def test_parse_case_two_cases_ped():
# GIVEN ped lines from multiple families
case_lines = [
"#Family ID\tIndividual ID\tPaternal ID\tMaternal ID\tSex\tPhenotype",
"636808\tADM1059A1\t0\t0\t1\t1",
"636808\tADM1059A2\tADM1059A1\tADM1059A3\t1\t2",
"636808\tADM1059A3\t0\t0\t2\t1",
"636809\tADM1059A3\t0\t0\t2\t1",
]
# WHEN parsing case info
with pytest.raises(PedigreeError):
# THEN it should raise since there are multiple families
parse_ped(case_lines)
def test_parse_case_two_cases_ped():
# GIVEN ped lines from multiple families
case_lines = [
"#Family ID\tIndividual ID\tPaternal ID\tMaternal ID\tSex\tPhenotype",
"636808\tADM1059A1\t0\t0\t1\t1",
"636808\tADM1059A2\tADM1059A1\tADM1059A3\t1\t2",
"636808\tADM1059A3\t0\t0\t2\t1",
"636809\tADM1059A3\t0\t0\t2\t1",
]
# WHEN parsing case info
with pytest.raises(PedigreeError):
# THEN it should raise since there are multiple families
parse_ped(case_lines)
def test_parse_ped_file(ped_file):
# GIVEN a pedigree with three samples
with open(ped_file, 'r') as case_lines:
# WHEN parsing out relevant sample info
family_id, samples = parse_ped(case_lines)
# THEN it should return correct family id
assert family_id == '643594'
# THEN it should return correct number of individuals
assert len(samples) == 3
def test_parse_ped_file(ped_file):
# GIVEN a pedigree with three samples
with open(ped_file, 'r') as case_lines:
# WHEN parsing out relevant sample info
family_id, samples = parse_ped(case_lines)
# THEN it should return correct family id
assert family_id == '643594'
# THEN it should return correct number of individuals
assert len(samples) == 3
# THEN assert the sex has been converted
for sample in samples:
assert sample['sex'] in REV_SEX_MAP
def test_parse_ped_file(ped_file):
# GIVEN a pedigree with three samples
with open(ped_file, 'r') as case_lines:
# WHEN parsing out relevant sample info
family_id, samples = parse_ped(case_lines)
# THEN it should return correct family id
assert family_id == '643594'
# THEN it should return correct number of individuals
assert len(samples) == 3
# THEN assert the sex has been converted
for sample in samples:
assert sample['sex'] in REV_SEX_MAP
def test_parse_ped():
# GIVEN a pedigree with three samples
case_lines = [
"#Family ID\tIndividual ID\tPaternal ID\tMaternal ID\tSex\tPhenotype",
"636808\tADM1059A1\t0\t0\t1\t1",
"636808\tADM1059A2\tADM1059A1\tADM1059A3\t1\t2",
"636808\tADM1059A3\t0\t0\t2\t1",
]
# WHEN parsing out relevant sample info
family_id, samples = parse_ped(case_lines)
# THEN it should return correct family id
assert family_id == '636808'
# THEN it should return correct number of individuals
assert len(samples) == 3
def test_parse_ped():
# GIVEN a pedigree with three samples
case_lines = [
"#Family ID\tIndividual ID\tPaternal ID\tMaternal ID\tSex\tPhenotype",
"636808\tADM1059A1\t0\t0\t1\t1",
"636808\tADM1059A2\tADM1059A1\tADM1059A3\t1\t2",
"636808\tADM1059A3\t0\t0\t2\t1",
]
# WHEN parsing out relevant sample info
family_id, samples = parse_ped(case_lines)
# THEN it should return correct family id
assert family_id == '636808'
# THEN it should return correct number of individuals
assert len(samples) == 3
def case(context, vcf, vcf_sv, owner, ped, update, config, no_variants):
"""Load a case into the database"""
if config is None and ped is None:
click.echo("You have to provide either config or ped file")
context.abort()
config_data = yaml.load(config) if config else {}
if not config_data:
config_data['analysis_date'] = datetime.datetime.now()
if ped:
with open(ped, 'r') as f:
family_id, samples = parse_ped(f)
config_data['family'] = family_id
if 'gene_panels' in config_data:
log.debug("handle whitespace in gene panel names")
config_data['gene_panels'] = [
panel.strip() for panel in config_data['gene_panels']
]
config_data['default_gene_panels'] = [
panel.strip() for panel in config_data['default_gene_panels']
]
log.info("Use family %s" % config_data['family'])
# check if the analysis is from a newer analysis
adapter = context.obj['adapter']
if 'owner' not in config_data:
if not owner:
click.echo("You have to specify the owner of the case")
context.abort()
else:
config_data['owner'] = owner
existing_case = adapter.case(institute_id=config_data['owner'],
display_name=config_data['family'])
if existing_case:
new_analysisdate = config_data.get('analysis_date')
if new_analysisdate and new_analysisdate > existing_case[
'analysis_date']:
log.info("updated analysis - updating existing case")
# update by default!
update = True
else:
log.warning("Case already exists in database")
context.abort()
else:
log.info("Case does not exist in database")
config_data['vcf_snv'] = vcf if vcf else config_data.get('vcf_snv')
config_data['vcf_sv'] = vcf_sv if vcf_sv else config_data.get('vcf_sv')
config_data['owner'] = config_data.get('owner')
config_data['rank_score_threshold'] = config_data.get(
'rank_score_threshold', 5)
if not (config_data.get('vcf_snv') or config_data.get('vcf_sv')):
log.warn("Please provide a vcf file (use '--vcf')")
context.abort()
if not config_data.get('owner'):
log.warn("Please provide an owner for the case (use '--owner')")
context.abort()
try:
load_scout(adapter, config_data, ped=ped, update=update)
except (IntegrityError, ValueError, ConfigError, KeyError) as error:
log.exception(error)
context.abort()
