Add comments and debugging

src/pathme/cli.py

@@ -8,7 +8,6 @@
 import click
 from bio2bel_chebi import Manager as ChebiManager
 from bio2bel_hgnc import Manager as HgncManager
-from pybel import from_pickle
 from tqdm import tqdm
 
 from pathme.constants import *
@@ -20,6 +19,7 @@
 from pathme.utils import get_files_in_folder, make_downloader, statistics_to_df, summarize_helper
 from pathme.wikipathways.rdf_sparql import get_wp_statistics, wikipathways_to_pickles
 from pathme.wikipathways.utils import get_file_name_from_url, get_wikipathways_files, unzip_file
+from pybel import from_pickle
 
 log = logging.getLogger(__name__)
 
@@ -127,12 +127,13 @@ def download():
 
 @wikipathways.command()
 @click.option('-c', '--connection', help="Defaults to {}".format(DEFAULT_CACHE_CONNECTION))
-@click.option('-v', '--verbose', is_flag=True)
+@click.option('-d', '--debug', is_flag=True, default=False, help='Debug mode')
 @click.option('-x', '--only-canonical', default=True, help='Parse only canonical pathways')
-def to_bel(connection, verbose, only_canonical):
+def to_bel(connection, debug, only_canonical):
     """Convert WikiPathways to BEL."""
     logging.basicConfig(level=logging.DEBUG, format="%(asctime)s - %(levelname)s - %(name)s - %(message)s")
-    if verbose:
+
+    if debug:
         log.setLevel(logging.DEBUG)
 
     log.info('Initiating HGNC Manager')

src/pathme/wikipathways/convert_to_bel.py

@@ -32,6 +32,8 @@ def convert_to_bel(nodes: Dict[str, Dict], complexes: Dict[str, Dict], interacti
         contact='daniel.domingo.fernandez@scai.fraunhofer.de',
     )
 
+    log.warning('Parsing %s', pathway_info['title'])
+
     nodes = nodes_to_bel(nodes, hgnc_manager)
     nodes.update(complexes_to_bel(complexes, nodes, graph))
 
@@ -62,7 +64,7 @@ def node_to_bel(node: Dict, hgnc_manager: Manager) -> BaseEntity:
         identifier = uri_id
 
     if isinstance(identifier, set):
-        print('Multiple identifier {}'.format(node['identifier']))
+        log.warning('Multiple identifier {}'.format(node['identifier']))
         # TODO: print the wikipathways bps that return a set because they are probably wrong.
         identifier = list(identifier)[0]
 
@@ -76,7 +78,7 @@ def node_to_bel(node: Dict, hgnc_manager: Manager) -> BaseEntity:
 
     # TODO: Deal with multiple names. Print for now to identify possible errors
     if isinstance(name, set):
-        print('Multiple name {}'.format(node['name']))
+        log.warning('Multiple name {}'.format(node['name']))
         name = list(name)[0]
 
     if 'Protein' in node_types:
@@ -140,6 +142,7 @@ def add_edges(graph: BELGraph, participants, nodes, att: Dict):
     edge_types = att['interaction_types']
     _, _, namespace, interaction_id = parse_id_uri(uri_id)
 
+    # Add reaction if the edge type is a Conversion
     if 'Conversion' in edge_types:
         reactants = set()
         products = set()
@@ -161,6 +164,7 @@ def add_edges(graph: BELGraph, participants, nodes, att: Dict):
         reaction_node = reaction(reactants=reactants, products=products)
         graph.add_node_from_data(reaction_node)
 
+    # Else it is a simple edge, process it
     else:
         for source, target in participants:
             if source in nodes:

src/pathme/wikipathways/rdf_sparql.py

@@ -223,7 +223,7 @@ def get_wp_statistics(resource_files, resource_folder, hgnc_manager) -> Tuple[
 
 
 def rdf_wikipathways_to_bel(rdf_graph, hgnc_manager) -> BELGraph:
-    """Convert RDF graph to BELGraph
+    """Convert RDF graph to BELGraph.
 
     :param rdf_graph: RDF graph
     :param bio2bel_hgnc.Manager: HGNC manager