def accuracy(ref, seq, balanced=False, min_coverage=0.0):
"""
Calculate the accuracy between `ref` and `seq`
"""
alignment = parasail.sw_trace_striped_32(seq, ref, 8, 4, parasail.dnafull)
counts = defaultdict(int)
q_coverage = len(alignment.traceback.query) / len(seq)
r_coverage = len(alignment.traceback.ref) / len(ref)
if r_coverage < min_coverage:
return 0.0
_, cigar = parasail_to_sam(alignment, seq)
for count, op in re.findall(split_cigar, cigar):
counts[op] += int(count)
if balanced:
accuracy = (counts['='] - counts['I']) / (counts['='] + counts['X'] +
counts['D'])
else:
accuracy = counts['='] / (counts['='] + counts['I'] + counts['X'] +
counts['D'])
return accuracy * 100
def update(self, annotation, prediction):
alignment = parasail.sw_trace_striped_32(prediction.label,
annotation.label, 8, 4,
parasail.dnafull)
counts = defaultdict(int)
_, cigar = self._parasail_to_sam(alignment, prediction.label)
r_coverage = len(alignment.traceback.ref) / len(annotation.label)
if r_coverage < self.min_coverage:
self.accuracy.append(0.0)
return 0.0
for count, op in re.findall(split_cigar, cigar):
counts[op] += int(count)
if self.balanced:
accuracy = (counts['='] - counts['I']) / (
counts['='] + counts['X'] + counts['D'])
else:
accuracy = counts['='] / (counts['='] + counts['I'] + counts['X'] +
counts['D'])
self.accuracy.append(accuracy)
return accuracy
def print_alignment(ref, seq):
"""
Print the alignment between `ref` and `seq`
"""
alignment = parasail.sw_trace_striped_32(ref, seq, 8, 4, parasail.dnafull)
print(alignment.traceback.query)
print(alignment.traceback.comp)
print(alignment.traceback.ref)
print(" Score=%s" % alignment.score)
return alignment.score
def parasail_alignment(query, ref):
"""Run a Smith-Waterman alignment between two sequences.
:param query: the query sequence.
:param ref: the reference sequence.
:returns: reference start co-ordinate, cigar string
"""
result = parasail.sw_trace_striped_32(query, ref, 5, 3, parasail.dnafull)
rstart, cigar = medaka.smolecule.parasail_to_sam(result, query)
return rstart, cigar
def align_read_to_haps(read, haps, g_open=5, g_ext=3, matrix=parasail.dnafull):
"""Get trimmed reads without reference.
:param read: str trimmed read sequence
:param haps: (padded ref, padded alt 1, ... padded alt n)
:param g_open: int, gap opening penalty
:param g_ext: int, gap extend penalty
:param matrix: int matrix shape (16, 16) substitution matrix.
:returns: [int, scores]
"""
scores = []
for i, hap in enumerate(haps):
algn = parasail.sw_trace_striped_32(read, hap, g_open, g_ext, matrix)
scores.append(algn.score)
return scores
def accuracy(ref, seq, balanced=False):
"""
Calculate the accuracy between `ref` and `seq`
"""
alignment = parasail.sw_trace_striped_32(ref, seq, 8, 4, parasail.dnafull)
counts = defaultdict(int)
_, cigar = parasail_to_sam(alignment, seq)
for count, op in re.findall(split_cigar, cigar):
counts[op] += int(count)
if balanced:
accuracy = (counts['='] - counts['I']) / (counts['='] + counts['X'] + counts['D'])
else:
accuracy = counts['='] / (counts['='] + counts['I'] + counts['X'] + counts['D'])
return accuracy * 100
def pair_align(reference, query, query_name, subs_mat, params):
""" Perform pairwise local alignment using parsail-python """
aln = parasail.sw_trace_striped_32(query, reference, params['gap_open'],
params['gap_extend'], subs_mat)
return process_alignment(aln, query, query_name, params)
def align(r, h, go=5, ge=3, matrix=parasail.dnafull):
return parasail.sw_trace_striped_32(r, h, go, ge, matrix).score
