alignment files including parents that are obsolete in Mondo

[sabrinatoro]

for example: slurp/ordo.tsv file. The parents reported for new terms includes

• MONDO id that are obsolete
• MONDO:0000001 (=disease), which is too high level.

These obsolete or too high-level terms should not be brought in as parents Example from slurp/ordo.tsv

MONDO:0958100 | autoinflammatory syndrome with acne and/or hidradenitis suppurativa | Orphanet:653434 | MONDO:equivalentTo | Autoinflammatory syndrome with acne and/or hidradenitis suppurativa |   |   | MONDO:8000033|MONDO:0017954|MONDO:0017370

Note: I understand that it might be problematic if only terms that have parents in Mondo are brought in. If it is the case, these terms can be brought in, but parents should not be included in the spreadsheet.

[matentzn]

• Obsolete terms should never be included in the report - I hope this was a syncing issue. (bug)
• Can you provide us with a simple exclude list for parents? "too high level" is hard to operationalise, I think it would be more effective to just say, exclude "disease" "anatomical disease" etc. I
• Can you provide an short statement on why "too high parents" are confusing? I am not that versed in curation, I would have though since a human needs to review the parents, they woul in any case reconsider the classification (whether teh parent field is empty or not).

[matentzn]

@twhetzel this should be prioritised right after ICD11 and MedGen probably. I lost a bit track of Joes priorities now, so I leave it to you to fold this into the schedule?

[sabrinatoro]

Can you provide us with a simple exclude list for parents? "too high level"

Currently, the "too high level" term that is consistently reported is MONDO:0000001 =disease = the highest term in the ontology (parent of "human disease" and "non-human animal disease").

Can you provide a short statement on why "too high parents" are confusing?

"disease" as a parent, is not specific enough to be useful. Every term should at least be either in the "human disease" branch or in the "non-human animal disease" branch. From a curation perspective, if a term has the parent "disease" or even "human disease", we will have to review this term and find a more specific parent, minimally one of the "high-level classification" term for human diseases (see list below)

Mondo ID	term name
MONDO:0002409	auditory system disorder'
MONDO:0002657	breast disorder'
MONDO:0045024	cancer or benign tumor'
MONDO:0004995	cardiovascular disorder'
MONDO:0019040	chromosomal disorder'
MONDO:0003900	connective tissue disorder
MONDO:0004335	digestive system disorder'
MONDO:0021147	disorder of development or morphogenesis'
MONDO:0002022	disorder of orbital region'
MONDO:0024458	disorder of visual system'
MONDO:0005151	endocrine system disorder'
MONDO:0005570	hematologic disorder'
MONDO:0003847	hereditary disease'
MONDO:0043543	iatrogenic disease'
MONDO:0700007	idiopathic disease'
MONDO:0005046	immune system disorder'
MONDO:0005550	infectious disease'
MONDO:0021166	inflammatory disease'
MONDO:0002051	integumentary system disorder'
MONDO:0005066	metabolic disease'
MONDO:0044970	mitochondrial disease'
MONDO:0006858	mouth disorder'
MONDO:0002081	musculoskeletal system disorder'
MONDO:0005071	nervous system disorder'
MONDO:0005137	nutritional disorder'
MONDO:0700003	obstetric disorder'
MONDO:0100366	occupational disorder'
MONDO:0024623	otorhinolaryngologic disease'
MONDO:0100086	perinatal disease'
MONDO:0029000	poisoning
MONDO:0021669	post-infectious disorder'
MONDO:0002025	psychiatric disorder'
MONDO:0043459	radiation-induced disorder'
MONDO:0005039	reproductive system disorder'
MONDO:0005087	respiratory system disorder'
MONDO:0002254	syndromic disease'
MONDO:0043839	ulcer disease'
MONDO:0044991	upper digestive tract disorder'
MONDO:0002118	urinary system disorder'

[matentzn]

From a curation perspective, if a term has the parent "disease" or even "human disease", we will have to review this term and find a more specific parent, minimally one of the "high-level classification" term for human diseases (see list below)

I think our SOP should really include a moment of pause here (this is exactly why I was asking). I personally hoped the "parent" was mere a suggestion and is always carefully reviewed during migration. This is why I was not originally worried to include very high level parents - because I knew someone was looking at them anyways and throw them out..

[sabrinatoro]

I think our SOP should really include a moment of pause here (this is exactly why I was asking). I personally hoped the "parent" was mere a suggestion and is always carefully reviewed during migration. This is why I was not originally worried to include very high level parents - because I knew someone was looking at them anyways and throw them out..

I see where you come from, and I want to reassure you that a curator reviews the list of suggested parents before creating the new terms. In some cases, 5 parents are suggested (IDs separated by a pipe), so it is a lot of copy/paste and manual removal. It is easy to recognize MONDO:0000001, and remove it from parents, so it is not a big issue. But since we will never add it as a parent, it is not useful to have it reported (but again, it might not be worth the technical work to exclude it from the parent list).

[matentzn]

This makes sense now, thank you @sabrinatoro :)