Dataset IRIs are based on CxG dataset UUID (#82)

* Refactor dataset_class URI logic * Add get_cxg_dataset_url method * Update assert statements * Update version to 0.2.23
INCATools · Aug 15, 2024 · 116dfc0 · 116dfc0
1 parent df3e807
commit 116dfc0
Show file tree

Hide file tree

Showing 5 changed files with 639 additions and 585 deletions.
diff --git a/pandasaurus_cxg/graph_generator/graph_generator.py b/pandasaurus_cxg/graph_generator/graph_generator.py
@@ -22,6 +22,7 @@
     colour_mapping,
     find_and_rotate_center_layout,
     generate_subgraph,
+    get_cxg_dataset_url,
     parse_citation_field_into_dict,
     remove_special_characters,
     select_node_with_property,
@@ -124,15 +125,19 @@ def generate_rdf_graph(self):
                 grouped_dict_uuid[str(uuid.uuid4())] = temp_dict
 
         # generate dataset entity and has_source property
-        dataset_class = URIRef(self.ns[str(uuid.uuid4())])
+        uns = self.ea.enricher_manager.anndata.uns
+        citation_dict = {}
+        if citation_field_name in uns.keys():
+            citation_dict = parse_citation_field_into_dict(uns[citation_field_name])
+            dataset_class = URIRef(get_cxg_dataset_url(citation_dict.get("download_link").split("/")[-1].split(".")[0]))
+        else:
+            dataset_class = URIRef(self.ns[str(uuid.uuid4())])
         self.graph.add((dataset_class, RDF.type, URIRef(DATASET.get("iri"))))
         self.graph.add((dataset_class, RDFS.label, Literal(DATASET.get("label"))))
-        uns = self.ea.enricher_manager.anndata.uns
         for key, value in uns.items():
             if not isinstance(value, str):
                 continue
             if key == citation_field_name:
-                citation_dict = parse_citation_field_into_dict(value)
                 for citation_key, citation_value in citation_dict.items():
                     self.graph.add(
                         (

diff --git a/pandasaurus_cxg/graph_generator/graph_generator_utils.py b/pandasaurus_cxg/graph_generator/graph_generator_utils.py
@@ -158,3 +158,19 @@ def parse_citation_field_into_dict(value: str) -> Dict[str, str]:
             }
         )
     return key_value_pairs
+
+
+def get_cxg_dataset_url(matrix_id: str) -> str:
+    """
+    Constructs a URL for a CellxGene dataset.
+
+    Given a matrix ID, this function returns the corresponding URL for accessing
+    the dataset on the CellxGene platform.
+
+    Args:
+        matrix_id: The unique identifier for the dataset matrix.
+
+    Returns:
+        The full URL to access the dataset on CellxGene.
+    """
+    return f"https://cellxgene.cziscience.com/e/{matrix_id}.cxg/"