def _assay_to_schema_name(assay_type, source_project):
'''
Given an assay name, and a source_project (may be None),
read all the schemas until one matches.
Return the schema name, but not the version.
>>> _assay_to_schema_name('PAS microscopy', None)
'stained'
>>> _assay_to_schema_name('snRNAseq', None)
'scrnaseq'
>>> _assay_to_schema_name('snRNAseq', 'HCA')
'scrnaseq-hca'
Or, if a match can not be found (try-except just for shorter lines):
>>> try: _assay_to_schema_name('PAS microscopy', 'HCA')
... except PreflightError as e: print(e)
No schema where 'PAS microscopy' is assay_type and 'HCA' is source_project
>>> try: _assay_to_schema_name('snRNAseq', 'Bad Project')
... except PreflightError as e: print(e)
No schema where 'snRNAseq' is assay_type and 'Bad Project' is source_project
>>> try: _assay_to_schema_name('Bad assay', None)
... except PreflightError as e: print(e)
No schema where 'Bad assay' is assay_type
>>> try: _assay_to_schema_name('Bad assay', 'HCA')
... except PreflightError as e: print(e)
No schema where 'Bad assay' is assay_type and 'HCA' is source_project
'''
for path in (Path(__file__).parent / 'table-schemas' /
'assays').glob('*.yaml'):
schema = load_yaml(path)
assay_type_enum = get_field_enum('assay_type', schema)
source_project_enum = get_field_enum('source_project', schema)
if assay_type not in assay_type_enum:
continue
if source_project_enum:
if not source_project:
continue
if source_project:
if not source_project_enum:
continue
if source_project not in source_project_enum:
continue
return re.match(r'.+(?=-v\d+)', path.stem)[0]
message = f"No schema where '{assay_type}' is assay_type"
if source_project is not None:
message += f" and '{source_project}' is source_project"
raise PreflightError(message)
def get_other_schema(schema_name, version, offline=None):
schema = load_yaml(_table_schemas_path / 'others' /
_get_schema_filename(schema_name, version))
names = [field['name'] for field in schema['fields']]
for field in schema['fields']:
_add_constraints(field,
optional_fields=[],
offline=offline,
names=names)
return schema
def _assay_to_schema_name(name):
'''
Given an assay name, read all the schemas until one matches.
Return the schema name, but not the version.
'''
for path in (Path(__file__).parent / 'table-schemas' / 'assays').glob('*.yaml'):
schema = load_yaml(path)
for field in schema['fields']:
if field['name'] == 'assay_type' and name in field['constraints']['enum']:
return path.stem.split('-v')[0]
raise PreflightError(f"Can't find schema where '{name}' is in the enum for assay_type")
def get_table_schema(schema_name, version, optional_fields=[], offline=None):
schema = load_yaml(_table_schemas_path / 'assays' /
f'{schema_name}-v{version}.yaml')
for field in schema['fields']:
_add_level_1_description(field)
_validate_level_1_enum(field)
_add_constraints(field, optional_fields, offline=offline)
_validate_field(field)
return schema
def get_directory_schema(directory_type):
schema = load_yaml(_directory_schemas_path / f'{directory_type}.yaml')
schema += [{
'pattern': r'extras/.*',
'description': 'Free-form descriptive information supplied by the TMC',
'required': False
}, {
'pattern': r'extras/thumbnail\.(png|jpg)',
'description':
'Optional thumbnail image which may be shown in search interface',
'required': False
}]
return schema
def get_table_schema(schema_name, version, optional_fields=[], offline=None):
schema = load_yaml(_table_schemas_path / 'assays' /
_get_schema_filename(schema_name, version))
names = [field['name'] for field in schema['fields']]
for field in schema['fields']:
_add_level_1_description(field)
_validate_level_1_enum(field)
_add_constraints(field, optional_fields, offline=offline, names=names)
_validate_field(field)
return schema
def get_other_schema(schema_name, version, offline=None):
schema = load_yaml(_table_schemas_path / 'others' /
f'{schema_name}-v{version}.yaml')
for field in schema['fields']:
_add_constraints(field, optional_fields=[], offline=offline)
return schema
def _assay_to_schema_name(assay_type, source_project):
'''
Given an assay name, and a source_project (may be None),
read all the schemas until one matches.
Return the schema name, but not the version.
>>> _assay_to_schema_name('PAS microscopy', None)
'stained'
>>> _assay_to_schema_name('snRNAseq', None)
'scrnaseq'
>>> _assay_to_schema_name('snRNAseq', 'HCA')
'scrnaseq-hca'
Or, if a match can not be found (try-except just for shorter lines):
>>> try: _assay_to_schema_name('PAS microscopy', 'HCA')
... except PreflightError as e: print(e)
No schema where 'PAS microscopy' is assay_type and 'HCA' is source_project
>>> try: _assay_to_schema_name('snRNAseq', 'Bad Project')
... except PreflightError as e: print(e)
No schema where 'snRNAseq' is assay_type and 'Bad Project' is source_project
>>> try: _assay_to_schema_name('Bad assay', None)
... except PreflightError as e: print(e)
No schema where 'Bad assay' is assay_type
>>> try: _assay_to_schema_name('Bad assay', 'HCA')
... except PreflightError as e: print(e)
No schema where 'Bad assay' is assay_type and 'HCA' is source_project
'''
for path in (Path(__file__).parent / 'table-schemas' /
'assays').glob('*.yaml'):
schema = load_yaml(path)
assay_type_fields = [
f for f in schema['fields'] if f['name'] == 'assay_type'
]
source_project_fields = [
f for f in schema['fields'] if f['name'] == 'source_project'
]
# Because names are unique, these list should not contain more than one field:
assert len(assay_type_fields) <= 1
assert len(source_project_fields) <= 1
if assay_type not in assay_type_fields[0]['constraints']['enum']:
continue
if source_project_fields:
if not source_project:
continue
if source_project:
if not source_project_fields:
continue
if source_project not in source_project_fields[0]['constraints'][
'enum']:
continue
return re.match(r'.+(?=-v\d+)', path.stem)[0]
message = f"No schema where '{assay_type}' is assay_type"
if source_project is not None:
message += f" and '{source_project}' is source_project"
raise PreflightError(message)
