test: get_ancestors

chanzuckerberg · Nov 19, 2023 · b9d6d5b · b9d6d5b
1 parent 45c7f34
commit b9d6d5b
Show file tree

Hide file tree

Showing 2 changed files with 178 additions and 122 deletions.
diff --git a/scripts/compute_mappings/compute_ancestor_mapping.py b/scripts/compute_mappings/compute_ancestor_mapping.py
@@ -14,35 +14,39 @@
 # In[ ]:
 import json
 from collections import defaultdict
-from typing import Dict, List
+from typing import Dict, Iterable, Iterator, List
 
 import yaml
 from owlready2 import get_ontology  # noqa
+from owlready2.entity import ThingClass
+from owlready2.namespace import Ontology
 
 # In[ ]:
 
 RO__PART_OF = "BFO_0000050"  # RO ontology property "part of"
 
 
-def get_ancestors(onto, class_name):  # type: ignore
-    z = onto.search_one(iri=f"http://purl.obolibrary.org/obo/{class_name}")
+def get_ancestors(onto: Ontology, class_name: str) -> Iterator[str]:
+    print("djh call")
+    entity: ThingClass = onto.search_one(iri=f"http://purl.obolibrary.org/obo/{class_name}")
 
-    def recurse(x):  # type: ignore
+    # print(f"djh: {entity} {type(entity)}")
+    # return entity
+    def recurse(x: ThingClass) -> Iterator[str]:
         for e in x.is_a:
             if hasattr(e, "value") and e.property.name == RO__PART_OF:
-                val = e.value.name.replace("obo.", "")
-                z = onto.search_one(iri=f"http://purl.obolibrary.org/obo/{val}")
+                z = onto.search_one(iri=f"http://purl.obolibrary.org/obo/{e.value.name.replace('obo.', '')}")
                 yield z.name
-                yield from recurse(z)  # type: ignore
+                yield from recurse(z)
 
-    yield z.name
-    yield from recurse(z)  # type: ignore
+    yield entity.name
+    yield from recurse(entity)
 
 
 # In[ ]:
 
 
-def create_ancestors_mapping(onto, classes, prefix=None) -> Dict[str, List[str]]:
+def create_ancestors_mapping(onto: Ontology, classes: Iterable[ThingClass], prefix: str = None) -> Dict[str, List[str]]:
     ancestors: Dict[str, List[str]] = defaultdict(list)
     for cls in classes:
         class_key = cls.name.replace("_", ":")
@@ -56,115 +60,104 @@ def create_ancestors_mapping(onto, classes, prefix=None) -> Dict[str, List[str]]
 # In[ ]:
 
 
-# Load owl.info to grab latest ontology sources
-owl_info_yml = "cellxgene_schema_cli/cellxgene_schema/ontology_files/owl_info.yml"
-with open(owl_info_yml, "r") as owl_info_handle:
-    owl_info = yaml.safe_load(owl_info_handle)
-
-
-# In[ ]:
-
-
-human_latest_key = owl_info["HsapDv"]["latest"]
-human_ontology = owl_info["HsapDv"]["urls"][human_latest_key]
-
-human_onto = get_ontology(human_ontology)
-human_onto.load()
-
-m_human = create_ancestors_mapping(human_onto, human_onto.classes(), "HsapDv")  # type: ignore
-
-
-# In[ ]:
-
-
-mouse_latest_key = owl_info["MmusDv"]["latest"]
-mouse_ontology = owl_info["MmusDv"]["urls"][mouse_latest_key]
-
-mouse_onto = get_ontology(mouse_ontology)
-mouse_onto.load()
-
-m_mouse = create_ancestors_mapping(mouse_onto, mouse_onto.classes(), "MmusDv")  # type: ignore
-
-
-# In[ ]:
-
-
-uberon_latest_key = owl_info["UBERON"]["latest"]
-uberon_ontology = owl_info["UBERON"]["urls"][uberon_latest_key]
-
-tissue_onto = get_ontology(uberon_ontology)
-tissue_onto.load()
-
-
-# In[ ]:
-
-
-uberon_classes = [
-    "UBERON_0007236",
-    "UBERON_0000106",
-    "UBERON_0014859",
-    "UBERON_0008264",
-    "UBERON_0007233",
-    "UBERON_0000112",
-    "UBERON_8000003",
-    "UBERON_0014857",
-    "UBERON_0009849",
-    "UBERON_0034920",
-    "UBERON_0000069",
-    "UBERON_0000109",
-    "UBERON_8000001",
-    "UBERON_0000068",
-    "UBERON_0018685",
-    "UBERON_0000107",
-    "UBERON_0007222",
-    "UBERON_0000092",
-    "UBERON_0018378",
-    "UBERON_0014864",
-    "UBERON_0004730",
-    "UBERON_0000111",
-    "UBERON_0007220",
-    "UBERON_0014405",
-    "UBERON_0014862",
-    "UBERON_8000000",
-    "UBERON_0000071",
-    "UBERON_0014860",
-    "UBERON_0012101",
-    "UBERON_0000113",
-    "UBERON_0014858",
-    "UBERON_0007232",
-    "UBERON_0000070",
-    "UBERON_0000110",
-    "UBERON_8000002",
-    "UBERON_0014856",
-    "UBERON_0004728",
-    "UBERON_0034919",
-    "UBERON_0000108",
-    "UBERON_0000066",
-    "UBERON_0004707",
-    "UBERON_0000105",
-    "UBERON_0018241",
-    "UBERON_0007221",
-    "UBERON_0014406",
-    "UBERON_0014863",
-    "UBERON_0004729",
-    "UBERON_0014861",
-]
-
-
-# In[ ]:
-
-
-uc = [tissue_onto.search_one(iri=f"http://purl.obolibrary.org/obo/{c}") for c in uberon_classes]
-
-m_uberon = create_ancestors_mapping(tissue_onto, uc, "UBERON")  # type: ignore
-
-
-# In[ ]:
-
-
-with open("development_stage_ontology_mapping.json", "w") as f:
-    d = {}
-    d.update(m_human)
-    d.update(m_mouse)
-    d.update(m_uberon)
-    json.dump(d, f)
+if __name__ == "__main__":
+    # Load owl.info to grab latest ontology sources
+    owl_info_yml = "cellxgene_schema_cli/cellxgene_schema/ontology_files/owl_info.yml"
+    with open(owl_info_yml, "r") as owl_info_handle:
+        owl_info = yaml.safe_load(owl_info_handle)
+
+    # In[ ]:
+
+    human_latest_key = owl_info["HsapDv"]["latest"]
+    human_ontology = owl_info["HsapDv"]["urls"][human_latest_key]
+
+    human_onto = get_ontology(human_ontology)
+    human_onto.load()
+
+    m_human = create_ancestors_mapping(human_onto, human_onto.classes(), "HsapDv")  # type: ignore
+
+    # In[ ]:
+
+    mouse_latest_key = owl_info["MmusDv"]["latest"]
+    mouse_ontology = owl_info["MmusDv"]["urls"][mouse_latest_key]
+
+    mouse_onto = get_ontology(mouse_ontology)
+    mouse_onto.load()
+
+    m_mouse = create_ancestors_mapping(mouse_onto, mouse_onto.classes(), "MmusDv")  # type: ignore
+
+    # In[ ]:
+
+    uberon_latest_key = owl_info["UBERON"]["latest"]
+    uberon_ontology = owl_info["UBERON"]["urls"][uberon_latest_key]
+
+    tissue_onto = get_ontology(uberon_ontology)
+    tissue_onto.load()
+
+    # In[ ]:
+
+    uberon_classes = [
+        "UBERON_0007236",
+        "UBERON_0000106",
+        "UBERON_0014859",
+        "UBERON_0008264",
+        "UBERON_0007233",
+        "UBERON_0000112",
+        "UBERON_8000003",
+        "UBERON_0014857",
+        "UBERON_0009849",
+        "UBERON_0034920",
+        "UBERON_0000069",
+        "UBERON_0000109",
+        "UBERON_8000001",
+        "UBERON_0000068",
+        "UBERON_0018685",
+        "UBERON_0000107",
+        "UBERON_0007222",
+        "UBERON_0000092",
+        "UBERON_0018378",
+        "UBERON_0014864",
+        "UBERON_0004730",
+        "UBERON_0000111",
+        "UBERON_0007220",
+        "UBERON_0014405",
+        "UBERON_0014862",
+        "UBERON_8000000",
+        "UBERON_0000071",
+        "UBERON_0014860",
+        "UBERON_0012101",
+        "UBERON_0000113",
+        "UBERON_0014858",
+        "UBERON_0007232",
+        "UBERON_0000070",
+        "UBERON_0000110",
+        "UBERON_8000002",
+        "UBERON_0014856",
+        "UBERON_0004728",
+        "UBERON_0034919",
+        "UBERON_0000108",
+        "UBERON_0000066",
+        "UBERON_0004707",
+        "UBERON_0000105",
+        "UBERON_0018241",
+        "UBERON_0007221",
+        "UBERON_0014406",
+        "UBERON_0014863",
+        "UBERON_0004729",
+        "UBERON_0014861",
+    ]
+
+    # In[ ]:
+
+    uc = [tissue_onto.search_one(iri=f"http://purl.obolibrary.org/obo/{c}") for c in uberon_classes]
+
+    m_uberon = create_ancestors_mapping(tissue_onto, uc, "UBERON")  # type: ignore
+
+    # In[ ]:
+
+    with open("development_stage_ontology_mapping.json", "w") as f:
+        d = {}
+        d.update(m_human)
+        d.update(m_mouse)
+        d.update(m_uberon)
+        json.dump(d, f)
diff --git a/scripts/compute_mappings/tests/test_compute_ancestor_mapping.py b/scripts/compute_mappings/tests/test_compute_ancestor_mapping.py
@@ -0,0 +1,63 @@
+from unittest.mock import Mock
+
+from owlready2 import Ontology
+
+from scripts.compute_mappings.compute_ancestor_mapping import RO__PART_OF, get_ancestors
+
+
+class TestAncestorMapping:
+    class FakeThingClass:
+        def __init__(self, name, is_a=None, value=None, property=None):
+            self.name = name
+            self.is_a = is_a or []
+            self.value = value
+            self.property = property
+
+    #
+    # graph:
+    #
+    #   A   B
+    #    \ /
+    #     C   D
+    #      \ /
+    #       E
+    #
+
+    # @patch("scripts.compute_mappings.compute_ancestor_mapping.Ontology.search_one")
+    def test_get_ancestors(self):
+        # Arrange
+        A = self.FakeThingClass("obo.A_")
+        B = self.FakeThingClass("obo.B_")
+        C_A = self.FakeThingClass("C_to_A", value=A, property=self.FakeThingClass(RO__PART_OF))
+        C_B = self.FakeThingClass("C_to_A", value=B, property=self.FakeThingClass(RO__PART_OF))
+
+        # Should be skipped if missing 'value' attr
+        C_NA = self.FakeThingClass(
+            "should be skipped", value=self.FakeThingClass("NA"), property=self.FakeThingClass(RO__PART_OF)
+        )
+        delattr(C_NA, "value")
+
+        # Should be skipped if property.name is not RO__PART_OF
+        C_NA_2 = self.FakeThingClass(
+            "should be skipped", value=self.FakeThingClass("NA_2"), property=self.FakeThingClass("not_ro_part_of")
+        )
+
+        C = self.FakeThingClass("C_", is_a=[C_NA, C_NA_2, C_A, C_B])
+        D = self.FakeThingClass("D_")
+        E_C = self.FakeThingClass("C_to_A", value=C, property=self.FakeThingClass(RO__PART_OF))
+        E_D = self.FakeThingClass("C_to_A", value=D, property=self.FakeThingClass(RO__PART_OF))
+        E = self.FakeThingClass("E_", is_a=[E_C, E_D])
+
+        return_values = [D, B, A, C, E]
+        attrs = {"search_one.side_effect": lambda **x: return_values.pop()}
+        ontology = Mock(spec=Ontology, **attrs)
+
+        # Act
+        ancestors = get_ancestors(ontology, "E:")
+
+        # Assert
+        assert list(ancestors) == ["E_", "C_", "obo.A_", "obo.B_", "D_"]
+        print(ontology.call_args_list)
+
+    # def test_create_ancestors_mapping(self):
+    #     create_ancestors_mapping()