def install_and_populate_database(name: str, connection: Optional[str] = None) -> Optional[str]:
"""Install biological database.
:param name: The name of the databae
:param connection: The optional database connection
"""
export_file = os.path.join(OUTPUT_DIR, f'{name}.csv')
summary_file = os.path.join(OUTPUT_DIR, f'{name}_summary.csv')
json_file = os.path.join(OUTPUT_DIR, f'{name}.bel.json')
if os.path.exists(export_file):
logger.info(f'{EMOJI} {name} has already been retrieved. See: {export_file}')
return export_file
if os.path.exists(json_file):
logger.info(f'{EMOJI} loaded {name} JSON: {json_file}')
graph = from_json_path(json_file)
df = to_pykeen_df(graph)
to_pykeen_path(df, export_file)
to_pykeen_summary_path(df, summary_file)
return export_file
if name in _PATHME_MODULES:
click.secho(
f'{EMOJI} You are trying to install {name} which is part of PathMe. '
'Due to the complexity of the installation, we refer to the tutorial at'
' https://github.com/PathwayMerger/PathMe to install it. Alternatively, you can download the dumps of '
f'{name} directly from the URLs at the DiffuPath README file.'
)
sys.exit(0)
module_name = f'bio2bel_{name}'
bio2bel_module = ensure_bio2bel_installation(module_name)
logger.debug(f'{EMOJI} imported {name}')
manager_cls = bio2bel_module.Manager
manager = manager_cls(connection=connection)
if issubclass(manager_cls, AbstractManager):
if not manager.is_populated():
logger.info(f'{EMOJI} populating {module_name}')
manager.populate()
else:
logger.debug(f'{EMOJI} {module_name} has already been populated')
logger.debug(f'{EMOJI} generating BEL for {module_name}')
graph = manager.to_bel()
logger.debug(f'Summary: {graph.number_of_nodes()} nodes / {graph.number_of_edges()} edges')
to_json_path(graph, json_file, indent=2)
logger.debug(f'{EMOJI} generating PyKEEN TSV for {module_name}')
df = to_pykeen_df(graph)
to_pykeen_summary_path(df, summary_file)
success = to_pykeen_path(df, export_file)
if success:
logger.debug(f'{EMOJI} wrote PyKEEN TSV to {export_file}')
return export_file
logger.warning(f'{EMOJI} no statements generated')
def install_bio2bel_module(name: str,
connection: Optional[str] = None,
rebuild: bool = False) -> Optional[str]:
"""Install Bio2BEL module.
:param name: The name of the Bio2BEL module
:param connection: The optional database connection
:param rebuild: Should the cache not be used? Defaults to False.
"""
module_name = _SPECIAL_CASES.get(name, f'bio2bel_{name}')
pykeen_df_path = os.path.join(biokeen_config.data_directory,
f'{name}.{biokeen_config.keen_tsv_ext}')
pykeen_df_summary_path = os.path.join(biokeen_config.data_directory,
f'{name}.keen.summary.json')
json_path = os.path.join(biokeen_config.data_directory, f'{name}.bel.json')
if os.path.exists(pykeen_df_path) and not rebuild:
logger.info(
f'{EMOJI} {module_name} has already been retrieved. See: {pykeen_df_path}'
)
return pykeen_df_path
if os.path.exists(json_path) and not rebuild:
logger.info(f'{EMOJI} loaded {module_name} JSON: {json_path}')
graph = from_json_path(json_path)
df = to_pykeen_df(graph)
to_pykeen_path(df, pykeen_df_path)
to_pykeen_summary_path(df, pykeen_df_summary_path)
return pykeen_df_path
bio2bel_module = ensure_bio2bel_installation(module_name)
logger.debug(f'{EMOJI} imported {module_name}')
manager_cls = bio2bel_module.Manager
if not issubclass(manager_cls, BELManagerMixin):
version = pkg_resources.get_distribution(module_name).version
logger.warning(
f'{EMOJI} {module_name} v{version} does not produce BEL')
sys.exit(1)
manager = manager_cls(connection=connection)
if issubclass(manager_cls, AbstractManager):
if not manager.is_populated():
logger.info(f'{EMOJI} populating {module_name}')
manager.populate()
else:
logger.debug(f'{EMOJI} {module_name} has already been populated')
logger.debug(f'{EMOJI} generating BEL for {module_name}')
graph = manager.to_bel()
logger.debug(
f'Summary: {graph.number_of_nodes()} nodes / {graph.number_of_edges()} edges'
)
to_json_path(graph, json_path, indent=2)
logger.debug(f'{EMOJI} generating PyKEEN TSV for {module_name}')
df = to_pykeen_df(graph)
to_pykeen_summary_path(df, pykeen_df_summary_path)
success = to_pykeen_path(df, pykeen_df_path)
if success:
logger.debug(f'{EMOJI} wrote PyKEEN TSV to {pykeen_df_path}')
return pykeen_df_path
logger.warning(f'{EMOJI} no statements generated')
def to_bel_json(self, path: str, **kwargs):
"""Export the BEL model as a node-link JSON file."""
pybel.to_json_path(self.to_bel(), path, **kwargs)
def get_graph(
self,
directory: Optional[str] = None,
use_cached: bool = True,
use_tqdm: bool = True,
) -> BELGraph:
"""Get the graph from all sources."""
if directory is None:
if self.directory is None:
raise ValueError
directory = self.directory
pickle_path = os.path.join(directory, f'{self.name}.bel.pickle')
if use_cached and os.path.exists(pickle_path):
return pybel.from_pickle(pickle_path)
rv = union(self.get_graphs(use_tqdm=use_tqdm))
self.metadata.update(rv)
pybel.to_pickle(rv, pickle_path)
nodelink_path = os.path.join(directory,
f'{self.name}.bel.nodelink.json')
pybel.to_json_path(rv, nodelink_path)
sif_path = os.path.join(directory, f'{self.name}.bel.sif')
pybel.to_sif_path(rv, sif_path)
gsea_path = os.path.join(directory, f'{self.name}.bel.gmt')
pybel.to_gsea_path(rv, gsea_path)
graphml_path = os.path.join(directory, f'{self.name}.bel.graphml')
pybel.to_graphml(rv, graphml_path)
try:
statements = pybel.to_indra_statements(rv)
except ImportError:
pass
else:
indra_path = os.path.join(directory, f'{self.name}.indra.pickle')
with open(indra_path, 'wb') as file:
pickle.dump(statements, file)
try:
from pybel_cx import to_cx_file
except ImportError:
pass
else:
cx_path = os.path.join(directory, f'{self.name}.bel.cx.json')
with open(cx_path, 'w') as file:
to_cx_file(rv, file)
try:
from pybel_tools.assembler.html import to_html
except ImportError:
pass
else:
html_path = os.path.join(directory, 'index.html')
with open(html_path, 'w') as file:
print(to_html(rv), file=file)
return rv