def call_debruijn_graph(self, windows, reads):
"""Helper function to call debruijn_graph module."""
windows_haplotypes = []
# Build and process de-Bruijn graph for each window.
for window in windows:
if window.end - window.start > self.config.ws_config.max_window_size:
continue
if not self.ref_reader.is_valid_interval(window):
continue
ref = self.ref_reader.bases(window)
# redacted
dbg_reads = [
read for read in reads
if ranges.ranges_overlap(window, utils.read_range(read))
]
with timer.Timer() as t:
graph = debruijn_graph.build(ref, dbg_reads,
self.config.dbg_config)
graph_building_time = t.GetDuration()
if not graph:
candidate_haplotypes = [ref]
else:
candidate_haplotypes = graph.candidate_haplotypes()
if candidate_haplotypes and candidate_haplotypes != [ref]:
candidate_haplotypes_info = realigner_pb2.CandidateHaplotypes(
span=window, haplotypes=candidate_haplotypes)
windows_haplotypes.append(candidate_haplotypes_info)
self.diagnostic_logger.log_graph_metrics(window, graph,
candidate_haplotypes,
graph_building_time)
return windows_haplotypes
def read_span(self):
if self._read_span is None and self.reads:
spans = [utils.read_range(r) for r in self.reads]
self._read_span = ranges.make_range(spans[0].reference_name,
min(s.start for s in spans),
max(s.end for s in spans))
return self._read_span
def test_read_range(self):
"""Tests reads have their ranges calculated correctly."""
start = 10000001
read = test_utils.make_read('AAACAG',
chrom='chrX',
start=start,
cigar='2M1I3M',
quals=range(10, 16),
name='read1')
self.assertEquals(ranges.make_range('chrX', start, start + 5),
utils.read_range(read))
read = test_utils.make_read('AAACAG',
chrom='chrX',
start=start,
cigar='2M16D3M',
quals=range(10, 16),
name='read1')
self.assertEquals(ranges.make_range('chrX', start, start + 5 + 16),
utils.read_range(read))
def assign_reads_to_assembled_regions(assembled_regions, reads):
"""Assign each read to the maximally overlapped window.
Args:
assembled_regions: list[AssemblyRegion], list of AssemblyRegion to assign
reads to. Does not assume AssemblyRegion are sorted.
reads: iterable[learning.genomics.genomics.Read], to be processed. Does
not assume the reads are sorted.
Returns:
[AssemblyRegion], information on assigned reads for each assembled region.
list[learning.genomics.genomics.Read], the list of unassigned reads.
"""
regions = [ar.region for ar in assembled_regions]
unassigned_reads = []
for read in reads:
read_range = utils.read_range(read)
window_i = ranges.find_max_overlapping(read_range, regions)
if window_i is not None:
assembled_regions[window_i].add_read(read)
else:
unassigned_reads.append(read)
return unassigned_reads