def machine(manager: Manager, agents: List[str], local: bool, host: str):
"""Get content from the INDRA machine and upload to BEL Commons."""
from indra.sources import indra_db_rest
from pybel import from_indra_statements
statements = indra_db_rest.get_statements(agents=agents)
click.echo('got {} statements from INDRA'.format(len(statements)))
graph = from_indra_statements(
statements,
name='INDRA Machine for {}'.format(', '.join(sorted(agents))),
version=time.strftime('%Y%m%d'),
)
click.echo('built BEL graph with {} nodes and {} edges'.format(
graph.number_of_nodes(), graph.number_of_edges()))
if 0 == len(graph):
click.echo('not uploading empty graph')
sys.exit(-1)
if local:
to_database(graph, manager=manager)
else:
resp = to_web(graph, host=host)
resp.raise_for_status()
def assess_completeness(
ids: Union[str, Tuple[str, str], List[Tuple[str, str]]],
reference_nodes: Union[Iterable[BaseEntity], BELGraph],
*,
verbose_indra_logger: bool = False,
) -> Optional[CompletenessSummary]:
"""Check INDRA if the given document has new interesting content compared to the graph.
:param ids: A CURIE (e.g., pmid:30606258, pmc:PMC6318896),
a pair of database/identifier (e.g. `('pmid', '30606258')`) or a list of pairs.
:param reference_nodes: A set of nodes or a BEL graph (which is a set of nodes)
:param verbose_indra_logger: Should the INDRA logger show any output below the WARNING level?
"""
if not verbose_indra_logger:
indra_logger.setLevel(logging.WARNING)
if isinstance(ids, str):
_ids = ids.split(':')
if len(_ids) != 2:
raise ValueError(f'String should be given as CURIE: {ids}')
ids = [_ids]
elif isinstance(ids, tuple):
ids = [ids]
# Normalize PMC database name as well as stringify all identifiers
ids = [
('pmcid' if db == 'pmc' else db, str(db_id))
for db, db_id in ids
]
stmts = get_statements_for_paper(ids=ids)
indra_graph = from_indra_statements(stmts)
indra_nodes = set(indra_graph)
if not indra_nodes:
print(f'INDRA did not return any results for {ids}')
return
return CompletenessSummary(
documents=ids,
reference_nodes=set(reference_nodes),
indra_nodes=indra_nodes,
)
def run_one(directory):
"""
:param str directory:
"""
if not os.path.isdir(directory):
return
if not os.path.exists(os.path.join(directory, 'config.yaml')):
return # Skip non-INDRA machine directories
log.info('running in %s', directory)
# Use config=None so it looks in the subdirectory
run_with_search_helper(directory, config=None)
default_config_fname = os.path.join(directory, 'config.yaml')
config = get_config(default_config_fname)
ndex_cred = get_ndex_cred(config)
name = ndex_cred.get('name') if ndex_cred is not None else None
model = load_model(directory)
stmts = model.get_statements()
if not stmts:
log.warning('no statements')
return
# Output CX
if name:
cx_str = assemble_cx(stmts, name)
with open(os.path.join(directory, 'output.cx'), 'w') as file:
print(cx_str, file=file)
# Output BEL gpickle
bel_graph = pybel.from_indra_statements(stmts)
pybel.to_pickle(bel_graph, os.path.join(directory, 'output.gpickle'))
return True