def test_make_read(self):
bases = 'ACG'
quals = [30, 40, 50]
cigar = '3M'
mapq = 42
chrom = 'chr10'
start = 123
name = 'myname'
read = test_utils.make_read(bases,
quals=quals,
cigar=cigar,
mapq=mapq,
chrom=chrom,
start=start,
name=name)
self.assertEqual(read.aligned_sequence, bases)
self.assertEqual(read.aligned_quality, quals)
self.assertEqual(list(read.alignment.cigar), [
cigar_pb2.CigarUnit(operation_length=3,
operation=cigar_pb2.CigarUnit.ALIGNMENT_MATCH)
])
self.assertEqual(read.alignment.mapping_quality, mapq)
self.assertEqual(read.alignment.position.reference_name, chrom)
self.assertEqual(read.alignment.position.position, start)
self.assertEqual(read.fragment_name, name)
def _add_cigar_unit(self, cigar, cigar_op, cigar_len):
if cigar_len <= 0:
return
if cigar and cigar[-1].operation == cigar_op:
cigar[-1].operation_length += cigar_len
else:
cigar_unit = cigar_pb2.CigarUnit(operation=cigar_op,
operation_length=cigar_len)
cigar.extend([cigar_unit])
def to_cigar_unit(source):
"""Creates a cigar_pb2 CigarUnit from source.
This function attempts to convert source into a CigarUnit protobuf. If
source is a string, it must be a single CIGAR string specification like
'12M'. If source is a tuple or a list, must have exactly two elements
(operation_length, opstr). operation_length can be a string or int, and must
be >= 1. opstr should be a single character CIGAR specification (e.g., 'M').
If source is already a CigarUnit, it is just passed through unmodified.
Args:
source: many types allowed. The object we want to convert to a CigarUnit
proto.
Returns:
CigarUnit proto with operation_length and operation set to values from
source.
Raises:
ValueError: if source cannot be converted or is malformed.
"""
try:
if isinstance(source, cigar_pb2.CigarUnit):
return source
elif isinstance(source, six.string_types):
l, op = source[:-1], source[-1]
elif isinstance(source, (tuple, list)):
l, op = source
else:
raise ValueError('Unexpected source', source)
if isinstance(op, six.string_types):
op = CHAR_TO_CIGAR_OPS[op]
l = int(l)
if l < 1:
raise ValueError('Length must be >= 1', l)
return cigar_pb2.CigarUnit(operation=op, operation_length=int(l))
except (KeyError, IndexError):
raise ValueError(
'Failed to convert {} into a CigarUnit'.format(source))
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import itertools
from absl.testing import absltest
from absl.testing import parameterized
from third_party.nucleus.protos import cigar_pb2
from third_party.nucleus.util import cigar
_CIGAR_TUPLES_AND_CIGAR_UNITS = [
((1, 'M'),
cigar_pb2.CigarUnit(
operation=cigar_pb2.CigarUnit.ALIGNMENT_MATCH, operation_length=1)),
((2, 'I'),
cigar_pb2.CigarUnit(
operation=cigar_pb2.CigarUnit.INSERT, operation_length=2)),
((3, 'D'),
cigar_pb2.CigarUnit(
operation=cigar_pb2.CigarUnit.DELETE, operation_length=3)),
((4, 'N'),
cigar_pb2.CigarUnit(
operation=cigar_pb2.CigarUnit.SKIP, operation_length=4)),
((5, 'S'),
cigar_pb2.CigarUnit(
operation=cigar_pb2.CigarUnit.CLIP_SOFT, operation_length=5)),
((6, 'H'),
cigar_pb2.CigarUnit(
operation=cigar_pb2.CigarUnit.CLIP_HARD, operation_length=6)),
def trim_cigar(cigar, ref_trim, ref_length):
"""Trim a cigar string to a certain reference length.
Args:
cigar: list of `nucleus.protos.CigarUnit`s of the original read alignment.
ref_trim: integer. Number of reference bases to trim off the beginning of
the read.
ref_length: integer. Number of reference bases to cover with the read, the
middle part that is not trimmed from the start or end of the read.
Returns:
new_cigar: list of `nucleus.protos.CigarUnit`s of the final read alignment,
after the left and/or right have been trimmed off.
read_trim: The number of bases of the read that are trimmed off.
new_read_length: The number of bases of the read that remain after trimming.
This is different from the final number of reference bases; for example,
an insertion makes the read longer without affecting the reference.
"""
# First consume the ref until the trim is covered.
trim_remaining = ref_trim
# Then consume the ref until the ref_length is covered.
ref_to_cover_remaining = ref_length
read_trim = 0
new_cigar = []
new_read_length = 0
for cigar_unit in cigar:
c_operation_length = cigar_unit.operation_length
# Each operation moves forward in the ref, the read, or both.
advances_ref = cigar_unit.operation in cigar_utils.REF_ADVANCING_OPS
advances_read = cigar_unit.operation in cigar_utils.READ_ADVANCING_OPS
ref_step = c_operation_length if advances_ref else 0
# First, use up each operation until the trimmed area is covered.
if trim_remaining > 0:
if ref_step <= trim_remaining:
# Fully apply to the trim.
trim_remaining -= ref_step
read_trim += c_operation_length if advances_read else 0
continue
else:
# Partially apply to finish the trim.
ref_step -= trim_remaining
read_trim += trim_remaining if advances_read else 0
# If trim finishes here, the rest of the ref_step can apply to the
# next stage and count towards covering the given ref window.
c_operation_length = ref_step
trim_remaining = 0
# Once the trim is done, start applying cigar entries to covering the ref
# window.
if trim_remaining == 0:
if ref_step <= ref_to_cover_remaining:
# Fully apply to the window.
new_cigar.append(
cigar_pb2.CigarUnit(operation=cigar_unit.operation,
operation_length=c_operation_length))
ref_to_cover_remaining -= ref_step
new_read_length += c_operation_length if advances_read else 0
else:
# Partially apply to finish the window.
c_operation_length = ref_to_cover_remaining
new_cigar.append(
cigar_pb2.CigarUnit(operation=cigar_unit.operation,
operation_length=c_operation_length))
new_read_length += c_operation_length if advances_read else 0
ref_to_cover_remaining = 0
break
return new_cigar, read_trim, new_read_length