def make_msa_kalign(family, mate):
logging.info('Aligning with kalign.')
try:
# Import in the child process in case there's any issue in the .so with shared state between
# processes (maybe not possible, but just in case).
from kalign import kalign
except ImportError:
logging.critical('Error importing kalign module. Check that the submodule is installed properly.')
raise
seqs = [pair['seq'+mate] for pair in family]
aligned_seqs = kalign.align(seqs)
return aligned_seqs
def make_msa_kalign(family, mate):
logging.info('Aligning with kalign.')
try:
# Import in the child process in case there's any issue in the .so with shared state between
# processes (maybe not possible, but just in case).
from kalign import kalign
except ImportError:
logging.critical('Error importing kalign module. Check that the submodule is installed properly.')
raise
seqs = [pair['seq'+mate] for pair in family]
aln_struct = kalign.align(seqs)
return [aln_struct.seqs[i] for i in range(aln_struct.nseqs)]
def realign_family_to_consensus(consensus, family, quals, validate=False):
unaligned_seqs = [consensus]
unaligned_seqs.extend([seq.replace('-', '') for seq in family])
aligned_seqs = kalign.align(unaligned_seqs)
if validate:
for input, output in zip(unaligned_seqs, aligned_seqs):
assert input == output.replace('-', ''), (
"Kalign may have returned sequences in a different order than they were given. Failed on "
"sequence: {}".format(input)
)
aligned_consensus = aligned_seqs[0]
aligned_family = aligned_seqs[1:]
aligned_quals = seqtools.transfer_gaps_multi(quals, aligned_family, gap_char_out=' ')
return aligned_consensus, aligned_family, aligned_quals