Can Paml Be Applied To Comparison Between Distantly Related Organisms?

Question

[Cacau]

I was thinking about whether it would be proper to use PAML to calculate the overall Ka/Ks value of a gene family (including eukaryotes, bacteria and archaea) that originated very early (in ancient bacteria). Also, can I test different models in different clades of this ancient gene family using PAML? Many thanks.

Answer 1

[14134125465346445]

Ka/Ks values saturate, specially after a certain number of synonymous mutations have occurred, due to the fact that after enough evolutionary time and high mutation rates, the same position in a sequence can mutate multiple times, and if the comparison is between very distant species, one won't be able to see the multiple events, but only the last difference in the comparison set.

There is no fix rule on how to run Ka/Ks, because different gene families have been seen to evolve at different rates. For example, myosin heavy chain gene families are used for evolutionary studies of deep nodes in the tree of life, because they have evolved very slowly along the years.

Also, if you have enough species in your dataset that split your long branches into shorter branches, that will increase the power of your comparison.

All said, I don't know of any gene family that can be compared across bacteria, archaea and eukarya that will give you meaningful Ka/Ks numbers. I think all genes might be too distant for that.

Answer 2

[Gabriel R.]

This question is not really bioinformatics related. It's more theoretical.

A basic question that comes to mind is that how would you compare really distant homologs ? How do you get proper alignments for such distant homologs ?