Non-alphabet characters aren't counted as gaps when computing sequence coverage

[nikkithadani]

I'm not sure if this is something we want to fix or just something we want to note (can be a real issue for some viral sequences) but sequences with large numbers of Xs or other non-residue characters are not counted as gaps when applying the sequence coverage threshold.

EVcouplings/evcouplings/align/protocol.py

Line 911 in 594b45a

keep_seqs = (1 - ali.count("-", axis="seq")) >= min_cov

[thomashopf]

Will be updated in future release together with more flexible sequence weight calculation handling