#313: improve get_head_join_node transformation to BC format

biocypher · Apr 22, 2024 · 42d5f4b · 42d5f4b
1 parent ae7ceb5
commit 42d5f4b
Show file tree

Hide file tree

Showing 3 changed files with 71 additions and 15 deletions.
diff --git a/biocypher/_ontology.py b/biocypher/_ontology.py
@@ -569,18 +569,22 @@ def _get_head_join_node(self, adapter: OntologyAdapter) -> str:
         """
 
         head_join_node = None
-        head_join_node_label = adapter.get_head_join_node().replace("_", " ")
+        user_defined_head_join_node_label = adapter.get_head_join_node()
+        head_join_node_label_in_bc_format = to_lower_sentence_case(
+            user_defined_head_join_node_label.replace("_", " ")
+        )
 
         if self._head_ontology._switch_label_and_id:
-            head_join_node = head_join_node_label
+            head_join_node = head_join_node_label_in_bc_format
         elif not self._head_ontology._switch_label_and_id:
-            for node, data in self._head_ontology.get_nx_graph().nodes(
+            for node_id, data in self._head_ontology.get_nx_graph().nodes(
                 data=True
             ):
-                if "label" in data and data["label"] == to_lower_sentence_case(
-                    head_join_node_label
+                if (
+                    "label" in data
+                    and data["label"] == head_join_node_label_in_bc_format
                 ):
-                    head_join_node = node
+                    head_join_node = node_id
                     break
 
         if head_join_node not in self._head_ontology.get_nx_graph().nodes:

diff --git a/test/ontologies/missing_label.ttl b/test/ontologies/missing_label.ttl
@@ -2,11 +2,11 @@
 @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
 
 owl:ID_0 a owl:Class ;
-     rdfs:label "Label_Root" ;
+     rdfs:label "Test_Missing_Label_Root" ;
      rdfs:comment "The class of OWL individuals." .
 
 owl:ID_1A a owl:Class ;
-     rdfs:label "Label_Level1A" ;
+     rdfs:label "Test_Missing_Label_Level1A" ;
      rdfs:subClassOf owl:ID_0 ;
      rdfs:comment "Level1A." .
 

diff --git a/test/test_ontology.py b/test/test_ontology.py
@@ -178,14 +178,14 @@ def test_multiple_parents(ontology_file):
 def test_missing_label_on_node():
     ontology_adapter = OntologyAdapter(
         ontology_file="test/ontologies/missing_label.ttl",
-        root_label="Label_Root",
+        root_label="Test_Missing_Label_Root",
     )
     result = ontology_adapter.get_nx_graph()
     # Expected hierarchy:
-    #  label root
-    #  ├── label level1a
+    #  test missing label root
+    #  ├── test missing label level1a
     # (└── level1B) <- missing label on this node (should not be part of the graph)
-    expected_edges = [("label level1a", "label root")]
+    expected_edges = [("test missing label level1a", "test missing label root")]
     for edge in expected_edges:
         assert edge in result.edges
     assert len(result.edges) == len(expected_edges)
@@ -229,7 +229,59 @@ def test_root_node_not_found():
     )
 
 
-def test_reverse_labels_from_yaml_config():
+def test_switch_id_and_label_from_yaml_config():
+    bc = BioCypher(
+        head_ontology={
+            "url": "test/ontologies/reverse_labels.ttl",
+            "root_node": "Label_Root",
+            "switch_label_and_id": True,
+        },
+        tail_ontologies={
+            "tail": {
+                "url": "test/ontologies/missing_label.ttl",
+                "head_join_node": "Label_Level1A",
+                "tail_join_node": "Test_Missing_Label_Root",
+                "switch_label_and_id": True,
+            }
+        },
+    )
+    expected_not_switched = [
+        "label level1b",
+        "label root",
+        "label level1a",
+        "test missing label level1a",
+    ]
+    for node in bc._get_ontology()._nx_graph.nodes:
+        assert node in expected_not_switched
+
+
+def test_mixed_switch_id_and_label_from_yaml_config():
+    bc = BioCypher(
+        head_ontology={
+            "url": "test/ontologies/reverse_labels.ttl",
+            "root_node": "Label_Root",
+            "switch_label_and_id": True,
+        },
+        tail_ontologies={
+            "tail": {
+                "url": "test/ontologies/missing_label.ttl",
+                "head_join_node": "Label_Level1A",
+                "tail_join_node": "Test_Missing_Label_Root",
+                "switch_label_and_id": False,
+            }
+        },
+    )
+    expected_not_switched = [
+        "label level1b",
+        "label root",
+        "label level1a",
+        "ID_1A",
+    ]
+    for node in bc._get_ontology()._nx_graph.nodes:
+        assert node in expected_not_switched
+
+
+def test_do_not_switch_id_and_label_from_yaml_config():
     bc = BioCypher(
         head_ontology={
             "url": "test/ontologies/reverse_labels.ttl",
@@ -240,7 +292,7 @@ def test_reverse_labels_from_yaml_config():
             "tail": {
                 "url": "test/ontologies/missing_label.ttl",
                 "head_join_node": "Label_Level1A",
-                "tail_join_node": "Label_Root",
+                "tail_join_node": "Test_Missing_Label_Root",
                 "switch_label_and_id": False,
             }
         },
@@ -260,7 +312,7 @@ def test_head_join_node_not_found():
             "tail": {
                 "url": "test/ontologies/missing_label.ttl",
                 "head_join_node": "not present",
-                "tail_join_node": "Label_Root",
+                "tail_join_node": "Test_Missing_Label_Root",
             }
         },
     )