Closed marybarker closed 1 year ago
Great idea! Yeah, it would be good to do Sankoff on a subset of the sites, leaving the remainder fixed. This will take a bit of extra thought(I've made an issue #55), since sites are not independent and hence it will require either a method to compute transition cost for fully resolved sequences or else a the full-length cost vector expansion, even when only a subset of the sites are being recomputed.
…ase of a tree shaped DAG, and added option to apply Sankoff only to a subset of DAG nodes