def test_vcf():
"""Parse the files and add fill the mongo db."""
families_path = '/vagrant/scout/tests/vcf_examples'
families = {}
# print(families_path)
# print(os.path.exists(families_path))
# for file in index(families_path):
# print(file)
i = 0
for root, dirs, files in os.walk(families_path):
print('root: %s, dirs: %s , files: %s' % (str(root), str(dirs), str(files)))
for f in files:
print('File: %s' % f)
if os.path.splitext(f)[-1] == '.ped':
if i in families:
families[i]['ped'] = os.path.join(root, f)
else:
families[i] = {'ped' : os.path.join(root, f)}
if os.path.splitext(f)[-1] == '.vcf':
if i in families:
families[i]['vcf'] = os.path.join(root, f)
else:
families[i] = {'vcf' : os.path.join(root, f)}
i += 1
for i in families:
my_family = ped_parser.FamilyParser(families[i]['ped'])
print(my_family.make_json(), type(my_family.make_json()))
pp(families)
def get_family(args):
"""Return the family"""
family_type = 'ped'
family_file = args.family_file[0]
my_family_parser = parser.FamilyParser(family_file, family_type)
# Stupid thing but for now when we only look at one family
return my_family_parser.families.popitem()[1]
def test_standard_trio(self):
"""Test if the file is parsed in a correct way."""
family_parser = parser.FamilyParser(open(self.trio_file.name, 'r'))
assert family_parser.header == [
'family_id', 'sample_id', 'father_id', 'mother_id', 'sex',
'phenotype'
]
assert 'healthyParentsAffectedSon' in family_parser.families
assert set(['proband', 'mother', 'father']) == set(
family_parser.families['healthyParentsAffectedSon'].individuals.
keys())
assert set(['proband', 'mother', 'father']) == set(
family_parser.families['healthyParentsAffectedSon'].trios[0])
def test_alternative_parser():
"""Test parsing a ped file with alternative formatting."""
# test default
with codecs.open('tests/fixtures/alternative.ped', 'r') as handle:
family_parser = parser.FamilyParser(handle, family_type='alt')
# we've only loaded one family
ped = family_parser.families.values()[0]
assert ped.family_id == 'family_id'
assert len(ped.individuals) == 1
sample = ped.individuals.values()[0]
assert sample.extra_info['Capture_kit'] == 'Agilent_SureSelect.V5'
def get_cases(self, cases_path):
"""Take a case file and return the case on the specified format."""
########### Loop over the case folders. Structure is described in documentation ###########
for root, dirs, files in os.walk(cases_path):
if files:
ped_file = None
vcf_file = None
zipped_vcf_file = None
case = None
for file in files:
if os.path.splitext(file)[-1] == '.ped':
ped_file = os.path.join(root, file)
case_parser = ped_parser.FamilyParser(ped_file)
case = case_parser.get_json()[0]
if os.path.splitext(file)[-1] == '.vcf':
vcf_file = os.path.join(root, file)
if os.path.splitext(file)[-1] == '.gz':
if os.path.splitext(file)[0][-1] == '.gz':
zipped_vcf_file = os.path.join(root, file)
# If no vcf we search for zipped files
if not vcf_file:
vcf_file = zipped_vcf_file
# If ped and vcf are not found exit:
if not (ped_file and vcf_file):
raise SyntaxError(
'Wrong folder structure in vcf directories. '
'Could not find ped and/or vcf files. '
'See documentation.')
# Store the path to variants as case id:s:
case['id'] = case['family_id']
case['vcf_path'] = vcf_file
self._cases.append(case)
return
def test_standard_trio_proband_missing_column(self):
"""Test if the file is parsed in a correct way."""
with pytest.raises(WrongLineFormat):
family_parser = parser.FamilyParser(open(self.trio_file.name, 'r'))
def test_standard_trio_missing_father(self):
"""Test if the file is parsed in a correct way."""
with pytest.raises(PedigreeError):
family_parser = parser.FamilyParser(open(self.trio_file.name, 'r'))