

I have muscle and liver tissue for 3 individuals per species. The idea is to try and find differentially expressed genes involved in body size that are common across all 3 clades (i.e. is there convergence on a macroevolutionary scale?).

I am worried that I will be comparing apples to oranges because they are different species, so my initial thought was to run de novo assemblies for each of the six species separately in Trinity (combining all the tissue and biological replicates per species). Then I read here that I should perhaps run TransDecoder to convert the output Trinity fastas to protein sequences, then run OrthoFinder to search for single copy orthologs between each species pair (i.e. A vs B). Then I think I can somehow do differential expression on the OrthoFinder output.

Am I on the right track? This is a very new field for me, so wanted to seek out some advice.

Thank you!

[GenoMax]

a large and a small species

For those of us who are not fish biologists what is the significance. We are talking about the size of the fish (seems to be the case) or their genomes?

[fish_enthusiast]

Ah, my apologies! I am talking about body size. The small bodied species exhibit miniaturization in body size. I am interested to see if there is convergence on a macroevolutionary scale of genes involved in miniaturized body size in this family of fish.

[GenoMax]

Following is simply aloud thinking. Is comparing two species for differential expression a valid experiment to determine differences in body sizes? Wouldn't the timing of gene expression be more important than their amount? Since these species are in the same clade they hopefully have similarly sized genomes.

[fish_enthusiast]

There is a similar study in nematodes which looked at gene expression levels between a large-bodied and small-bodied species to identify genes involved in body size evolution (https://academic.oup.com/gbe/article/15/4/evad063/7128094), so I think that comparing two different fish would be alright as well.

[GenoMax]

This bit from the paper you linked is important. Only the orthologs are being considered.

Of the coding genes (20,173 genes in C. elegans and 21,443 genes in C. inopinata), 12,399 genes were inferred to be one-to-one orthologs and used in subsequent analyses.

[fish_enthusiast]

Thanks! Yeah, that's why I was thinking to use OrthoFinder to get the orthologs, but still a little confused on how to go from that output to DE analysis. It looks like that paper used RSEM and edgeR. I was just curious if anyone here had any experience with that specific pipeline. I will likely have to dig into this paper a bit more.

[dthorbur]

Does each of your 6 fish species even have a transcriptome annotated?

If so, then a workflow could start with 2 different analyses: i) DE analysis of each species individually, ii) orthology assignment among all species. Then you could overlap the DE results with the one-to-one orthologs found among each species pair.

If not, you could feasibly use a single reference transcriptome of a similar species, but I wouldn't recommend this. Plus, you'd then lose the ability to identify one-to-one orthologs which underpins your potential results.

I have some other worries about your approach too. Are the samples for each species taking at the same life stage, and if so, what life stage have you sampled? Did you sample in the wild, or are these lab reared?

[fish_enthusiast]

I have not yet annotated the transcriptomes, but do plan to once the assemblies are complete.

All species are wild caught adults from the aquarium trade. I agree its not an ideal setup, but these fish don't breed in captivity and are found in remote tropical regions, making field collection difficult.

[dthorbur]

If you are using adults I don't see how adult DEGs can be associated with body size. Surely the vast majority DEGs associated with body size will expressed during development. I would suggest your data is not appropriate to ask the question you want to ask.

I'm not sure what would be a more appropriate question if you only have adults that have spent time in captivity.