def align_to_ref(ref_seq, base_seq):
"""Aligns a sequence to the reference and caches the result for fast
lookup later. Returns a tuple (base_seq, ref_seq) properly aligned.
ref_seq -- The reference sequence to use as a guide.
query_seq -- The query sequence.
Returns:
query_aln -- The aligned query sequence.
ref_aln -- The aligned reference sequence.
"""
seqs = [('query', base_seq), ('ref', ref_seq)]
aligned = dict(GeneralSeqTools.call_muscle(seqs))
return aligned['query'], aligned['ref']
def get_region(seq, reference, regions = None):
if regions == None:
regions = [(300, 400)]
tmp_seqs = [('conc', reference),
('guess', seq)]
aligned = dict(GeneralSeqTools.call_muscle(tmp_seqs))
out = []
for _, start, stop in regions:
conc_pos = 0
align_start = None
for align_pos, l in enumerate(aligned['conc']):
if l != '-':
conc_pos += 1
if conc_pos == start:
align_start = align_pos
if conc_pos == stop:
align_stop = align_pos
break
yield seq[align_start:align_stop].replace('-', '')
def test_muscle_basic_call():
seqs = [('test1', 'ATCGATTGC'), ('test2', 'ATCGATGC')]
aln = [('test1', 'ATCGATTGC'), ('test2', 'ATCGA-TGC')]
res = list(GeneralSeqTools.call_muscle(seqs))
eq_(res, aln)