v4.2.7

• Fix validation reporting to handle odd validation message parameters which are not scalars
• Also coerce a few specific validation message report call parameter inputs to strings
• Skipping Ontology (UberGraph) unit tests due to some unknown change in the resource (requires review since this could break some extended use cases of the library)

TRAPI and Biolink Model release checking in TRAPI Response

• Extracted test (from the graph-validation-test-runner project a.k.a. former SRI_Testing harness) for schema-level validation of TRAPI and Biolink Model versioning, moved into check_compliance_of_trapi_response() method.

fuzzy concept node identifier matching

• resolve_testcase_node() method calling get_aliases() now reports certain outlier cases of missing or fuzzy ("non-canonical" namespace letter case) matching of subject or object concept node identifier matching.
• updated bmt to 1.4.3 to cover some fresh issues relating to (deliberately) unknown elements in reasoner-validator test data

Full May 28, 2024 TRAPI 1.5 release plus some enhanced support for OneHop edge data discovery

• Update TRAPI to full minor release 1.5
• Refactored validator of test case node (onehop) a bit further to sharpen scope, including new resolve_testcase_node() method readability and to ensure full reporting of validation error message inside
• Bug fix to ontology - test for null values returned from the node normalizer; add caching for the parent concept lookup
• refactor to break apart the testcase edge validation; added; added validation of subclassed instances of a query identifier (should be allowed); identified 'query_id' not yet used for results validation.

Compensate for NodeNormalization identifier case discrepancy, affecting testcase node matches

• In some fringe cases, the Node Normalizer sends by equivalent identifiers with CURIE namespaces letter case which departs from expectation. This makes comparisons to aliases a bit fragile. In such cases, the original identifier letter case preference is taken as the desired one to return in the alias list.
• Some missing unit tests added against utils

Finer grained matching of node categories in validation of One Hop test output

• Validator case_\* functions renamed to testcase_\* for clarity of function
• testcase_node comparisons tweaked to check exact matches or matches against all parent categories of the specified testcase category and observed KG node categories. Less precise (more generic) categorical matches trigger a validation warning.

Partly restructured repo layout; extraction of 'ontology_kp' code from graph-validation-test-runners repo

• consolidated utility functions into 'utils'
• extracted 'ontology_kp' from graph-validation-test-runners repository (former SRI_Testing ontology_kp) and into reasoner-validator under 'biolink'; copied associated unit tests over from graph-validation-test-runners repository.

v4.1.9

• temporarily removed biolink:treats predicate validation check for biolink:supporting_graphs attribute value, pending community-wide implementation of the recently specified 'treats' predicate knowledge graph annotation standards.

v4.1.8

• BMT update to 1.4.1 (with default Biolink Model 4.2.1)
• implemented rudimentary biolink:treats predicate validation check for biolink:supporting_graph attribute value.

Enhanced mixin predicate handling

• Fixed mixin predicates to be allowed in both Query Graph and Knowledge Graph contexts; however, this may be overridden by explicitly setting strict_validation to true