def _aggregate(self, covgs, threshold=5):
del_phylo_groups = []
for phylo_group, covg_dict in covgs.items():
percent_coverage = covg_dict["percent_coverage"]
length = covg_dict["length"]
bases_covered = self._bases_covered(percent_coverage, length)
total_bases = covg_dict["total_bases"]
total_percent_covered = round(bases_covered / total_bases, 3)
_median = covg_dict.get("median", [0])
minimum_percentage_coverage_required = percent_coverage_from_expected_coverage(
self.expected_depth) * self.threshold.get(
phylo_group, DEFAULT_THRESHOLD)
if total_percent_covered < minimum_percentage_coverage_required or median(
_median) < 0.1 * self.expected_depth:
# Remove low coverage nodes
_index = [
i for i, d in enumerate(_median)
if d > 0.1 * self.expected_depth
]
percent_coverage = [percent_coverage[i] for i in _index]
length = [length[i] for i in _index]
bases_covered = self._bases_covered(percent_coverage, length)
_median = [_median[i] for i in _index]
total_percent_covered = round(bases_covered / total_bases, 3)
if total_percent_covered > threshold:
if phylo_group == "Mycobacterium_llatzerense": # Mistake in panel
phylo_group = "Mycobacterium_mucogenicum"
covgs[phylo_group] = {
"percent_coverage": total_percent_covered,
"median_depth": median(_median)
}
else:
del_phylo_groups.append(phylo_group)
for phylo_group in del_phylo_groups:
del covgs[phylo_group]
def _log_post_het_or_alt(self, llk, expected_depth, sequence_coverage):
expected_percentage_coverage = percent_coverage_from_expected_coverage(
expected_depth)
minimum_percentage_coverage_required = expected_percentage_coverage * \
sequence_coverage.percent_coverage_threshold
if sequence_coverage.percent_coverage > minimum_percentage_coverage_required:
return self._log_post_hom_ref(llk)
else:
return MIN_LLK
def hom_alt_lik(self, variant_probe_coverage):
if variant_probe_coverage.alternate_percent_coverage < 100 * \
percent_coverage_from_expected_coverage(max(self.expected_depths)):
return MIN_LLK
else:
hom_alt_liks = []
# Either alt+cov or alt_covg + contam_covg
for expected_depth in self.expected_depths:
hom_alt_liks.append(
log_lik_R_S_coverage(
variant_probe_coverage.alternate_median_depth,
variant_probe_coverage.reference_median_depth,
expected_depth, expected_depth * self.error_rate / 3))
for contamination in self.contamination_depths:
hom_alt_liks.append(
log_lik_R_S_coverage(
variant_probe_coverage.alternate_median_depth,
variant_probe_coverage.reference_median_depth,
expected_depth + contamination,
(expected_depth + contamination) *
self.error_rate / 3))
return max(hom_alt_liks)
def test_percentage_coverage():
assert percent_coverage_from_expected_coverage(
100) > percent_coverage_from_expected_coverage(10)
assert percent_coverage_from_expected_coverage(100) == 1
assert percent_coverage_from_expected_coverage(1) < 1