Closed Anaphory closed 7 years ago
The multiple sequence alignment tries iterative global alignment of the sequences in each descendant node in the tree (in random order), with the alignment penalties between larger groups being the best possible alignment between each possible pair in the group. This probably turns calculation-expensive quite fast.
I have extracted the core
online_pmi
functionality into a class and added a multiple sequence aligmnent merger to it. I'm not sure the data model is good yet, but it seems to work okay.