def _process_annotations(
data,
remove_ungrounded: bool = False,
skip_namespaces: Optional[Collection[str]] = None,
) -> None:
"""Process the annotations in a PyBEL edge data dictionary."""
cell_line_entities = data[ANNOTATIONS].get('CellLine')
if cell_line_entities:
ne = []
for entity in cell_line_entities:
if entity[NAMESPACE] == 'CellLine':
_namespaces = [
'efo',
# 'clo', # FIXME implement CLO in PyOBO then uncomment
]
g_prefix, g_identifier, g_name = pyobo.ground(
_namespaces, entity[IDENTIFIER])
if g_prefix and g_identifier:
ne.append(
Entity(namespace=g_prefix,
identifier=g_identifier,
name=g_name))
elif not remove_ungrounded:
logger.warning('could not ground CellLine: "%s"',
entity[IDENTIFIER])
ne.append(entity)
data[ANNOTATIONS]['CellLine'] = ne
# fix text locations
text_location = data[ANNOTATIONS].get('TextLocation')
if text_location:
data[ANNOTATIONS]['TextLocation'] = [
text_location_labels.get(entity.identifier, entity)
for entity in text_location
]
# remap category names
data[ANNOTATIONS] = {
_BEL_ANNOTATION_PREFIX_CATEGORY_MAP.get(category, category): entities
for category, entities in data[ANNOTATIONS].items()
}
# fix namespaces that were categories before
for category, entities in data[ANNOTATIONS].items():
if category in CATEGORY_BLACKLIST:
continue
ne = []
for entity in entities:
if not isinstance(entity, dict):
raise TypeError(f'entity should be a dict. got: {entity}')
nn = _BEL_ANNOTATION_PREFIX_MAP.get(entity[NAMESPACE])
if nn is not None:
entity[NAMESPACE] = nn
_process_concept(concept=entity, skip_namespaces=skip_namespaces)
ne.append(entity)
data[ANNOTATIONS][category] = ne
def test_ground(self):
"""Test grounding a ChEBI entry by name and synonym."""
for query in ("Fusilade II", "fluazifop-P-butyl"):
with self.subTest(query=query), mock_id_name_mapping, mock_id_synonyms_mapping:
prefix, identifier, name = pyobo.ground("chebi", query)
self.assertIsNotNone(prefix)
self.assertIsNotNone(identifier)
self.assertIsNotNone(name)
self.assertEqual("chebi", prefix)
self.assertEqual("132964", identifier)
self.assertEqual("fluazifop-P-butyl", name)
def test_ground(self):
"""Test grounding a ChEBI entry by name and synonym."""
for query in ('Fusilade II', 'fluazifop-P-butyl'):
with self.subTest(
query=query
), mock_id_name_mapping, mock_id_synonyms_mapping:
prefix, identifier, name = pyobo.ground('chebi', query)
self.assertIsNotNone(prefix)
self.assertIsNotNone(identifier)
self.assertIsNotNone(name)
self.assertEqual('chebi', prefix)
self.assertEqual('132964', identifier)
self.assertEqual('fluazifop-P-butyl', name)
def get_identifier(
namespace: str,
name: str) -> Union[Tuple[str, None, str], Tuple[str, str, str]]:
if namespace in {'SFAM', 'SCOMP'}:
return 'fplx', bel_fplx.get(name), name
if namespace in {'SCHEM', 'CHEBI'}:
prefix, identifier, name = pyobo.ground('chebi', name)
return prefix or namespace, identifier, name
norm_namespace = bioregistry.normalize_prefix(namespace)
if norm_namespace is None:
raise ValueError(f'could not normalize {namespace}')
namespace = norm_namespace
if namespace in MISSING_NAMESPACE:
return namespace, None, name
try:
name_id_mapping = pyobo.get_name_id_mapping(namespace)
except:
logger.info('missing namespace: %s', namespace)
MISSING_NAMESPACE.add(namespace)
return namespace, None, name
if name_id_mapping:
identifier = name_id_mapping.get(name)
if identifier:
return namespace, identifier, name
elif (namespace, name) not in MISSING:
MISSING.add((namespace, name))
logger.debug('missing lookup for %s ! %s', namespace, name)
return namespace, None, name
elif namespace not in MISSING_NAMESPACE:
logger.info('empty namespace: %s', namespace)
MISSING_NAMESPACE.add(namespace)
return namespace, None, name
def _process_annotations(data, remove_ungrounded: bool = False) -> None:
"""Process the annotations in a PyBEL edge data dictionary."""
grounded_category_curie_polarity = []
ungrounded_category_name_polarity = []
for category, names in data[ANNOTATIONS].items():
if category == 'CellLine':
_namespaces = [
'efo',
# 'clo', # FIXME implement CLO import and add here
]
for name, polarity in names.items():
g_prefix, g_identifier, g_name = pyobo.ground(_namespaces, name)
if g_prefix and g_identifier:
grounded_category_curie_polarity.append((
category, Entity(namespace=g_prefix, identifier=g_identifier, name=g_name), polarity,
))
else:
ungrounded_category_name_polarity.append((category, name, polarity))
elif category in _BEL_ANNOTATION_PREFIX_MAP:
norm_prefix = _BEL_ANNOTATION_PREFIX_MAP[category]
norm_category = _BEL_ANNOTATION_PREFIX_CATEGORY_MAP.get(category, category)
for name, polarity in names.items():
_, identifier, _ = pyobo.ground(norm_prefix, name)
if identifier:
grounded_category_curie_polarity.append((
norm_category, Entity(namespace=norm_prefix, identifier=identifier, name=name), polarity,
))
else:
ungrounded_category_name_polarity.append((norm_category, name, polarity))
elif normalize_prefix(category):
norm_prefix = normalize_prefix(category)
for name, polarity in names.items():
_, identifier, _ = pyobo.ground(norm_prefix, name)
if identifier:
grounded_category_curie_polarity.append((
category, Entity(namespace=norm_prefix, identifier=identifier, name=name), polarity,
))
else:
ungrounded_category_name_polarity.append((category, name, polarity))
else:
if category not in _UNHANDLED_ANNOTATION:
logger.warning('unhandled annotation: %s', category)
_UNHANDLED_ANNOTATION.add(category)
if isinstance(names, dict):
for name, polarity in names.items():
ungrounded_category_name_polarity.append((category, name, polarity))
else:
ungrounded_category_name_polarity.append((category, names, True))
data[ANNOTATIONS] = defaultdict(dict)
for category, entity, polarity in grounded_category_curie_polarity:
data[ANNOTATIONS][category][entity.curie] = polarity
data[ANNOTATIONS] = dict(data[ANNOTATIONS])
if not remove_ungrounded and ungrounded_category_name_polarity:
data[FREE_ANNOTATIONS] = defaultdict(dict)
for category, name, polarity in ungrounded_category_name_polarity:
data[FREE_ANNOTATIONS][category][name] = polarity
data[FREE_ANNOTATIONS] = dict(data[FREE_ANNOTATIONS])
def get_bel() -> pybel.BELGraph:
"""Get the HMDD data."""
# category mir disease pmid description
path = ensure_path(PREFIX, URL)
df = pd.read_csv(
path,
sep='\t',
dtype=str,
encoding="ISO-8859-1",
)
failed_mirnas = 0
mirna_to_dsl = {}
mirnas = df['mir'].unique()
it = tqdm(mirnas, desc='mapping miRNA names')
for text in it:
_, identifier, name = pyobo.ground('mirbase', text)
if identifier is None:
it.write(f'[mirbase] could not ground: {text}')
failed_mirnas += 1
continue
mirna_to_dsl[text] = pybel.dsl.MicroRna(
namespace='mirbase',
identifier=identifier,
name=name,
)
logger.info(f'failed on {failed_mirnas}/{len(mirnas)} miRNAs')
failed_diseases = 0
disease_to_dsl = {}
diseases = df['disease'].unique()
it = tqdm(diseases, desc='mapping disease names')
for text in it:
prefix, identifier, name = pyobo.ground(['mondo', 'doid', 'efo', 'hp', 'mesh'], text)
if identifier is None and ', ' in text:
i = text.index(', ')
left, right = text[:i], text[i + 2:]
x = f'{right} {left}'
prefix, identifier, name = pyobo.ground(['mondo', 'doid', 'efo', 'hp', 'mesh'], x)
if identifier is None and ', ' in x:
x2 = ' '.join(z.strip() for z in text.split(',')[::-1])
prefix, identifier, name = pyobo.ground(['mondo', 'doid', 'efo', 'hp', 'mesh'], x2)
if identifier is None:
it.write(f'could not ground {text}')
failed_diseases += 1
continue
disease_to_dsl[text] = pybel.dsl.Pathology(
namespace=prefix,
identifier=identifier,
name=name,
)
logger.info(f'failed on {failed_diseases}/{len(diseases)} diseases')
rv = pybel.BELGraph(name='HMDD', version=VERSION)
for _category, mir, disease, pmid, text in df.values:
source = mirna_to_dsl.get(mir)
target = disease_to_dsl.get(disease)
if not source or not target:
continue
rv.add_regulates(
source,
target,
citation=pmid,
evidence=text,
)
return rv