def test_missing_column():
reader = SourcesReader(
input_dir='./test_data/input_data/CLINICAL',
config_dir='./test_data/input_data/config/invalid_sources_config/missing_column')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
assert 'Column \'tumor_type\' not found' in str(excinfo.value)
def test_duplicate_identifier():
reader = SourcesReader(
input_dir='./test_data/input_data/CLINICAL',
config_dir='./test_data/input_data/config/invalid_sources_config/duplicate_identifier')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
assert 'Duplicate identifier' in str(excinfo.value)
def test_biomaterial_cannot_be_derived_from_self():
reader = SourcesReader(
input_dir='./test_data/input_data/CLINICAL',
config_dir='./test_data/input_data/config/invalid_sources_config/biomaterial_derived_from_self')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
assert 'Invalid data for Biomaterial with id BM6' in str(excinfo.value)
def test_invalid_number():
reader = SourcesReader(
input_dir='./test_data/input_data/CLINICAL',
config_dir='./test_data/input_data/config/invalid_sources_config/invalid_number')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
assert 'Invalid data for Biosource with id BS6' in str(excinfo.value)
def test_non_existing_file():
reader = SourcesReader(
input_dir='./test_data/input_data/CLINICAL',
config_dir='./test_data/input_data/config/invalid_sources_config/non_existing_file')
with pytest.raises(ReaderException) as excinfo:
reader.read_subject_data()
assert 'File not found' in str(excinfo.value)
def test_derived_values_in_sources():
reader = SourcesReader(
input_dir='./test_data/input_data/CLINICAL',
config_dir='./test_data/input_data/config/invalid_sources_config/derived_biomaterial_values')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
assert 'Derived value fields not allowed in source files' \
in str(excinfo.value)
def test_invalid_date():
reader = SourcesReader(
input_dir='./test_data/input_data/CLINICAL',
config_dir='./test_data/input_data/config/invalid_sources_config/invalid_date')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
assert 'Error parsing biomaterial_date from source biomaterial_with_invalid_date.tsv:biomaterial_date with id BM15'\
in str(excinfo.value)
def test_wrong_file_format():
reader = SourcesReader(
input_dir='./test_data/input_data/CLINICAL',
config_dir='./test_data/input_data/config/invalid_sources_config/wrong_file_format')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
# The columns in the biosource.csv file are not correctly parsed, resulting
# in a missing identifier column
assert 'Identifier column \'biosource_id\' not found' in str(excinfo.value)
def test_biomaterial_biosource_mismatch():
"""
Biomaterial BM2 is derived form BM1,
but originates from a different biosource.
"""
reader = SourcesReader(
input_dir='./test_data/input_data/bios_biom_mismatch_data/source_data',
config_dir='./test_data/input_data/bios_biom_mismatch_data/config')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
assert '' in str(excinfo.value)
def test_diagnosis_biosource_patient_mismatch():
"""
Individual contains P1 and P2.
Diagnosis contains D1 linked to P2.
Biosource contains BS1 linked to P1 and D1.
"""
reader = SourcesReader(
input_dir='./test_data/input_data/ind_bios_diag_mismatch_data/source_data',
config_dir='./test_data/input_data/ind_bios_diag_mismatch_data/config')
with pytest.raises(DataException) as excinfo:
reader.read_subject_data()
assert '' in str(excinfo.value)
def sources2csr(input_dir, output_dir, config_dir):
logger.info('sources2csr')
try:
reader = SourcesReader(input_dir=input_dir, config_dir=config_dir)
subject_registry = reader.read_subject_data()
add_derived_values(subject_registry)
subject_registry_writer = SubjectRegistryWriter(output_dir)
subject_registry_writer.write(subject_registry)
study_registry = reader.read_study_data()
study_registry_writer = StudyRegistryWriter(output_dir)
study_registry_writer.write(study_registry)
except Exception as e:
logger.error(e)
sys.exit(1)