def _new_rec(self, name: str, chrom: str, start: int,
attrs: Optional[Dict[str, Any]]) -> AlignedSegment:
"""Generates a new AlignedSegment. Sets the segment up with the correct
header and adds the RG attribute if not contained in attrs.
Args:
name: the name of the read/template
chrom: the chromosome to which the read is mapped
start: the start position of the read on the chromosome
attrs: an optional dictionary of SAM attributes with two-char keys
Returns:
AlignedSegment: an aligned segment with name, chrom, pos, attributes the
read group, and the unmapped flag all set appropriately.
"""
if chrom is not sam.NO_REF_NAME and chrom not in self._seq_lookup:
raise ValueError(
f"{chrom} is not a valid chromosome name in this builder.")
rec = AlignedSegment(header=self._samheader)
rec.query_name = name
rec.reference_name = chrom
rec.reference_start = start
rec.mapping_quality = self.mapping_quality
if chrom == sam.NO_REF_NAME or start == sam.NO_REF_POS:
rec.is_unmapped = True
attrs = attrs if attrs else dict()
if "RG" not in attrs:
attrs["RG"] = self.rg_id()
rec.set_tags(list(attrs.items()))
return rec
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
AEQ(BZ.insertRegions, VZ.insertRegions)
def testRead(self):
BZR = self.BZ.read()
VZR = self.VZ.read()
EQ(BZR.basecalls(), VZR.basecalls())
# Mockup some bam records reflecting the internal "pulse BAM" spec
from mock import Mock
from pysam import AlignedSegment
from pbcore.io import BamAlignment
from PRmm.io.ZmwReadStitcherIO import StitchedZmw, FeatureDesc
pulsePeer = AlignedSegment()
pulsePeer.is_unmapped=True
pulsePeer.seq = "GATTACAGATTACA"
pulsePeer.qname = "FakePulseRead"
tags = dict(
RG="00000000",
np=1,
qs=0,
qe=14,
rq=0.80,
sn=[2.0, 3.0, 5.0, 6.0],
ip=[15]*14,
pw=[16]*14,
zm=42,
cx=2,
# Now, the pulse stuff
pc = " GAgggTTACAcccGATaaaTACA",
AEQ(BZ.insertRegions, VZ.insertRegions)
def testRead(self):
BZR = self.BZ.read()
VZR = self.VZ.read()
EQ(BZR.basecalls(), VZR.basecalls())
# Mockup some bam records reflecting the internal "pulse BAM" spec
from mock import Mock
from pysam import AlignedSegment
from pbcore.io import BamAlignment
from PRmm.io.ZmwReadStitcherIO import StitchedZmw, FeatureDesc
pulsePeer = AlignedSegment()
pulsePeer.is_unmapped = True
pulsePeer.seq = "GATTACAGATTACA"
pulsePeer.qname = "FakePulseRead"
tags = dict(
RG="00000000",
np=1,
qs=0,
qe=14,
rq=0.80,
sn=[2.0, 3.0, 5.0, 6.0],
ip=[15] * 14,
pw=[16] * 14,
zm=42,
cx=2,
# Now, the pulse stuff
pc=" GAgggTTACAcccGATaaaTACA",