Closed fedorgrab closed 6 months ago
Meta question: how does the data get ingested into postgres? We should document that in the readme
Meta question: how does the data get ingested into postgres? We should document that in the readme
About that, we have cell data lifecycle scripts, one of which is 'ingest'. I believe there should be significant refactoring there (they've been used for bq ingest). Also, this seems like it should be a separate PR to me.
Meta question: how does the data get ingested into postgres? We should document that in the readme
About that, we have cell data lifecycle scripts, one of which is 'ingest'. I believe there should be significant refactoring there (they've been used for bq ingest). Also, this seems like it should be a separate PR to me.
Yep, totally agreed. Just wanted to make sure it's on the radar
Meta question: how does the data get ingested into postgres? We should document that in the readme
About that, we have cell data lifecycle scripts, one of which is 'ingest'. I believe there should be significant refactoring there (they've been used for bq ingest). Also, this seems like it should be a separate PR to me.
Yep, totally agreed. Just wanted to make sure it's on the radar
Oh totally there's even an old issue for it.
Changes:
Cell Metadata models and querying. Check out how a temp table query could be executed in SQLalchemy, no raw SQL required, which seems p cool.
Admin restructure. Since we now getting more and more models in admin, seems reasonable to group them together. Now admin page nav bar looks like:
Where All user related models go under Users, all ML realted models go under ML Management and all Cell Metadata models go under Cell Data management.
Consensus Engine which includes ontology_aware (new method that uses postgres for cell metadata fetching), and the old method which is now called cell_type_count. Old method should be deprecated and rewritten with use of postgres as well, but not in this PR.:)
In local environment SQLAlchemy will print all SQL queries
Inspired by @nmalfroy 's PR (#141), adding more pydantic models to wrap internal data flows into the python data classes