Question regarding finding orthologs using proteins of interest and OrthoFinder

Question

[Introvertebrate]

Hello all!

I am currently working on a project in which I am trying to identify orthologs in the proteome of my species using a set of 70 proteins of interest from another species. As an initial step toward identifying putative homologs (i.e., proteins with considerable sequence similarity), I first blasted these proteins of interest against my proteome and kept all hits with an Evalue <1E-10. I now want pull out all the actual orthologs from the proteins identified in my species with blast and was thinking of using OrthoFinder for this task. It looks like people mainly use OrthoFinder to identify orthologous groups and orthologs between the full proteomes of multiple species. My question is, would using OrthoFinder to identify orthologs between a fasta containing the sequences of the 70 proteins of interest and a second fasta containing the the protein sequences of the blast hits from my species be appropriate/accurate?

Thanks for any help!

Answer 1

[lieven.sterck]

it will technically work yes and might even give you "meaningful" results.

You have to understand however the severe bias you are creating. The reason most people use full proteomes is that it allows ORTHOFinder to have all info it needs to do its thing. If for some reason your initial filtering is not accurate (eg it includes some false positives) than it might get assigned simply because its correct 'partner' is not present and might thus go to the best available.

Perhaps you could into a more basic approach? some very basic clustering for instance?

[Introvertebrate]

Thanks for your response. That makes sense; I certainly don't want to introduce severe bias into my results. Do you mean using something like CD-HIT-EST-2D to cluster my blast dataset and proteins of interest based on similarity?