Hello,

I have protein alignments and a phylogenetic tree, we suspect that for some species in a (more or less) conserved region, the sequences vary whereas at the same positions for other species this region is well conserved.

By looking at the alignment I found good candidates but I need a method and a metric to back this statement, and also possibly discover other regions with a similar effect at different positions, in different alignments.

I have few ideas on how to use the tree and windows over the alignment to find nodes most locally dissimilar to a consensus, but I imagine there is already a name and a method to do it.

It's maybe trivial but I don't find the right keywords to google (or to search is there is not already an answer to this problem here).

Thank you.