def __init__(self, locus, index, start, stop, seq):
"""Build new HomAltAllele."""
self.locus = locus
self.index = index
self.start = start
self.stop = stop
self.seq = normalize_seq(seq)
def __init__(self, locus, index, start, stop, seq, phase=None):
"""Build new HetAltAllele."""
self.locus = locus
self.index = index
self.start = start
self.stop = stop
self.seq = normalize_seq(seq)
self.phase = phase
def build_match_strings(ref, start, stop, allele, mode='sensitive', debug=False):
"""Build allele matching strings."""
alts = allele.alts
if debug:
print(' Allele: start={}, stop={}, size={}, ref={}, seq={}'.format(
allele.start,
allele.stop,
allele.stop - allele.start,
allele.ref,
','.join(alts),
), file=sys.stderr)
super_ref = normalize_seq(ref[start:stop])
# Require reference matches within the wobble zone + padding built into each normalized allele
if mode == 'specific':
super_alleles = [normalize_seq(ref[start:allele.start] + alt + ref[allele.stop:stop]) for alt in alts]
elif mode == 'sensitive':
super_alleles = [
normalize_seq(
'*' * (allele.min_start - start)
+ ref[allele.min_start:allele.start]
+ alt
+ ref[allele.stop:allele.max_stop]
+ '*' * (stop - allele.max_stop)
) for alt in alts
]
else:
raise ValueError(f'invalid match mode specified: {mode}')
if debug:
print(' MODE:', mode, file=sys.stderr)
print(' SUPER ALLELES:', super_alleles, file=sys.stderr)
print(' SUPER REF: ', super_ref, file=sys.stderr)
assert all(len(a) == stop - start - len(allele.ref) + len(alt) for a, alt in zip(super_alleles, alts))
return super_ref, super_alleles
def __init__(self, start, stop, seq):
self.start = start
self.stop = stop
self.seq = normalize_seq(seq)
def seq(self):
return normalize_seq(self.ref[self.start:self.stop])
def __init__(self, start, stop, seq, phase=None):
self.start = start
self.stop = stop
self.seq = normalize_seq(seq)
self.phase = phase
def find_allele(ref, allele, superlocus, debug=False):
# FASTPATH: Avoid constructing the graph match if the allele and the superlocus
# match perfectly.
if (len(superlocus) == 1 and allele.start == superlocus[0].start
and allele.stop == superlocus[0].stop
and allele.alleles[1] in superlocus[0].alleles[1:]
and 'PASS' in superlocus[0].record.filter):
i = superlocus[0].alleles.index(allele.alleles[1])
z = superlocus[0].allele_indices.count(i)
return z
# Bounds come from normalized extremes
start, stop = get_superlocus_bounds([[allele], superlocus])
if debug:
print(' Allele: start={}, stop={}, size={}, seq={}'.format(allele.start, allele.stop, allele.stop-allele.start, allele.alleles[1]), file=sys.stderr)
# Require reference matches within the wobble zone + padding built into each normalized allele
super_allele = ('*' * (allele.min_start - start)
+ ref[allele.min_start:allele.start]
+ allele.alleles[1]
+ ref[allele.stop:allele.max_stop]
+ '*' * (stop - allele.max_stop))
super_allele = normalize_seq(super_allele)
assert len(super_allele) == stop - start - len(allele.alleles[0]) + len(allele.alleles[1])
# Create genotype sets for each superlocus
try:
graph, constraints = generate_graph(ref, start, stop, superlocus, debug)
graph = list(graph)
if debug:
for i, (start, stop, alleles) in enumerate(graph):
print(' GRAPH{:02d}: start={}, stop={}, alleles={}'.format(i, start, stop, alleles), file=sys.stderr)
print(file=sys.stderr)
paths = generate_paths(graph, debug=debug)
if debug:
paths = list(paths)
for i, p in enumerate(paths):
print(' PATH{:02d}: {}'.format(i, p), file=sys.stderr)
print(file=sys.stderr)
except OverlapError:
return None
# Generate the set of diploid genotypes (actually haplotypes)
genos = set(generate_genotypes(paths, constraints, debug))
# Apply matcher to each pair of allele
matches = [(fancy_match(super_allele, a1), fancy_match(super_allele, a2))
for a1, a2 in genos]
# Find the highest zygosity
z = max(((a1 or 0) + (a2 or 0)) for a1, a2 in matches)
# If no match, check for the presense of any nocalls
if not z and any(None in m for m in matches):
z = None
if debug:
print(' ALLELE:{} {}'.format(len(super_allele), super_allele), file=sys.stderr)
for i, (g, m) in enumerate(zip(genos, matches)):
print(' GENO{:02d}:{} {}'.format(i, tuple(map(len, g)), g), file=sys.stderr)
print(' MATCH{:02d}: {}'.format(i, m), file=sys.stderr)
print(file=sys.stderr)
print(' ZYGOSITY: {}'.format(z), file=sys.stderr)
return z
def seq(self):
"""Return the sequence of this allele."""
return normalize_seq(self.ref[self.start:self.stop])
def test_normalize_seq():
"""Test sequence normalization."""
assert normalize_seq('') == ''
assert normalize_seq('ACGT') == 'ACGT'
assert normalize_seq('ACGTNacgtNRYSWKMBDHV') == 'ACGTNACGTNNNNNNNNNNN'
def test_normalize_seq():
assert normalize_seq('') == ''
assert normalize_seq('ACGT') == 'ACGT'
assert normalize_seq('ACGTNacgtNRYSWKMBDHV') == 'ACGTNACGTNNNNNNNNNNN'
def __init__(self, start, stop, seq):
self.start = start
self.stop = stop
self.seq = normalize_seq(seq)
def seq(self):
return normalize_seq(self.ref[self.start:self.stop])
def __init__(self, start, stop, seq, phase=None):
self.start = start
self.stop = stop
self.seq = normalize_seq(seq)
self.phase = phase