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(window):
continue
ref = self.ref_reader.query(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 query(self, region):
"""Returns an iterator for going through the reads in the region.
Args:
region: nucleus.genomics.v1.Range. The query region.
Returns:
An iterator over nucleus.genomics.v1.Read protos.
"""
# redacted
return (read for read in self.reads
if ranges.ranges_overlap(region, utils.read_range(read)))
def query(self, region):
"""Iterate over records overlapping a query region.
Args:
region: third_party.nucleus.protos.Range, query region.
Returns:
An iterator over third_party.nucleus.protos.Read
"""
# redacted
return (read for read in self.reads
if ranges.ranges_overlap(region, utils.read_range(read)))
def query(self, region):
"""Iterate over records overlapping a query region.
Args:
region: third_party.nucleus.protos.Range, query region.
Returns:
An iterator over third_party.nucleus.protos.Read
"""
# redacted
return (read for read in self.reads
if ranges.ranges_overlap(region, utils.read_range(read)))
def check_overlaps(chr1, start1, end1, chr2, start2, end2, expected):
nbp = end1 - start1
read = test_utils.make_read('A' * nbp,
chrom=chr1,
start=start1,
cigar='{}M'.format(nbp))
region = ranges.make_range(chr2, start2, end2)
self.assertEqual(utils.read_overlaps_region(read, region),
expected)
# This check ensures we get the same result calling ranges.ranges_overlap.
self.assertEqual(
ranges.ranges_overlap(region, utils.read_range(read)),
expected)
def variants_overlap(variant1, variant2):
"""Returns True if the range of variant1 and variant2 overlap.
This is equivalent to:
ranges_overlap(variant_range(variant1), variant_range(variant2))
Args:
variant1: third_party.nucleus.protos.Variant we want to compare for overlap.
variant2: third_party.nucleus.protos.Variant we want to compare for overlap.
Returns:
True if the variants overlap, False otherwise.
"""
return ranges.ranges_overlap(variant_range(variant1), variant_range(variant2))
def check_overlaps(chr1, start1, end1, chr2, start2, end2, expected): i1 = ranges.make_range(chr1, start1, end1) i2 = ranges.make_range(chr2, start2, end2) self.assertEquals(ranges.ranges_overlap(i1, i2), expected) self.assertEquals(ranges.ranges_overlap(i2, i1), expected)
def query(self, region):
return iter(variant for variant in self.variants
if ranges.ranges_overlap(
variant_utils.variant_range(variant), region))
def query(self, region):
return iter(
variant for variant in self.variants
if ranges.ranges_overlap(variant_utils.variant_range(variant), region)
)