Bugfix/kbdev 1424 ncit parent records name

[mathieulemieux]

No description provided.

[mathieulemieux]

When tested on a blank DB, ~25k/150k records less than current GKB, probably because now the aliassed record are not an alias of themselves anymore.

[mathieulemieux]

These warnings in the output log:

[...]
2026-02-13T16:13:19.033Z warn: one or more records in relationship not loaded: c121583, c62277
[...]

Seems to be related to parent records being obsolete, but still referenced as parent in the child record:

$ header="id xmlTag parents synonyms definition name conceptStatus semanticType conceptInSubset"
$ grep -i $'^c121583\t' Thesaurus.txt | tojson -H $header
[
  {
    "id": "C121583",
    "xmlTag": "<http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#C121583>",
    "parents": "C62277",
    "synonyms": "Non-Cutaneous Leiomyosarcoma|Leiomyosarcoma (Excluding Skin)",
    "definition": "A leiomyosarcoma arising from an anatomic site other than skin.",
    "name": "",
    "conceptStatus": "",
    "semanticType": "Neoplastic Process",
    "conceptInSubset": ""
  }
]

$ grep -i $'^c62277\t' Thesaurus.txt | tojson -H $header
[
  {
    "id": "C62277",
    "xmlTag": "<http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#C62277>",
    "parents": "C45230",
    "synonyms": "Connective and Soft Tissue Neoplasm by Obsolete Classification|Connective and Soft Tissue Neoplasm (Antiquated)",
    "definition": "A connective and soft tissue neoplasm that is categorized using nomenclature that is no longer in use or diagnostic criteria that are outdated.",
    "name": "",
    "conceptStatus": "Obsolete_Concept",
    "semanticType": "Neoplastic Process",
    "conceptInSubset": ""
  }
]

Making child concept (here C121583) without a --SubClassOf-> <parent>.
Not sure what we should do about it. Any thoughts @elewis2 ? Actually I suggest we keep it that way.

Actually, not sure what to do about it @elewis2. We are already keeping old terminology in GraphKb, but don't upload obsolete terms; I'm wondering why? When a term is available but not linked to a parent, that's problematic, like here a sarcoma not linked to cancer.

[mathieulemieux]

So far, upload tests on new DB are going well. I'll post the log here when done.

[mathieulemieux]

ncit-output-dev.log

Warnings about some 1864 SubClassOf not being loaded are normal. Parent term not part of our selective upload.

[mathieulemieux]

ok, re-ran the loader with changes made to facilitate the datafix. Exact same results as last run. Here is the log:
ncit_26feb.log

[mathieulemieux]

Loader haas been run on staging 2 days ago, with the --ignoreSynonyms flag. So far evrething seems ok

[mathieulemieux]

@sshu and @elewis2, anything else you would need to help with the review?