def get_release_date(data: Submission) -> date:
if 'study' in data.get_entity_types():
for study in data.get_entities('study'):
if 'release_date' in study.attributes:
release_date = date.fromisoformat(study.attributes['release_date'])
if release_date > date.today():
return release_date
def get_ena_files(self, data: Submission) -> Dict[str, Tuple[str, str]]:
ena_files = {}
for entity_type, converter in self.conversion_map:
ena_type = converter.ena_type.upper()
ena_set = etree.XML(f'<{ena_type}_SET />')
for entity in data.get_entities(entity_type):
ena_conversion = self.convert_entity(converter, data, entity)
ena_set.append(ena_conversion)
if len(ena_set) > 0:
ena_files[ena_type] = (f'{ena_type}.xml', self.make_ena_file(ena_set))
return ena_files
def validate_data(self, data: Submission):
for entity_type, converter in self.converter.conversion_map:
ena_type = converter.ena_type.upper()
entities = data.get_entities(entity_type)
logging.info(f'Validating {len(entities)} {entity_type}(s) against ENA {ena_type} schema')
for entity in entities:
schema = self.ena_schema[ena_type]
ena_set = etree.XML(f'<{ena_type}_SET />')
ena_set.append(self.converter.convert_entity(converter, data, entity))
if not schema(ena_set):
self.add_errors(schema, ena_type, entity_type, entity)
def make_manifests(self, submission: Submission) -> Dict[str, str]:
manifests = {}
for run_experiment in submission.get_entities('run_experiment'):
samples = submission.get_linked_entities(run_experiment, 'sample')
studies = submission.get_linked_entities(run_experiment, 'study')
if len(samples) == 1 and len(studies) == 1 and 'uploaded_file_1' in run_experiment.attributes:
sample = samples.pop()
study = studies.pop()
sample_accession = sample.get_first_accession(SAMPLE_ACCESSION_PRIORITY)
study_accession = study.get_accession('ENA_Study')
if sample_accession and study_accession:
file_name, content = self.make_manifest(run_experiment, sample_accession, study_accession)
manifests[file_name] = content
return manifests
def validate_data(self, data: Submission):
entities = data.get_entities('run_experiment')
logging.info(f'Validating file checksums for {len(entities)} run(s)')
for entity in entities:
self.validate_entity(entity)
def validate_data(self, data: Submission):
entities = data.get_entities('sample')
logging.info(f'Validating taxonomy against scientific name in {len(entities)} sample(s)')
for entity in entities:
self.validate_entity(entity)