Is BLASTp the quickest method to get a rough idea on ortholog from protein sequence?

Question

[greyman]

I know that when species are not very close to each other, hmmer will be a great tool to search for ortholog. However, I have more than 100 different types of gene of interest. I just want a really quick search to find "potential" orthologs from two organisms which belong to the same phylum. I wonder will blastp work if I apply stringent evalue cutoff 1e-10 and piden of 50? Any criticism, suggestion on other easy tools and opinion are welcome.

[natasha.sernova]

BlastX is a little bit better if I remember correctly See the example below:
BlastP vs BlastX

And a close question was asked some time ago

Ortholog Protein Finding Tool

Answer 1

[lieven.sterck]

if you want a quick and dirty approach then yes, blastP is the most easy/crude approach you can get indeed.

Moreover, being a blast hit is a per-requiste for any ortholog gene/protein. Usually then some more advances processing is done on the output to filter the correct or most likely orthologs.

[greyman]

Hi, I am not sure filtering by query coverage or the alignment length would be a better way, the qcovhsp i got are pretty low , lesser than 50 .... do you have any suggestion on downstream processing?

[lieven.sterck]

well, you can have a look at that link posted above. OrthoFinder is good way to go.

if you only stick to blast, then you can consider doing a best-reciprocal blast analysis (you blast A vs B and then B vs A, and if A->B gives the same tophit as B->A for a query gene, it's an extra indication already).

alternatively you can apply the LiRost criterion (not sure that is official name though), basically coming down to filtering on alignable region and %identity of it. From top of my head I think it's something like 150 residues length with a %ID of >30%

[greyman]

Thank you for the suggestion for blastp. For the suggestion on OrthoFInder, I have successfully run it and it generated a slightly different result than the reciprocal blast analysis, do I choose one that is both verified by blastp and OrthoFInder or simply rely on the output of OrthoFinder to be enough?

As I am only select a subset of genes from each species, will it introduces some bias in OrthoFinder? I was thinking of using blastp to filter the output from OrthoFinder, maybe blastp is unnecessary in this case?

[lieven.sterck]

orthofinder over reciprocal blast for me.

but indeed if you want to be extra conservative you can take the intersection, though that comes with additional concerns (and perhaps missed orthologs)