searching orthologous transcripts

Question

[Esaie]

Hello All, I'm searching orthology relations between transcripts in differents species. My questions is that : At what similarity percentages i can say that 2 transcripts are orthologous after having aligned them? Tank your for the ansews.

Answer 1

[Jean-Karim Heriche]

You can't define orthology from an alignment. By definition, orthology is a relationship on a tree (orthologs evolved from a common ancestor by speciation) so the proper way to find orthologs is to build a phylogenetic tree. Now some applications use heuristics based on alignments for scalability reasons. Depending on your data, you could
1- use an existing database of orthologous sequences,
2- add your sequences to existing phylogenetic trees e.g. in TreeFam.
3- build a phylogenetic tree
4- use one of the pairwise alignment-based pipelines e.g. OrthoMCL, InParanoid...

[Esaie]

Tank you, Exactly, i want to build a phylogenetic tree. Firstly, i do an sequences alignment, after that, I construct sequences cluster, the question is that: at what similarity level proteins can be consider to belong to the same cluster? I have saw that OrthoMCL use 50% of similatity level to include sequences in the same cluster OrthoMCL algo. Why the choose this similarity level? it's consistant? Tank you

Answer 2

[Antonio R. Franco]

There are not answer for that. Not an easy one.

You have very very conserved orthologous genes such as histones (higher than 95% homology), and you have other orthologous proteins such as the growth hormone that are very little conserved among species, even though they are probed orthologous proteins whose activity has been assayed and test it to be the same.

And you have other kind of orthologous proteins all in the middle range of homology between the two..