def process_clinical_data(input_dir: str, output_dir: str) -> List[str]:
"""
Reads subject registry data from input_dir and transforms the data
to clinical data files for cBioPortal.
Writes the generated data files to output_dir.
Returns the list of sample identifiers in the clinical data.
"""
# Clinical data
subject_registry_reader = SubjectRegistryReader(input_dir)
subject_registry: CentralSubjectRegistry = subject_registry_reader.read_subject_registry(
)
# Transform patient file
patient_clinical_data, patient_clinical_header = transform_patient_clinical_data(
subject_registry)
write_clinical(patient_clinical_data, patient_clinical_header, 'patient',
output_dir, STUDY_ID)
# Transform sample file
sample_clinical_data, sample_clinical_header = transform_sample_clinical_data(
subject_registry)
write_clinical(sample_clinical_data, sample_clinical_header, 'sample',
output_dir, STUDY_ID)
return sample_clinical_data['SAMPLE_ID'].unique().tolist()
def csr2transmart(input_dir: str, output_dir: str, config_dir: str,
study_id: str, top_tree_node: str):
logger.info('csr2transmart')
try:
logger.info('Reading configuration data...')
ontology_config = read_configuration(config_dir)
logger.info('Reading CSR data...')
subject_registry_reader = SubjectRegistryReader(input_dir)
subject_registry: CentralSubjectRegistry = subject_registry_reader.read_subject_registry(
)
study_registry_reader = StudyRegistryReader(input_dir)
study_registry: StudyRegistry = study_registry_reader.read_study_registry(
)
logger.info('Mapping CSR to Data Collection...')
mapper = CsrMapper(study_id, top_tree_node)
collection: DataCollection = mapper.map(subject_registry,
study_registry,
ontology_config.nodes)
logger.info('Writing files to {}'.format(output_dir))
copy_writer = TransmartCopyWriter(str(output_dir))
copy_writer.write_collection(collection)
logger.info('Done.')
except Exception as e:
logger.error(e)
sys.exit(1)
def mapped_data_collection() -> DataCollection:
input_dir = './test_data/input_data/CSR2TRANSMART_TEST_DATA'
config_dir = './test_data/input_data/config'
study_id = 'CSR'
top_tree_node = '\\Central Subject Registry\\'
ontology_config: OntologyConfig = read_configuration(config_dir)
subject_registry_reader = SubjectRegistryReader(input_dir)
subject_registry: CentralSubjectRegistry = subject_registry_reader.read_subject_registry(
)
study_registry_reader = StudyRegistryReader(input_dir)
study_registry: StudyRegistry = study_registry_reader.read_study_registry()
mapper = CsrMapper(study_id, top_tree_node)
return mapper.map(subject_registry, study_registry, ontology_config.nodes)
def test_serialising_and_deserialising_biomaterials(
registry_with_biosources_and_biomaterials, tmp_path):
subject_registry = add_ngs_data(registry_with_biosources_and_biomaterials,
'./test_data/input_data/CLINICAL')
writer = SubjectRegistryWriter(tmp_path.as_posix())
writer.write(subject_registry)
reader = SubjectRegistryReader(tmp_path.as_posix())
subject_registry = reader.read_subject_registry()
assert len(subject_registry.entity_data['Biomaterial']) == 2
assert Counter(subject_registry.entity_data['Biomaterial']
[0].library_strategy) == Counter({
'CNV': 2,
'SNV': 1
})
def sample_clinical_data() -> pd.DataFrame:
input_dir = './test_data/input_data/CSR2CBIOPORTAL_TEST_DATA'
subject_registry_reader = SubjectRegistryReader(input_dir)
subject_registry: CentralSubjectRegistry = subject_registry_reader.read_subject_registry()
return transform_sample_clinical_data(subject_registry)[0]