Job request: 2026

######################################

# This script defines the project pipeline - it specifys the execution orders for all the code in this
# repo using a series of actions.

######################################

version: '3.0'

expectations:
  population_size: 1000000

actions:

  # Extract data for study population flow chart
  generate_study_population_flow_chart_data:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_flow_chart
    outputs:
      highly_sensitive:
        cohort: output/input_flow_chart.csv

  # Calculate numbers for study population flow chart
  flow_chart:
    run: r:latest -e 'rmarkdown::render("analysis/R/Markdown/Study_definition_flow_chart.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [generate_study_population_flow_chart_data]
    outputs:
      moderately_sensitive:
        html: output/Study_definition_flow_chart.html
        
  # Extract study data
  generate_study_population:
    run: cohortextractor:latest generate_cohort --study-definition study_definition
    outputs:
      highly_sensitive:
        cohort: output/input.csv
        
  # Process data
  data_process:
    run: r:latest analysis/R/Scripts/00_process_data.R
    needs: [generate_study_population]
    outputs:
      highly_sensitive:
        data1: output/data/data_all.rds
        data2: output/data/data_modelling.rds
        
  # Summarise data
  data_properties:
    run: r:latest analysis/R/Scripts/01_data_properties.R output/data/data_all.rds output/data_properties
    needs: [generate_study_population, data_process]
    outputs:
      moderately_sensitive:
        datasummary: output/data_properties/data_all*.txt
        
  # More data summaries
  data_summaries:
    run: r:latest -e 'rmarkdown::render("analysis/R/Markdown/Data_summaries.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [generate_study_population, data_process]
    outputs:
      moderately_sensitive:
        html: output/Data_summaries.html

  # # Coxph models
  # cox_models:
  #   run: r:latest analysis/R/Scripts/02a_models_coxph.R
  #   needs: [generate_study_population, data_process]
  #   outputs:
  #     highly_sensitive :
  #       models: output/models/testing/mod*.rds
        
  # # Coxme model
  # coxme_model:
  #   run: r:latest analysis/R/Scripts/02b_models_coxme.R
  #   needs: [generate_study_population, data_process]
  #   outputs:
  #     highly_sensitive :
  #       models: output/models/testing/mod_test*.rds
  
  # # Model summaries
  # cox_models_summaries:
  #   run: r:latest -e 'rmarkdown::render("analysis/R/Markdown/Model_comparisons.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
  #   needs: [generate_study_population, data_process, cox_models, coxme_model]
  #   outputs:
  #     moderately_sensitive:
  #       html: output/Model_comparisons.html
  
  # AFT vs Cox models feasibility
  aft_vs_cox_poc:
    run: r:latest analysis/R/Scripts/02c_models_cox_vs_aft_poc.R
    needs: [generate_study_population, data_process]
    outputs:
      highly_sensitive :
        models: output/models/testing/cox_vs_aft/mod*.rds
      moderately_sensitive:
        tables: output/models/testing/cox_vs_aft/table*.csv
        plots: output/models/testing/cox_vs_aft/plot*.svg

  # AFT models
  aft_vs_cox_models:
    run: r:latest analysis/R/Scripts/02d_models_cox_vs_aft.R
    needs: [generate_study_population, data_process]
    outputs:
      highly_sensitive :
        models: output/models/cox_vs_aft/mod*.rds
        
  # Model comparison
  aft_vs_cox_comparison:
    run: r:latest -e 'rmarkdown::render("analysis/R/Markdown/Cox_vs_aft_model_comparison.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [generate_study_population, data_process, aft_vs_cox_models]
    outputs:
      moderately_sensitive:
        html: output/Cox_vs_aft_model_comparison.html
        
  # Final (stratified cox) model
  model_final:
    run: r:latest analysis/R/Scripts/03_model_final.R
    needs: [generate_study_population, data_process]
    outputs:
      highly_sensitive :
        models: output/models/final/mod*.rds
 
  # Strata summaries
  strata_summary:
    run: r:latest analysis/R/Scripts/04_strata_summary.R
    needs: [data_process, model_final]
    outputs:
      moderately_sensitive:
        plots: output/models/final/plot_strata*.svg
  
  # Results summary
  results_summary:
    run: r:latest -e 'rmarkdown::render("analysis/R/Markdown/Results_summary.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [data_process, model_final]
    outputs:
      moderately_sensitive:
        html: output/Results_summary.html
        
  # Table and figure
  table_and_figure:
    run: r:latest analysis/R/Scripts/05_table_and_figure.R
    needs: [generate_study_population, data_process, model_final]
    outputs:
      moderately_sensitive:
        tables: output/models/final/tab*.csv
        plots: output/models/final/plot*.svg