Background: I have a list of species that I want to put through homology inference. The goal of homology inference is to investigate the evolution of a trait on a species tree. I want to use the quantitative aspect of the data like CNV and recreate nodal CNV using cafe or count. And use gene trees of the homology to see if I can discern divergence patterns across gene families of interest.

The issue is some species only have RNA-seq generated transcriptions (peptides and cDNA) while others have genomes. The issue of missing introns aside, how can I circumvent this gap?

Should I:

  1. Just use the cDNA data with genomic data? compare nucleic acid to nucleic acid. I need to eventually make gene trees as well and as such MSA is important.The list of species is huge and I anticipate seg-faults during MSA if not during homology inference itself.

  2. Use emboss or something else to convert the nucleic acid data to peptides.And then compare peptides to peptides? Which makes the homology inference → MSA (hopefully pal2nal and gene tree inference later) a little easier. I have never done translation before so I do not know the pitfalls per se.

My list of species is green plants.

Can anyone share their experience if you have done something like this before?

  • 1
    $\begingroup$ I think you need to use proteins/amino acids. $\endgroup$
    – M__
    Commented Jan 17, 2023 at 22:09

1 Answer 1


Personally, I would use blastx against a local protein database. This converts query cDNA->amino acids and therefore combines points 1 and 2 without need to perform a translation in advance.

Genetic distance and saturation might become an issue here depending whether the search is outside the family and especially Order

MAFFT vs Muscle Super5. I would try "Muscle5" or rather the latest Muscle under the "Super5" algorithm. It outperforms everything.

  • 1
    $\begingroup$ Thanks for suggesting Muscle5, I will have to try it out this time. $\endgroup$
    – Sudoh
    Commented Jan 17, 2023 at 22:39
  • 2
    $\begingroup$ this is the right workflow. deleting my answer. $\endgroup$ Commented Jan 17, 2023 at 22:45
  • 2
    $\begingroup$ This, or tblastx to compare translated peptide which avoids the need for a local protein database. $\endgroup$
    – terdon
    Commented Jan 18, 2023 at 12:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.