Missing relationships in concept_relationship

[ChaoPang]

Hi @swined @cgreich @cukarthik @dimshitc, I noticed that there are missing relationships in concept_relationship in the lastest vocabulary because those relationships do exist in an older version of the vocab (v5.0 22-APR-19). Here is an example below, where concept 40795740 is connected to 43055483 via an intermediate concept 43055731,

(This is from v5.0 22-APR-19)

Row	concept_id_1	concept_id_2	relationship_id	valid_start_date	valid_end_date	invalid_reason
1	43055731	43055483	Subsumes	1970-01-01	2099-12-31	null
2	40795740	43055731	Subsumes	2012-12-28	2099-12-31	null

But the second row is missing in the vocab (v5.0 06-NOV-20 although this is not the latest vocab, this relationship is missing from Athena too). This is only one example, I think there might be more such missing relationships, I am happy to provide such a list to you. I am wondering if you could take a look at this issue?

[mik-ohdsi]

Chao,
thanks for highlighting this. We will take a look into this.
Thanks!
Mik (for the vocabulary team)

[Alexdavv]

@ChaoPang
Please find the same issue and explanations here

[ChaoPang]

@Alexdavv thanks for linking up to the old issue. Just want to get a clarification on this, is there a plan to put LOINC parts (LONIC components) back into the hierarchy any time soon? The reason I ask is that we have analyses relying on concept_ancestor specifically on the hierarchy related to LONIC components, that's why we tried to reconstruct the missing hierarachy based on concept_relationship as the temporary solution before this gets fixed in concept_ancestor for good. However, during this reconstruction, we discovered those missing relationships.

[Alexdavv]

Just want to get a clarification on this, is there a plan to put LOINC parts (LONIC components) back into the hierarchy any time soon?

Not yet. We intentionally used SNOMED modeling in LOINC. It was discussed here.

If we really want the parts (and SNOMED attributes) to be the parents of the actual Measurements we need to make them Classifications. At the same time, we have to confirm they're not used as Standard concepts.
Makes sense?