Job request: 5692

version: "3.0"

expectations:
  population_size: 5000

actions:
  generate_study_population_1:
    run: cohortextractor:latest generate_cohort --study-definition study_definition --index-date-range "2019-09-01 to 2020-05-01 by month" --output-format feather
    outputs:
      highly_sensitive:
        cohort: output/input_*.feather

  generate_study_population_2:
    run: cohortextractor:latest generate_cohort --study-definition study_definition --index-date-range "2020-06-01 to 2021-02-01 by month" --output-format feather
    outputs:
      highly_sensitive:
        cohort: output/input*.feather

  generate_study_population_3:
    run: cohortextractor:latest generate_cohort --study-definition study_definition --index-date-range "2021-03-01 to 2021-09-01 by month" --output-format feather
    outputs:
      highly_sensitive:
        cohort: output/inpu*.feather

  generate_study_population_ethnicity:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_ethnicity --output-format feather
    outputs:
      highly_sensitive:
        cohort: output/input_ethnicity.feather

  join_ethnicity_region:
    run: python:latest python analysis/join_ethnicity_region.py
    needs:
      [
        generate_study_population_1,
        generate_study_population_2,
        generate_study_population_3,
        generate_study_population_ethnicity,
      ]
    outputs:
      highly_sensitive:
        cohort: output/inp*.feather

  filter_population:
    run: python:latest python analysis/filter_population.py
    needs: [join_ethnicity_region]
    outputs:
      highly_sensitive:
        cohort: output/input_filtered_*.feather

  calculate_numerators:
    run: python:latest python analysis/calculate_numerators.py
    needs: [filter_population]
    outputs:
      highly_sensitive:
        cohort: output/indicator_e_f_*.feather

  calculate_composite_indicators:
    run: python:latest python analysis/composite_indicators.py
    needs: [calculate_numerators, filter_population]
    outputs:
      moderately_sensitive:
        counts: output/*_composite_measure.csv

  generate_measures:
    run: cohortextractor:latest generate_measures --study-definition study_definition --output-dir=output
    needs: [filter_population]
    outputs:
      moderately_sensitive:
        measure_csv: output/measure_*_rate.csv

  generate_measures_demographics:
    run: python:latest python analysis/calculate_measures.py
    needs: [calculate_numerators, filter_population]
    outputs:
      moderately_sensitive:
        counts: output/indicator_measure_*.csv
        measure_csv: output/measure*_rate.csv
        demographics: output/demographics_summary_*.csv
  
  produce_stripped_measures:
    run: python:latest python analysis/stripped_measures.py
    needs:
      [
        generate_measures,
        generate_measures_demographics
      ]
    outputs:
      moderately_sensitive:
        measures: output/measure_stripped_*.csv

  generate_summary_counts:
    run: python:latest python analysis/summary_statistics.py
    needs:
      [
        filter_population,
        generate_measures,
        generate_measures_demographics,
        calculate_numerators,
      ]
    outputs:
      moderately_sensitive:
        patient_count: output/patient_count_*.json
        practice_count: output/practice_count_*.json
        summary: output/indicator_summary_statistics_*.json

  generate_plots:
    run: python:latest python analysis/plot_measures.py
    needs:
      [
        generate_measures,
        generate_measures_demographics,
        calculate_composite_indicators,
      ]
    outputs:
      moderately_sensitive:
        counts: output/figures/plot_*.jpeg
        combined: output/figures/combined_plot_*.png
        demographics: output/demographic_aggregates.csv
        medians: output/medians.json

  generate_notebook:
    run: jupyter:latest jupyter nbconvert /workspace/analysis/report.ipynb --execute --to html --template basic --output-dir=/workspace/output --ExecutePreprocessor.timeout=86400 --no-input
    needs: [generate_plots, generate_summary_counts]
    outputs:
      moderately_sensitive:
        notebook: output/report.html

  generate_dem_notebook:
    run: jupyter:latest jupyter nbconvert /workspace/analysis/demographic_report.ipynb --execute --to html --template basic --output-dir=/workspace/output --ExecutePreprocessor.timeout=86400 --no-input
    needs: [generate_plots]
    outputs:
      moderately_sensitive:
        notebook: output/demographic_report.html

  plot_Q1_comparisons:
    run: r:latest analysis/generate_demographic_slope_plot.R
    needs: [generate_plots]
    outputs:
      moderately_sensitive:
        plots: output/figures/SLOPE_*.png

  run_tests:
    run: python:latest python -m pytest --junit-xml=output/pytest.xml --verbose
    outputs:
      moderately_sensitive:
        log: output/pytest.xml

  test_population:
    run: python:latest python analysis/test_population.py
    needs: [filter_population]
    outputs:
      moderately_sensitive:
        counts: output/population_counts.csv
        count: output/patient_count_check.json