project.yaml

version: '3.0'

# Ignore this`expectation` block. It is required but not used, and will be removed in future versions.
expectations:
  population_size: 1000

actions:
  study_dates:
    run: r:latest analysis/metadates.R
    outputs:
      highly_sensitive:
        study_dates_json: output/study_dates.json

  generate_dataset_dm_algo:
    run: ehrql:v1 generate-dataset analysis/dataset_definition_dm_algo.py --output output/dataset_dm_algo.arrow
    needs: 
    - study_dates
    outputs:
      highly_sensitive:
        dataset: output/dataset_dm_algo.arrow
  
  data_process_dm_algo:
    run: r:latest analysis/data_process_dm_algo.R
    needs:
    - generate_dataset_dm_algo
    outputs:
      highly_sensitive:
        rds: output/data/data_processed_dm_algo.rds

  generate_dataset:
    run: ehrql:v1 generate-dataset analysis/dataset_definition_landmark.py --output output/dataset.arrow
    needs: 
    - study_dates
    outputs:
      highly_sensitive:
        dataset: output/dataset.arrow
  
  data_process:
    run: r:latest analysis/data_process.R
    needs:
    - generate_dataset
    - data_process_dm_algo
    outputs:
      highly_sensitive:
        rds: output/data/data_processed.rds
      moderately_sensitive:
        csv: output/data_properties/*.csv