Add wikidata nodes and edges, update add_resources endpoint example, …

…use better variable names, revert kwarg addition to 'from_obo_path' call

mira/dkg/api.py

@@ -13,7 +13,7 @@
 
 from mira.dkg.client import AskemEntity, Entity, Relation
 from mira.dkg.utils import DKG_REFINER_RELS
-from mira.dkg.construct import process_resource
+from mira.dkg.construct import add_resource_to_dkg
 
 __all__ = [
  "api_blueprint",
@@ -368,16 +368,17 @@ def add_relations(
  )
  def add_resources(
  request: Request,
- resource_list: List[str] = Body(
+ resource_prefix_list: List[str] = Body(
  ...,
  description="A of resources to add to the DKG",
  title="Resource Prefixes",
- example=["probonto"],
+ example=["probonto", "wikidata", "eiffel", "geonames", "ncit",
+ "nbcbitaxon"],
  )
  ):
- for resource in resource_list:
+ for resource_prefix in resource_prefix_list:
  # nodes and edges will be a list of dicts
- nodes, edges = process_resource(resource.lower())
+ nodes, edges = add_resource_to_dkg(resource_prefix.lower())
  # node_info and edge_info are dictionaries that will be
  # unpacked when creating instances of entities and relations
  entities = [Entity(**node_info) for node_info in nodes]

mira/dkg/client.py

@@ -410,7 +410,6 @@ def add_relation(self, relation):
 
  self.create_tx(create_relation_query)
 
-
  def create_single_property_node_index(
  self,
  index_name: str,

mira/dkg/construct.py

@@ -426,10 +426,55 @@ def extract_cso_nodes_edges():
 
 
 def extract_wikidata_nodes_edges():
- pass
+ wikidata_nodes, wikidata_edges = [], []
+ for wikidata_id, label, description, synonyms, xrefs in tqdm(
+ get_unit_terms(), unit="unit"):
+ synonyms_list = [Synonym(value=value, type="") for value in synonyms]
+ xrefs_list = [Xref(id=_id, type="oboinowl:hasDbXref") for _id in xrefs]
+ wikidata_nodes.append(
+ {
+ "id": f"wikidata:{wikidata_id}",
+ "name": label,
+ "type": "class",
+ "description": description,
+ "synonyms": synonyms_list,
+ "xrefs": xrefs_list,
+ "obsolete": False
+ }
+ )
+
+ for (wikidata_id, label, description, synonyms, xrefs, value, formula,
+ symbols) in tqdm(get_physical_constant_terms()):
+ synonym_types, synonym_values = [], []
+ for syn in synonyms:
+ synonym_values.append(syn)
+ synonym_types.append("oboInOwl:hasExactSynonym")
+ for symbol in symbols:
+ synonym_values.append(symbol)
+ synonym_types.append("debio:0000031")
+
+ synonyms_list = [Synonym(value=value, type=type) for value, type
+ in zip(synonym_values, synonym_types)]
+ xrefs_list = [Xref(id=_id, type="oboinowl:hasDbXref") for _id in xrefs]
+ if value:
+ properties = {"debio:0000042": [str(value)]}
+ else:
+ properties = {}
+ wikidata_nodes.append(
+ {
+ "id": f"wikidata:{wikidata_id}",
+ "name": label,
+ "obsolete": False,
+ "type": "class",
+ "description": description,
+ "synonyms": synonyms_list,
+ "xrefs": xrefs_list,
+ "properties": properties
+ }
+ )
 
 
-def process_resource(resource_prefix: str):
+def add_resource_to_dkg(resource_prefix: str):
  if resource_prefix == "probonto":
  return extract_probonto_nodes_edges()
  elif resource_prefix == "geonames":
@@ -443,6 +488,7 @@ def process_resource(resource_prefix: str):
  elif resource_prefix == "cso":
  return extract_cso_nodes_edges()
  elif resource_prefix == "wikidata":
+ # combine retrieval of wikidata constants and units
  return extract_wikidata_nodes_edges()
 
 

mira/dkg/resources/cso.py

@@ -23,10 +23,9 @@ def get_cso_obo() -> Obo:
  )
  download(url=URL, path=PATH)
  # use https://github.com/pyobo/pyobo/pull/159
- kwargs = {"default_prefix": "cso"}
- return from_obo_path(PATH, prefix="cso", strict=False, **kwargs)
+ return from_obo_path(PATH, prefix="cso", default_prefix="cso", strict=False)
 
 
 if __name__ == "__main__":
  for term in get_cso_obo():
- print(term)
+ print(term)