The September 2021 CA Edition is now available in the SNOMED International Browser!

0199b0ce-50c0-cae8-b314-34c558501195	20210930	0	20621000087109	733073007	844580005	SubClassOf(:844580005 :27841000087102)
3a6f0f27-adf8-50a6-af49-4488bf1884f7	20210930	1	20621000087109	733073007	27841000087102	SubClassOf(:844580005 :27841000087102)

51a8f312-980e-0df3-d685-b3cf68bfb974	20210930	0	20621000087109	733073007	23021000087106	SubClassOf(:23021000087106 :27841000087102)
1e07e7e1-1ac3-754c-bbbd-199e90bff17d	20210930	1	20621000087109	733073007	27841000087102	SubClassOf(:23021000087106 :27841000087102)

It is not a good idea to use GCI axiom for such a requirement as it is in the later change below:

Referenced component id: 27841000087102
OWL expression: SubClassOf(:844580005 :27841000087102)

There is a section on the "Referenced component for an Axiom" in the OWL Guide . We have specified how the referenced component should be allocated for each axiom. The referenced component of a GCI axiom can be in the following two situations. It is explicitly stated that C is not a precoorinated concept, but the concept 84458005 is a precoordinated concept. Therefore, this is not a GCI axiom.

• If an axiom is in the form of SubClassOf(C D) where concept C is NOT a precoordinated concept and D is a precoordinated concept. This is a General Concept Inclusion (GCI) axiom. Because concept C is a sufficient but not necessary condition for concept D, the concept ID of D should be the referencedComponentId for this axiom.
• If an axiom is in the form of SubClassOf(C D) or EquivalentClasses(C D) where both C and D are NOT precoordinated concepts, this is a GCI axiom and the referencedComponentId should be 733929006 |General concept inclusion axiom (metadata)|.

Our technical team asked me a similar question if C could be a precoordinated concept when they were implementing GCI axioms. I think we can make the OWL Guide more explicit to state what is not allowed as well to avoid potential misinterpretation, such as:

If C is a precoordinated concept, this SubclassOf axiom is NOT a GCI axiom and the concept D should not be assigned as the referencedComponentId. This would be a normal SubclassOf axiom and C is the referenced component.

The referencedComponentId has no impact on the semantics of the axioms and it is not used for any reasoning. The international tool follows the above statements. Therefore, it is correct that the axioms are still represented as multiple subclasses axioms against the concept 84458005.

However, we could introduce a QA in our toolset to check if C is a precoordinated concept. If it is not, concept D should not be assigned as a referencedComponentId. This will identify the issues you reported here and provide feedback on axioms promoted from extensions.

The original modelling is correct because it is an additional axiom for the concept 844580005 from the international release. This approach conforms to the axiom addition without inclusion in the OWL Guide as below:

Axiom without inclusion - a new axiom without inclusion of the axiom from the international release as part of the extension axiom. For example:

• SubClassOf(A C) is an axiom in the international release.
• SubClassOf(A D) is a new axiom in the extension.

• Add another subclass axiom to the core concept 844580005 |Antigen of inactivated whole Human alphaherpesvirus 3 (substance)|, in the Canadian extension module. This approach is compliant with extension management policies.
• Avoid modifying the core concept and just add two General Concept Inclusion (GCI) axioms to the Canadian primitive concept 27841000087102 |Antigen of non-live Human alphaherpesvirus 3 (substance)|, representing that the Canadian concept is a superclass of the core concept (instead of saying that the core concept is a subclass of the Canadian grouper).

SubClassOf(
	:27841000087102 |Antigen of non-live Human alphaherpesvirus 3 (substance)|
	:260214004 |Antigen of Human alphaherpesvirus 3 (substance)|
)
SubClassOf(
	:23021000087106
	:27841000087102 |Antigen of non-live Human alphaherpesvirus 3 (substance)|
)
SubClassOf(
	:844580005
	:27841000087102 |Antigen of non-live Human alphaherpesvirus 3 (substance)|
)

1. We will research what might be different in this concept definition compared with IE concept definitions containing GCIs, confusing the browser, and tools importing the axioms. I will follow up on that matter in one or two business days after the long weekend. Note that the inferred form is correct. Only the axiom representation seems to be confusing in some tool implementations.
2. The use case (grouping a core concept under an extension concept) pattern has been used elsewhere in national extensions. Still, we can discuss whether the preferred pattern to represent the intended semantics would be through GCIs or secondary axioms at the Vaccines working group.
3. Debbie has previously reported problems with her tooling importing Canadian definitions. We have been researching those reports and will review this one. Some of the patterns that cause problems were discontinued in the current version of the extension, even if they are semantically correct and conformant with the specifications (to give time for further evaluation and discussion). A few others might require collaboration with all parties involved to understand the best way better forward.