- BiolinkValidator.PREDICATE_INCLUSIONS added (just with "biolink:int…

…eracts_with") to bypass 'mixin' Biolink Model validation error (pending full community review of the 'mixin' status of this predicate); unit tests modified to suit - plus a small internal DRY refactor in test suit with respect to LATEST_BIOLINK_MODEL_VERSION otherwise preparing for patch release 3.9.1
NCATSTranslator · Sep 11, 2023 · 8d0b452 · 8d0b452
1 parent b59956c
commit 8d0b452
Show file tree

Hide file tree

Showing 4 changed files with 113 additions and 48 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -2,6 +2,12 @@
 
 The Reasoner Validator package is evolving along with progress in TRAPI and Biolink Model standards within the NCATS Biomedical Knowledge Translator.
 
+## 3.9.1
+- Upgrade to Biolink Model Toolkit 1.1.2
+- Removed all residual references to 'sanitize_trapi_response()' (warning: validation with pre-release versions of TRAPI 1.4 earlier than 1.4.2 may trigger some funny false positive validation messages)
+- BiolinkValidator.PREDICATE_INCLUSIONS added (just with "biolink:interacts_with") to bypass 'mixin' Biolink Model validation error (pending full community review of the 'mixin' status of this predicate); unit tests modified to suit
+- plus a small internal DRY refactor in test suit with respect to LATEST_BIOLINK_MODEL_VERSION
+
 ## 3.9.0
 - Detect missing knowledge_graph names (resolves part of issue#35)
 - detection of uninformative QNodes (resolve issue#14)

diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "reasoner-validator"
-version = "3.9.0"
+version = "3.9.1"
 description = "Validation tools for Reasoner API"
 authors = [
     "Richard Bruskiewich <richard.bruskiewich@delphinai.com>",

diff --git a/reasoner_validator/biolink/__init__.py b/reasoner_validator/biolink/__init__.py
@@ -1225,6 +1225,10 @@ def validate_sources(self, edge_id: str, edge: Dict) -> Optional[str]:
 
             return source_trail  # may be empty if required RetrievalSource 'sources' entries are missing
 
+    # TODO: 11-Sept-2023: Certain specific 'mixin' predicates used in Knowledge Graphs are being validated for now
+    #       as 'warnings', for short term validation purposes (see reasoner-validator issue #97)
+    PREDICATE_INCLUSIONS = ["biolink:interacts_with"]
+
     def validate_predicate(
             self,
             edge_id: str,
@@ -1233,41 +1237,46 @@ def validate_predicate(
             source_trail: Optional[str] = None
     ):
         """
+        Validates predicates based on their meta-nature: existence, mixin, deprecation, etc.
+        with notable hard-coded explicit PREDICATE_INCLUSIONS exceptions.
+
         :param edge_id: str, identifier of the edge whose predicate is being validated
         :param predicate: str, putative Biolink Model predicate to be validated
         :param source_trail: str, putative Biolink Model predicate to be validated
         :param graph_type: TRAPIGraphType, type of TRAPI graph component being validated
         :return: None (validation communicated via class instance of method)
         """
-        graph_type_context: str = graph_type.name.lower()
-        if graph_type_context != "input_edge":
-            graph_type_context += ".edge"
-        context: str = f"{graph_type_context}.predicate"
-
-        # Validate the putative predicate as *not* being abstract, deprecated or a mixin
-        biolink_class = self.validate_element_status(
-            graph_type=graph_type,
-            context=context,
-            identifier=predicate,
-            edge_id=edge_id,
-            source_trail=source_trail
-        )
-        if biolink_class:
-            if not self.bmt.is_predicate(predicate):
-                self.report(
-                    code=f"error.{context}.invalid",
-                    source_trail=source_trail,
-                    identifier=predicate,
-                    edge_id=edge_id
-                )
-            elif self.minimum_required_biolink_version("2.2.0") and \
-                    not self.bmt.is_translator_canonical_predicate(predicate):
-                self.report(
-                    code=f"warning.{context}.non_canonical",
-                    source_trail=source_trail,
-                    identifier=predicate,
-                    edge_id=edge_id
-                )
+        if predicate not in self.PREDICATE_INCLUSIONS:
+
+            graph_type_context: str = graph_type.name.lower()
+            if graph_type_context != "input_edge":
+                graph_type_context += ".edge"
+            context: str = f"{graph_type_context}.predicate"
+
+            # Validate the putative predicate as *not* being abstract, deprecated or a mixin
+            biolink_class = self.validate_element_status(
+                graph_type=graph_type,
+                context=context,
+                identifier=predicate,
+                edge_id=edge_id,
+                source_trail=source_trail
+            )
+            if biolink_class:
+                if not self.bmt.is_predicate(predicate):
+                    self.report(
+                        code=f"error.{context}.invalid",
+                        source_trail=source_trail,
+                        identifier=predicate,
+                        edge_id=edge_id
+                    )
+                elif self.minimum_required_biolink_version("2.2.0") and \
+                        not self.bmt.is_translator_canonical_predicate(predicate):
+                    self.report(
+                        code=f"warning.{context}.non_canonical",
+                        source_trail=source_trail,
+                        identifier=predicate,
+                        edge_id=edge_id
+                    )
 
     @staticmethod
     def build_source_trail(sources: Optional[Dict[str, List[str]]]) -> Optional[str]:

diff --git a/tests/test_biolink_compliance_validation.py b/tests/test_biolink_compliance_validation.py
@@ -69,12 +69,16 @@ def test_minimum_required_biolink_version():
 
 
 def test_message():
-    reporter = BiolinkValidator(prefix="Test Message", trapi_version="v1.3", biolink_version="v3.5.4")
+    reporter = BiolinkValidator(
+        prefix="Test Message",
+        trapi_version="v1.3",
+        biolink_version=f"v{LATEST_BIOLINK_MODEL_VERSION}"
+    )
     assert reporter.get_trapi_version() == "v1.3.0"
 
     # Note: BMT is a bit tricky in resolving Biolink Model versions:
     # the version is reported without the 'v' prefix of the GitHub release!
-    assert reporter.get_biolink_version() == "3.5.4"
+    assert reporter.get_biolink_version() == LATEST_BIOLINK_MODEL_VERSION
 
     assert not reporter.has_messages()
     reporter.report("info.compliant")
@@ -2884,6 +2888,7 @@ def test_validate_biolink_curie_in_qualifiers(
             {
                 "nodes": {
                     "HGNC:3059": {
+                        "name": "Heparin binding EGF like growth factor",
                         "categories": [
                            "biolink:Gene"
                         ],
@@ -2900,7 +2905,7 @@ def test_validate_biolink_curie_in_qualifiers(
                 "edges": {
                     "edge_1": {
                         "subject": "HGNC:3059",
-                        "predicate": "biolink:interacts_with",
+                        "predicate": "biolink:increases_amount_or_activity_of",
                         "object": "HGNC:391",
                         "attributes": [
                             {
@@ -2921,10 +2926,55 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 20: predicate is abstract - not allowed in Knowledge Graphs
+            # Query 20: predicate is a mixin - not allowed in Knowledge Graphs, but "biolink:interacts_with"
+            #           is tagged as an exception (see https://github.com/NCATSTranslator/reasoner-validator/issues/97)
+            {
+                "nodes": {
+                    "HGNC:3059": {
+                        "name": "Heparin binding EGF like growth factor",
+                        "categories": [
+                           "biolink:Gene"
+                        ],
+                        "description": "heparin binding EGF like growth factor"
+                    },
+                    "HGNC:391": {
+                        "name": "AKT serine/threonine kinase 1",
+                        "categories": [
+                            "biolink:Gene"
+                        ],
+                        "description": "AKT serine/threonine kinase 1"
+                    }
+                },
+                "edges": {
+                    "edge_1": {
+                        "subject": "HGNC:3059",
+                        "predicate": "biolink:interacts_with",
+                        "object": "HGNC:391",
+                        "attributes": [
+                            {
+                                "attribute_type_id": "biolink:stoichiometry",
+                                "value": 2
+                            }
+                        ],
+                        "sources": [
+                            {
+                                "resource_id": "infores:molepro",
+                                "resource_role": "primary_knowledge_source"
+                            }
+                        ]
+                    }
+                }
+            },
+            ""  # predicate "biolink:interacts_with" is now on
+                # the PREDICATE_INCLUSIONS list so 'mixin' validation is skipped
+        ),
+        (
+            LATEST_BIOLINK_MODEL_VERSION,
+            # Query 21: predicate is abstract - not allowed in Knowledge Graphs
             {
                 "nodes": {
                     "PMID:1234": {
+                        "name": "article title",
                         "categories": [
                            "biolink:InformationContentEntity"
                         ],
@@ -2962,7 +3012,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 21: predicate is non-canonical
+            # Query 22: predicate is non-canonical
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
                 "edges": {
@@ -2989,7 +3039,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 22: 'object' id is missing from the nodes catalog
+            # Query 23: 'object' id is missing from the nodes catalog
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
                 "edges": {
@@ -3016,7 +3066,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 23: attribute 'attribute_type_id' is missing
+            # Query 24: attribute 'attribute_type_id' is missing
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
                 "edges": {
@@ -3042,7 +3092,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 24: attribute 'value' is missing?
+            # Query 25: attribute 'value' is missing?
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
                 "edges": {
@@ -3069,7 +3119,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 25: 'attribute_type_id' is not a CURIE
+            # Query 26: 'attribute_type_id' is not a CURIE
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
                 "edges": {
@@ -3096,7 +3146,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 26: 'attribute_type_id' is a Biolink (node) category - not sure yet
+            # Query 27: 'attribute_type_id' is a Biolink (node) category - not sure yet
             #           whether this reflects "best practices" so issue a warning for now
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
@@ -3124,7 +3174,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 27: 'attribute_type_id' is a Biolink (edge) predicate - not sure yet
+            # Query 28: 'attribute_type_id' is a Biolink (edge) predicate - not sure yet
             #           whether this reflects "best practices" so issue a warning for now
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
@@ -3152,7 +3202,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 28: 'attribute_type_id' is not a 'biolink:association_slot' (biolink:synonym is a node property)
+            # Query 29: 'attribute_type_id' is not a 'biolink:association_slot' (biolink:synonym is a node property)
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
                 "edges": {
@@ -3179,7 +3229,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 29: 'attribute_type_id' has a 'biolink' CURIE prefix and
+            # Query 30: 'attribute_type_id' has a 'biolink' CURIE prefix and
             #           is an association_slot, so it should pass
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
@@ -3211,7 +3261,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             LATEST_BIOLINK_MODEL_VERSION,
-            # Query 30: 'attribute_type_id' has a CURIE prefix namespace unknown to Biolink?
+            # Query 31: 'attribute_type_id' has a CURIE prefix namespace unknown to Biolink?
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
                 "edges": {
@@ -3236,7 +3286,7 @@ def test_validate_biolink_curie_in_qualifiers(
             },
             "warning.knowledge_graph.edge.attribute.type_id.non_biolink_prefix"
         ),
-        (   # Query 31:  # An earlier Biolink Model won't recognize a category not found in its specified release
+        (   # Query 32:  # An earlier Biolink Model won't recognize a category not found in its specified release
             "1.8.2",
             {
                 # Sample nodes
@@ -3265,7 +3315,7 @@ def test_validate_biolink_curie_in_qualifiers(
             },
             "error.knowledge_graph.node.category.unknown"
         ),
-        (   # Query 32:  # 'attribute_type_id' has a CURIE prefix namespace unknown to Biolink but...
+        (   # Query 33:  # 'attribute_type_id' has a CURIE prefix namespace unknown to Biolink but...
             SUPPRESS_BIOLINK_MODEL_VALIDATION,
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
@@ -3295,7 +3345,7 @@ def test_validate_biolink_curie_in_qualifiers(
         ),
         (
             SUPPRESS_BIOLINK_MODEL_VALIDATION,
-            # Query 33: 'attribute_type_id' is not a 'biolink:association_slot'
+            # Query 34: 'attribute_type_id' is not a 'biolink:association_slot'
             #           (biolink:synonym is a node property) but...
             {
                 "nodes": SIMPLE_SAMPLE_NODES,
@@ -3326,7 +3376,7 @@ def test_validate_biolink_curie_in_qualifiers(
         (
             LATEST_BIOLINK_MODEL_VERSION,
 
-            # Query 34: Knowledge Graph dangling nodes error
+            # Query 35: Knowledge Graph dangling nodes error
             {
                 # Sample nodes with extra  unused node
                 'nodes': SAMPLE_NODES_WITH_UNUSED_NODE,