cellxgene-schema must update validation for disease_ontology_term_id

[brianraymor]

Context

See disease_ontology_term_id.

4.0.0 was:

This MUST be a MONDO term or "PATO:0000461" for normal or healthy

4.1.0 adds further requirements:

This MUST be one of:

• "PATO:0000461" for normal or healthy
• the most accurate child of "MONDO:0000001"for disease
• "MONDO:0021178" for injury or preferably its most accurate child

[joyceyan]

@brianraymor @danieljhegeman Is there any way for us to know if the MONDO term provided is the most accurate term for the disease / injury? Or would the work on the validation side just be to verify that a MONDO term provided is a child of either MONDO:0000001 or MONDO:0021178?

[brianraymor]

Is there any way for us to know if the MONDO term provided is the most accurate term for the disease / injury?

No. This language is a common pattern for schema fields for curators to encourage the submitter to not use a general term (which often happens in the first iteration). Curators will often review the research and request the submitter to update the precision.

[danieljhegeman]

most accurate child

@brianraymor please confirm: a child is a direct descendant, and not a multi-level descendant, e.g. B is a child of A but C is not a child of A in: A > B > C where A is the foremost ancestor. cc @joyceyan

[danieljhegeman]

Implicitly V6 is not a child of V9

[brianraymor]

Let's use a concrete, simple case - MONDO.

It's any descendant of the disease root:

which unfortunately allows non-human animal disease to be specified for human observations at the moment per my comments in #cell-science-data-wrangling.

[danieljhegeman]

@brianraymor Ok, thanks for clarifying. I'll make a ticket to change the language in the schema and code base.

[danieljhegeman]

(original issue: Joyce's feature change) Approved by Lattice