def merge_annotated_clusters(
biggest: pysam.AlignedSegment,
other: pysam.AlignedSegment) -> pysam.AlignedSegment:
"""Merges 2 annotated clusters together.
Merges 2 annotated aligned segments, each representing a cluster. Merges the
smaller into the larger. Adds the read number of the 2nd cluster to the first.
Args:
biggest: The larger of the 2 clusters, with a higher read number.
other: The smaller of the 2 clusters, with a lower read number.
Returns:
The annotated aligned segment representing the merged cluster.
"""
merged_id = biggest.get_tag(CLUSTER_ID_TAG)
if not merged_id.endswith("+"):
merged_id = merged_id + "+"
biggest.set_tag(CLUSTER_ID_TAG, merged_id, "Z")
total_reads = biggest.get_tag(NUM_READS_TAG) + other.get_tag(NUM_READS_TAG)
biggest.set_tag(NUM_READS_TAG, total_reads, "i")
return biggest
def set_qc_fail(rec: pysam.AlignedSegment, tool: Callable[..., Any],
reason: str) -> None:
"""Sets the QC fail flag, and adds tags containing the tool name and reason for failing.
Args:
rec: the record to fail
tool: the tool (as a callable) that failed this record
reason: the reason for failing
"""
assert '\t' not in reason, f"Reason may not contain tabs: {reason}"
rec.is_qcfail = True
rec.set_tag(QcFailToolTag, tool.__name__)
rec.set_tag(QcFailReasonTag, reason)
def _cleanup(rec: AlignedSegment, tags_to_invalidate: Iterable[str]) -> None:
"""Removes extended tags from a record that may have become invalid after clipping."""
for tag in tags_to_invalidate:
rec.set_tag(tag, None)
def add_tags(alignedSegment: pysam.AlignedSegment) -> pysam.AlignedSegment:
""" Takes an AlignedSegment and add percent identity and alignment length as tags
alignment length = MID
mismatches = NM
percent identity = (MID - NM) / MID
The percent identity is a value between 0.0 and 1.0
If the segment is unmapped then it is returned as with a percent identity of 0
and an alignment length of 0.
:param alignedSegment: The pysam AlignedSegment object
:return: alignedSegment: The updated pysam AlignedSegment object
"""
# Assuming that if the id tag is present that the other tags are also there.
if alignedSegment.has_tag('id'):
return alignedSegment
if alignedSegment.is_unmapped:
alignedSegment.set_tag('id', 0.0, 'f')
alignedSegment.set_tag('al', 0, 'i')
alignedSegment.set_tag('qc', 0.0, 'f')
return alignedSegment
alnlength = sum(alignedSegment.get_cigar_stats()[0][0:3])
query_covered_bases = sum(alignedSegment.get_cigar_stats()[0][0:2])
query_length = alignedSegment.infer_read_length()
mismatches = alignedSegment.get_tag('NM')
percid = (alnlength - mismatches) / float(alnlength)
qcov = query_covered_bases / float(query_length)
alignedSegment.set_tag('id', percid, 'f')
alignedSegment.set_tag('qc', qcov, 'f')
alignedSegment.set_tag('al', alnlength, 'i')
return alignedSegment