fix(metacyc): remove reaction nodes with non-canonical mcID

Some metabolic reactions have compartments, substrates and/or other information in their mcID and
displayName. They always have a related reaction that doesn't have those parts, so we remove those
extra reactions found in the SBML file.

fix #19

metabolike/db/metacyc.py

@@ -414,6 +414,17 @@ def merge_nodes(
             val=attr_val,
         )
 
+    def delete_bad_reaction_nodes(self):
+        """
+        Delete reaction nodes with non-canonical mcIds and all their relationships.
+        """
+        self.write(
+            """
+            MATCH (r:Reaction) WHERE r.displayName <> r.canonicalId
+            DETACH DELETE r;
+            """
+        )
+
     def get_view_of_pathway(self, pathway_id: str):
         """
         Get the view of a pathway.

metabolike/parser/metacyc.py

@@ -185,6 +185,10 @@ def setup(self, create_db: bool = True, **kwargs):
                     self.reaction_to_graph(rxn, rxn_dat)
                     logger.debug(f"Added extra info for reaction {rxn}")
 
+                # Remove additional reactions in the SBML file that have
+                # non-canonical IDs
+                self.db.delete_bad_reaction_nodes()
+
                 # Read pathways file if given
                 if self.input_files["pathways"]:
                     logger.info("Creating pathway links")
@@ -354,6 +358,13 @@ def reaction_to_graph(self, rxn_id: str, rxn_dat: Dict[str, List[List[str]]]):
             return
         lines = rxn_dat[canonical_id]
         props: Dict[str, Union[str, List[str]]] = {"canonicalId": canonical_id}
+
+        # Don't bother with parsing the other attributes if the canonical ID
+        # is not the displayName (the node will be deleted anyway)
+        if canonical_id != rxn_id:
+            self.db.add_props_to_node("Reaction", "displayName", rxn_id, props)
+            return
+
         for k, v in lines:
             # SYNONYMS is a special case because it is a list
             if k in {"SYNONYMS", "TYPES"}: