def _organize_calls(out_file, hla_base, data):
"""Prepare genotype calls, reporting best call along with quality metrics.
"""
hla_truth = get_hla_truthset(data)
align_file = dd.get_align_bam(data)
sample = dd.get_sample_name(data)
with file_transaction(data, out_file) as tx_out_file:
with open(tx_out_file, "w") as out_handle:
writer = csv.writer(out_handle)
writer.writerow(["sample", "locus", "mismatches", "options", "alleles", "p-groups", "expected",
"validates"])
for genotype_file in glob.glob("%s.HLA-*.gt" % (hla_base)):
hla_locus = os.path.basename(genotype_file).replace(
"%s.hla.HLA-" % os.path.basename(align_file), "").replace(".gt", "")
with open(genotype_file) as in_handle:
total_options = set([])
for i, line in enumerate(in_handle):
_, aone, atwo, m = line.split("\t")[:4]
pgroups = (hla_groups.hla_protein(aone, data), hla_groups.hla_protein(atwo, data))
if i == 0:
call_alleles = [aone, atwo]
call_pgroups = pgroups
mismatches = m
total_options.add(pgroups)
if len(total_options) > 0:
truth_alleles = tz.get_in([sample, hla_locus], hla_truth, [])
writer.writerow([sample, hla_locus, mismatches, len(total_options),
";".join(call_alleles), ";".join(call_pgroups),
";".join(truth_alleles), _matches_truth(call_alleles, truth_alleles, data)])
return out_file
def _organize_calls(out_file, hla_base, data):
"""Prepare genotype calls, reporting best call along with quality metrics.
"""
hla_truth = get_hla_truthset(data)
sample = dd.get_sample_name(data)
with file_transaction(data, out_file) as tx_out_file:
with open(tx_out_file, "w") as out_handle:
writer = csv.writer(out_handle)
writer.writerow(["sample", "locus", "mismatches", "options", "alleles", "p-groups", "expected",
"validates"])
for genotype_file in glob.glob("%s.HLA-*.gt" % (hla_base)):
hla_locus = os.path.basename(genotype_file).replace(
"%s.HLA-" % os.path.basename(hla_base), "").replace(".gt", "")
with open(genotype_file) as in_handle:
total_options = set([])
for i, line in enumerate(in_handle):
_, aone, atwo, m = line.split("\t")[:4]
pgroups = (hla_groups.hla_protein(aone, data), hla_groups.hla_protein(atwo, data))
if i == 0:
call_alleles = [aone, atwo]
call_pgroups = pgroups
mismatches = m
total_options.add(pgroups)
if len(total_options) > 0:
truth_alleles = tz.get_in([sample, hla_locus], hla_truth, [])
writer.writerow([sample, hla_locus, mismatches, len(total_options),
";".join(call_alleles), ";".join(call_pgroups),
";".join(truth_alleles), matches_truth(call_alleles, truth_alleles, data)])
return out_file
def _matches_truth(call_alleles, truth_alleles, data):
"""Flexibly check if truth and call alleles match, using p-groups.
"""
if not truth_alleles:
return ""
else:
call_pgroups = [hla_groups.hla_protein(x, data) for x in call_alleles]
t_cmp = set([hla_groups.hla_protein(x, data) for x in truth_alleles])
c_cmp = set(call_pgroups + [x[:-1] for x in call_pgroups if x.endswith("P")])
return "yes" if len(t_cmp.intersection(c_cmp)) == len(set(call_pgroups)) else "no"
def matches_truth(call_alleles, truth_alleles, data):
"""Flexibly check if truth and call alleles match, using p-groups.
"""
if not truth_alleles:
return ""
else:
def _remove_p(x):
return x[:-1] if x.endswith("P") else x
t_cmp = set([_remove_p(hla_groups.hla_protein(x, data)) for x in truth_alleles])
c_cmp = set([_remove_p(hla_groups.hla_protein(x, data)) for x in call_alleles])
return "yes" if len(t_cmp.intersection(c_cmp)) == len(t_cmp) else "no"
def _matches_truth(call_alleles, truth_alleles, data):
"""Flexibly check if truth and call alleles match, using p-groups.
"""
if not truth_alleles:
return ""
else:
call_pgroups = [hla_groups.hla_protein(x, data) for x in call_alleles]
t_cmp = set([hla_groups.hla_protein(x, data) for x in truth_alleles])
c_cmp = set(call_pgroups +
[x[:-1] for x in call_pgroups if x.endswith("P")])
return "yes" if len(t_cmp.intersection(c_cmp)) == len(
set(call_pgroups)) else "no"