def set_pair_info(r1: AlignedSegment,
r2: AlignedSegment,
proper_pair: bool = True) -> None:
"""Resets mate pair information between reads in a pair. Requires that both r1
and r2 are mapped. Can be handed reads that already have pairing flags setup or
independent R1 and R2 records that are currently flagged as SE reads.
Args:
r1: read 1
r2: read 2 with the same queryname as r1
"""
assert not r1.is_unmapped, f"Cannot process unmapped mate {r1.query_name}/1"
assert not r2.is_unmapped, f"Cannot process unmapped mate {r2.query_name}/2"
assert r1.query_name == r2.query_name, f"Attempting to pair reads with different qnames."
for r in [r1, r2]:
r.is_paired = True
r.is_proper_pair = proper_pair
r1.is_read1 = True
r1.is_read2 = False
r2.is_read2 = True
r2.is_read1 = False
for src, dest in [(r1, r2), (r2, r1)]:
dest.next_reference_id = src.reference_id
dest.next_reference_start = src.reference_start
dest.mate_is_reverse = src.is_reverse
dest.mate_is_unmapped = False
dest.set_tag("MC", src.cigarstring)
insert_size = isize(r1, r2)
r1.template_length = insert_size
r2.template_length = -insert_size
def _make_read_unmapped(rec: AlignedSegment) -> None:
"""Removes mapping information from a read."""
if rec.is_reverse:
quals = rec.query_qualities
quals.reverse()
rec.query_sequence = dnautils.reverse_complement(rec.query_sequence)
rec.query_qualities = quals
rec.is_reverse = False
rec.reference_id = sam.NO_REF_INDEX
rec.reference_start = sam.NO_REF_POS
rec.cigar = None
rec.mapping_quality = 0
rec.template_length = 0
rec.is_duplicate = False
rec.is_secondary = False
rec.is_supplementary = False
rec.is_proper_pair = False
rec.is_unmapped = True
def _set_mate_info(self, r1: pysam.AlignedSegment,
r2: pysam.AlignedSegment) -> None:
"""Sets the mate information on a pair of sam records.
Handles cases where both reads are mapped, one of the two reads is unmapped or both reads
are unmapped.
Args:
r1: the first read in the pair
r2: the sceond read in the pair
"""
for rec in r1, r2:
rec.template_length = 0
rec.is_proper_pair = False
if r1.is_unmapped and r2.is_unmapped:
# If they're both unmapped just clean the records up
for rec, other in [(r1, r2), (r2, r1)]:
rec.reference_id = sam.NO_REF_INDEX
rec.next_reference_id = sam.NO_REF_INDEX
rec.reference_start = sam.NO_REF_POS
rec.next_reference_start = sam.NO_REF_POS
rec.is_unmapped = True
rec.mate_is_unmapped = True
rec.is_proper_pair = False
rec.mate_is_reverse = other.is_reverse
elif r1.is_unmapped or r2.is_unmapped:
# If only one is mapped/unmapped copy over the relevant stuff
(m, u) = (r1, r2) if r2.is_unmapped else (r2, r1)
u.reference_id = m.reference_id
u.reference_start = m.reference_start
u.next_reference_id = m.reference_id
u.next_reference_start = m.reference_start
u.mate_is_reverse = m.is_reverse
u.mate_is_unmapped = False
u.set_tag("MC", m.cigarstring)
m.next_reference_id = u.reference_id
m.next_reference_start = u.reference_start
m.mate_is_reverse = u.is_reverse
m.mate_is_unmapped = True
else:
# Else they are both mapped
for rec, other in [(r1, r2), (r2, r1)]:
rec.next_reference_id = other.reference_id
rec.next_reference_start = other.reference_start
rec.mate_is_reverse = other.is_reverse
rec.mate_is_unmapped = False
rec.set_tag("MC", other.cigarstring)
if r1.reference_id == r2.reference_id:
r1p = r1.reference_end if r1.is_reverse else r1.reference_start
r2p = r2.reference_end if r2.is_reverse else r2.reference_start
r1.template_length = r2p - r1p
r2.template_length = r1p - r2p
# Arbitrarily set proper pair if the we have an FR pair with isize <= 1000
if r1.is_reverse != r2.is_reverse and abs(
r1.template_length) <= 1000:
fpos, rpos = (r2p, r1p) if r1.is_reverse else (r1p, r2p)
if fpos < rpos:
r1.is_proper_pair = True
r2.is_proper_pair = True