biodata-models

Installation

biodata-models provides controlled vocabulary and registry models for biodata-schema. You shouldn't generally install this package in isolation.

Contributing

Install the dev dependencies:

pip install -e .[dev]

How to add a new model class

tl;dr

Add new classes to the _generators/models/*.csv files or create new files containing Enum-derived classes directly in the src folder.

Run ./run_all.sh in the top-level folder to rebuild models from their CSV files.

Details

The model class files, brain_atlas.py etc, are auto-generated. You should never need to modify the class files directly.

Instead, take a look at the jinja2 templates in the folder _generators/templates. The filename of the template is used to pull the corresponding .csv file and populate the data DataFrame. In the template you can pull data from the various columns and use them to populate each of the fields in your class.

To re-build all the models, run the run_all.sh bash script in the root folder, which loops through the template files and runs them through the generate_code function.

There are a few special cases, e.g. if data are missing in columns they will show up as float: nan. See the organizations.txt template for examples of how to handle this.

Documentation

Internal registries need to be enumerated in the aind-data-schema file src/aind_data_schema/utils/docs/registries_generator.py in the variable registries. This list controls what classes will have documentation automatically generated and cross-referenced correctly.

If you add a new external registry, you need to write the documentation manually in the aind-data-schema file docs/source/biodata_models/external.md.