def canonicalize(curie: str):
"""Return the best CURIE."""
# TODO maybe normalize the curie first?
norm_prefix, norm_identifier = normalize_curie(curie)
if norm_prefix is None or norm_identifier is None:
return jsonify(
query=curie,
normalizable=False,
)
norm_curie = f'{norm_prefix}:{norm_identifier}'
rv = dict(query=curie)
if norm_curie != curie:
rv['norm_curie'] = norm_curie
if norm_curie not in graph:
rv['found'] = False
else:
result_curie = canonicalizer.canonicalize(norm_curie)
rv.update(
found=True,
result=result_curie,
mappings=url_for(
f'.{all_mappings.__name__}',
source_curie=norm_curie,
target_curie=result_curie,
),
)
return jsonify(rv)
def _process_interactor(s: str) -> Optional[str]:
prefix, identifier = normalize_curie(s)
if prefix is None:
logger.warning('could not parse %s', s)
return
if prefix == 'ncbigene':
return identifier
elif prefix == 'biogrid':
ncbigene_identifier = _map_ncbigene(identifier)
if ncbigene_identifier is not None:
return ncbigene_identifier
elif identifier in BIOGRID_NCBIGENE_REMAPPING:
remapped = BIOGRID_NCBIGENE_REMAPPING[identifier]
if not remapped:
logger.debug('tried but failed curation on %s', s)
return remapped
else:
logger.warning('need to curate: %s', s)
return
elif prefix == 'uniprot':
if identifier in UNIPROT_NCBIGENE_REMAPPING:
remapped = UNIPROT_NCBIGENE_REMAPPING[identifier]
if not remapped:
logger.debug('tried but failed curation on %s', s)
return remapped
else:
logger.warning('need to curate: %s', s)
return
else:
logger.warning('unhandled interactor: %s (%s:%s)', s, prefix,
identifier)
def test_strip_prefix(self):
"""Test stripping prefixes works."""
self.assertEqual(("go", "1234"), normalize_curie("GO:1234"))
self.assertEqual(("go", "1234"), normalize_curie("go:1234"))
self.assertEqual((None, None), normalize_curie("1234"))
self.assertEqual(("go", "1234"), normalize_curie("GO:GO:1234"))
self.assertEqual(("pubmed", "1234"), normalize_curie("pubmed:1234"))
# Test remapping
self.assertEqual(("pubmed", "1234"), normalize_curie("pmid:1234"))
self.assertEqual(("pubmed", "1234"), normalize_curie("PMID:1234"))
def test_strip_prefix(self):
"""Test stripping prefixes works."""
self.assertEqual(('go', '1234'), normalize_curie('GO:1234'))
self.assertEqual(('go', '1234'), normalize_curie('go:1234'))
self.assertEqual((None, None), normalize_curie('1234'))
self.assertEqual(('go', 'GO:1234'), normalize_curie('GO:GO:1234'))
self.assertEqual(('pubmed', '1234'), normalize_curie('pubmed:1234'))
# Test remapping
self.assertEqual(('pubmed', '1234'), normalize_curie('pmid:1234'))
self.assertEqual(('pubmed', '1234'), normalize_curie('PMID:1234'))
def _help_resolve(curie: str) -> Mapping[str, Any]:
prefix, identifier = normalize_curie(curie)
if prefix is None:
return dict(
query=curie,
success=False,
message='Could not identify prefix',
)
miriam = get_identifiers_org_link(prefix, identifier)
id_name_mapping = get_id_name_mapping(prefix)
if id_name_mapping is None:
rv = dict(
query=curie,
prefix=prefix,
identifier=identifier,
success=False,
)
if miriam:
rv.update(
dict(
miriam=miriam,
message=
'Could not find id->name mapping for prefix, but still able to report Identifiers.org link',
))
else:
rv['message'] = 'Could not find id->name mapping for prefix'
return rv
name = id_name_mapping.get(identifier)
if name is None:
return dict(
query=curie,
prefix=prefix,
identifier=identifier,
success=False,
message='Could not look up identifier',
)
return dict(
query=curie,
prefix=prefix,
identifier=identifier,
name=name,
success=True,
miriam=miriam,
)
def _iter_process_xrefs(s: str) -> Iterable[Tuple[str, str]]:
"""Take a string with pipe-delimited curies and split/normalize them.
Compact Uniform Identfiers (CURIE) examples:
- hgnc:12345
- ncbigene:12345
- uniprot:P12345
- ec-code:1.2.3.15
Goal:
make hgnc:1234|ncbigene:1245|uniprot...:12345" into a list of tuples
"""
for curie in s.split('|'):
curie = curie.strip()
prefix, identifier = normalize_curie(curie)
if prefix is not None:
yield prefix, identifier
def resolve(self, curie: str, resolve_alternate: bool = False) -> Mapping[str, Any]:
"""Return the results and summary when resolving a CURIE string."""
prefix, identifier = normalize_curie(curie)
if prefix is None:
return dict(
query=curie,
success=False,
message='Could not identify prefix',
)
miriam_link = get_identifiers_org_link(prefix, identifier)
obofoundry_link = get_obofoundry_link(prefix, identifier)
ols_link = get_ols_link(prefix, identifier)
if not self.has_prefix(prefix):
rv = dict(
query=curie,
prefix=prefix,
identifier=identifier,
success=False,
message=f'Could not find id->name mapping for {prefix}',
)
if miriam_link:
rv['miriam'] = miriam_link
if obofoundry_link:
rv['obofoundry'] = obofoundry_link
if ols_link:
rv['ols'] = ols_link
return rv
name = self.get_name(prefix, identifier)
if name is None and resolve_alternate:
primary_id = self.get_primary_id(prefix, identifier)
if primary_id != identifier:
miriam_link = get_identifiers_org_link(prefix, primary_id)
obofoundry_link = get_obofoundry_link(prefix, primary_id)
ols_link = get_ols_link(prefix, primary_id)
name = self.get_name(prefix, primary_id)
if name is None:
return dict(
query=curie,
prefix=prefix,
identifier=identifier,
success=False,
miriam=miriam_link,
obofoundry=obofoundry_link,
ols=ols_link,
message='Could not look up identifier',
)
return dict(
query=curie,
prefix=prefix,
identifier=identifier,
name=name,
success=True,
miriam=miriam_link,
obofoundry=obofoundry_link,
ols=ols_link,
)
