% of similarity to be orthologous

[Antony03]

Hello,

I have a philosophic question: What is the minimum similarity (% of amino acids) to consider two genes orthologous? Are there some studies about that?

Thank you

[Skiaphrene]

Hi Antony03,


Your question is perhaps not precise enough: do you mean %identity or %positivity (if you want to use a substitution matrix - See the BLAST glossary here if you need to)? Do you really want to look at the amino acid level and not the nucleotide level for comparing genes?

Also, your question is highly context-dependent! A given level of sequence similarity (as measured by identity, positivity, or anything else you can come up with) might be good evidence for orthology in one case (e.g. 80% identity between two sequences of enzyme proteins from two very distantly-related organisms) but not good evidence in another (e.g. 80% identity between two sequences of enzyme proteins from two strains of the same bacterial species - which sounds more like paralogs to me). I'd say you would have to relate your alignment results back to what you'd expect to see given the evolutionary relationships of the organisms you're looking at. Not something that's really captured by an e-value...

If you must have a guideline, the lab where I did my PhD advocated a sequence identity of >=30% over at least 80% of the length of the shortest of the 2 protein sequences as evidence of possible orthology, going down to 15% as a minimum mark, when comparing protein sequences between Prokaryotes. But I would not dare extrapolate that to Eukaryotes!

You might like this article on the "Twilight zone of protein sequence alignments".


Hope this helps a bit (maybe others will have other references?),

-- Alex

[rhinoceros]

Hello,

I have a philosophic question: What is the minimum similarity (% of amino acids) to consider two genes orthologous? Are there some studies about that?

Thank you

Like 0%? Some viral protein-coding genes are considered orthologous although no sequence similarity can be detected even after exhaustive psi-blast. However, higher-level structures like e.g. MCP-fold can still reveal common origin..