feat(release): generate descendant mapping for tissues and cells

[Bento007]

Reason for Change

• feat(API): Generate mapping #15

Changes

• add script to generate descendant mapping. It's based off a script in single-cell-curation
• update GHA to install the local API to run generate descendant mapping.

Testing steps

• Added test cases
• Compared to tissue_descendants.json and cell_type_descendant.json generating by the single-cell-curation repo. Results:
• verified PR is auto generated for decendant mapping updates chore: update ontology decendant mappings #116
• Add GHA run every monday at midnight, by workflow_dispatch, or if the ontology assets are updated

------Comparing Ontology Guide and single-cell-curation Cell Type Descendant
KEYS: In Ontology Guide not in single-cell-curation
	 {'CL:4030031'}
DESCENDANT: In Ontology Guide  not in single-cell-curation
	 CL:0000738 {'CL:2000054'}
	 CL:0002320 {'CL:0002554'}
	 CL:0002319 {'CL:4033051', 'CL:4033052'}
	 CL:0000117 {'CL:4033051', 'CL:4033052'}
	 CL:0000057 {'CL:0002554'}
	 CL:0000540 {'CL:4033051', 'CL:4033052'}
	 CL:0000219 {'CL:2000054'}
	 CL:0000542 {'CL:2000054'}
	 CL:0000988 {'CL:2000054'}

KEYS: In single-cell-curation not in Ontology Guide
	 None
DESCENDANT: In single-cell-curation not in Ontology Guide

Notes for Reviewer

• Terms in the locally generated cell_type_descendants.json not in cellxgene_schema/cell_type_descendants.json
  • CL:4030031: interstitial cell
  • CL:0002494: cardiocyte was recently added to fix: CL_0002494 cell subclass term single-cell-data-portal#6756 and fix: CL_0002494 cell subclass term single-cell-curation#779. Not sure why it isn't showing up in the old version of the descendant mapping script.

[nayib-jose-gloria]

this gets us to parity with the current system, but I'm still concerned about the fact this step means the mappings become outdated as soon as a new CL term is introduced.

[nayib-jose-gloria]

to resolve, we'd either need to make the artifacts larger (how much larger?) to map all CL terms or perhaps we can set-up a mechanism to run this script periodically and update the mappings regularly.

[Bento007]

Based on the needs of the frontend end, we can update the descendant mappings outside of schema update. We should run this at a regular cadence.

[nayib-jose-gloria]

is there an equivalent need for cell culture? if not, why do we also tag cell culture terms?

[nayib-jose-gloria]

i'm assuming doing this achieves parity with the current set-up, but just to confirm--we don't want to include self as a descendant nor do we want to include an empty list? wouldn't this cause certain terms to be "orphaned" and not appear in the filters? or is that not how it works?

[nayib-jose-gloria]

should this be a dictionary if the value can be derived from the key? thinking this should be a set and, where needed, we can append " (organoid)"

[nayib-jose-gloria]

might be a comp bio question but--in doing this, we will always mark a term as having both ontology_term_id and ontology_term_id (organoid) as descendants, if the ontology_term_id is in the accept list and is an organoid. Is that intended? It sounds like that could make sense, but I'm not certain