Database, models, Swagger API

[jmillanacosta]

• Created pydantic models for all sources of data (methods, tools, case studies, regulatory questions, zenodo, bioportal, compound...)
• Added seeding step of an in-memory SQLite instance when building docker, writes a cache every night
• Added a Swagger API
• Add a test (api_test.py) testing for data completeness and endpoint health.

TODO

• Don't use cached raw user content but GitHub API for the data JSONs?

[github-actions]

API check -- 2026-04-17 15:41 UTC

Entity counts

Entity	Count
Tools	50
Methods	20
Case studies	3
Regulatory questions	6
Stage explanations	7

Validation (field completeness)

Entity	Entries	Avg complete	Full
tools	50	42.9%	0/50
methods	20	69.7%	0/20
case_studies	3	87.5%	0/3
regulatory_questions	6	75.0%	0/6
stage_explanations	7	100.0%	7/7

Route health

Method	Route	Status
GET	/tools/	ok
GET	/tools/cdkdepict	ok
GET	/methods/	ok
GET	/methods/5_cfda_assay_to_determine_cytotoxicity	ok
GET	/regulatory-questions/	ok
GET	/stages/	ok
GET	/casestudies/	ok
GET	/casestudies/kidney	ok
GET	/compounds/Q2270	ok
GET	/compounds/Q2270/properties	ok
GET	/compounds/Q2270/identifiers	ok
GET	/compounds/Q2270/toxicology	ok
GET	/compounds/Q2270/experimental-data	ok
GET	/data/	ok
GET	/validation/	ok
GET	/validation/tools	ok

Result: PASS

[jmillanacosta]

@Maddocent @marvinm2 this data model can be the base for new MCP routes, thinking of e.g. an integrated site chatbot.

What's missing?

Please feel free to fork and develop on top of this branch