Job request: 6866

version: '3.0'

expectations:
  population_size: 1000

actions:
  
  generate_study_population:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population
    outputs:
      highly_sensitive:
        cohort: output/input_general_population.csv

  generate_study_population_covid_primarycare:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_covid_primarycare
    outputs:
      highly_sensitive:
        cohort: output/input_covid_primarycare.csv
  
  generate_study_population_covid_SGSS:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_covid_SGSS
    outputs:
      highly_sensitive:
        cohort: output/input_covid_SGSS.csv
  
  generate_study_population_covid_admission:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_covid_admission
    outputs:
      highly_sensitive:
        cohort: output/input_covid_admission.csv
  
  # generate_study_population_covid_icu:
  #   run: cohortextractor:latest generate_cohort --study-definition study_definition_covid_icu
  #   outputs:
  #     highly_sensitive:
  #       cohort: output/input_covid_icu.csv

  generate_study_population_covid_death_ons:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_covid_death_ons
    outputs:
      highly_sensitive:
        cohort: output/input_covid_death_ons.csv
  
  generate_study_population_covid_death_cpns:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_covid_death_cpns
    outputs:
      highly_sensitive:
        cohort: output/input_covid_death_cpns.csv
  


  # process_1_0:
  #   run: r:latest analysis/process_1_0.R
  #   needs: [generate_study_population_covid_primarycare, generate_study_population_covid_SGSS]
  #   outputs:
  #     highly_sensitive:
  #       cohort1: output/covid_infection_1_1.csv
  #       cohort2: output/covid_infection_1_2.csv
  #       cohort3: output/covid_infection_1_3.csv
  #       cohort4: output/covid_infection_1_4.csv
  #       cohort5: output/covid_infection_1_5.csv
  #       cohort6: output/covid_infection_1_6.csv
  #       cohort7: output/covid_infection_1_7.csv
  #       cohort8: output/covid_infection_1_8.csv
  #       cohort9: output/covid_infection_1_9.csv
  #       cohort10: output/covid_infection_1_10.csv
  #       cohort11: output/covid_infection_2_1.csv
  #       cohort12: output/covid_infection_2_2.csv
  #       cohort13: output/covid_infection_2_3.csv
  #       cohort14: output/covid_infection_2_4.csv
  #       cohort15: output/covid_infection_2_5.csv
  #       cohort16: output/covid_infection_2_6.csv
  #       cohort17: output/covid_infection_2_7.csv
  #       cohort18: output/covid_infection_2_8.csv
  #       cohort19: output/covid_infection_2_9.csv
  #       cohort20: output/covid_infection_2_10.csv

  process_1:
    run: r:latest analysis/process_1.R
    needs: [generate_study_population_covid_primarycare, generate_study_population_covid_SGSS]
    outputs:
      highly_sensitive:
       # cohort1.1: output/case_covid_infection.csv
        cohort1.2: output/control_covid_infection.csv 

  process_1.2:
    run: r:latest analysis/process_1.2.R
    needs: [process_1]
    outputs:
      highly_sensitive:
       # cohort1.1: output/case_covid_infection.csv
        cohort1: output/covid_infection_2002.csv
        cohort2: output/covid_infection_2003.csv
        cohort3: output/covid_infection_2004.csv
        cohort4: output/covid_infection_2005.csv
        cohort5: output/covid_infection_2006.csv
        cohort6: output/covid_infection_2007.csv
        cohort7: output/covid_infection_2008.csv
        cohort8: output/covid_infection_2009.csv
        cohort9: output/covid_infection_2010.csv
        cohort10: output/covid_infection_2011.csv
        cohort11: output/covid_infection_2012.csv
        cohort12: output/covid_infection_2101.csv
        cohort13: output/covid_infection_2102.csv
        cohort14: output/covid_infection_2103.csv
        cohort15: output/covid_infection_2104.csv
        cohort16: output/covid_infection_2105.csv
        cohort17: output/covid_infection_2106.csv
        cohort18: output/covid_infection_2107.csv
        cohort19: output/covid_infection_2108.csv
        cohort20: output/covid_infection_2109.csv
        cohort21: output/covid_infection_2110.csv 
        cohort22: output/covid_infection_2111.csv 
        cohort23: output/covid_infection_2112.csv 

  process_2:
    run: r:latest analysis/process_2.R
    needs: [generate_study_population_covid_admission]
    outputs:
      highly_sensitive:
        cohort1: output/case_covid_admission.csv
        cohort2: output/case_covid_ICU.csv

  process_3:
    run: r:latest analysis/process_3.R
    needs: [process_2,generate_study_population_covid_death_ons,generate_study_population_covid_death_cpns]
    outputs:
      highly_sensitive:
        cohort: output/case_covid_icu_death.csv

  check_input_icu: # check above 
    run: r:latest -e 'rmarkdown::render("analysis/check_input_icu.Rmd", knit_root_dir = "/workspace", output_dir = "output")'
    needs: [process_2]
    outputs:
      moderately_sensitive:
        html: output/check_input_icu.html

  check_input: # check above 
    run: r:latest -e 'rmarkdown::render("analysis/check_input.Rmd", knit_root_dir = "/workspace", output_dir = "output")'
    needs: [matching_3,matching_2,extract_variables_outcome_3_control,extract_variables_outcome_3_case,extract_variables_outcome_2_control,extract_variables_outcome_2_case,merge_outcome2_3,process_1,process_2,process_3,generate_study_population, generate_study_population_covid_primarycare, generate_study_population_covid_SGSS,generate_study_population_covid_admission,generate_study_population_covid_death_ons,generate_study_population_covid_death_cpns]
    outputs:
      moderately_sensitive:
        html: output/check_input.html
        
  # matching_1_1:
  #   run: python:latest python analysis/matching_outcome_1_1.py
  #   needs: [generate_study_population, process_1]
  #   outputs:
  #     moderately_sensitive: 
  #       matching_report1: output/matching_report_general_population_infection_1_1.txt
  #     highly_sensitive: 
  #       combined1: output/matched_combined_general_population_infection_1_1.csv

  matching_2:
    run: python:latest python analysis/matching_outcome_2.py
    needs: [process_2,process_1]
    outputs:
      moderately_sensitive: 
        matching_report2: output/matching_report_infection_hosp.txt
      highly_sensitive: 
        combined2: output/matched_combined_infection_hosp.csv

  matching_3:
    run: python:latest python analysis/matching_outcome_3.py
    needs: [process_3,process_2]
    outputs:
      moderately_sensitive: 
        matching_report3: output/matching_report_hosp_icu_death.txt 
      highly_sensitive: 
        combined3: output/matched_combined_hosp_icu_death.csv

  ## seperate matched by case and contorl
  pre_extract_variables_outcome_2:
    run: r:latest analysis/pre_extract_variables_outcome_2.R
    needs: [matching_2]
    outputs:
      highly_sensitive:
        cohort1: output/matched_outcome_2_case.csv
        cohort2: output/matched_outcome_2_control.csv

  pre_extract_variables_outcome_3:
    run: r:latest analysis/pre_extract_variables_outcome_3.R
    needs: [matching_3]
    outputs:
      highly_sensitive:
        cohort1: output/matched_outcome_3_case.csv
        cohort2: output/matched_outcome_3_control.csv

  extract_variables_outcome_2_case: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_2_case
    needs: [pre_extract_variables_outcome_2]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_2_case.csv
  
  extract_variables_outcome_2_control: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_2_control
    needs: [pre_extract_variables_outcome_2]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_2_control.csv

  extract_variables_outcome_3_case: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_3_case
    needs: [pre_extract_variables_outcome_3]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_3_case.csv
  
  extract_variables_outcome_3_control: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_3_control
    needs: [pre_extract_variables_outcome_3]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_3_control.csv

  merge_outcome2_3: # create variables and merge subsets
    run: r:latest analysis/merge_outcome2_3.R
    needs: [matching_3,matching_2,extract_variables_outcome_3_control,extract_variables_outcome_3_case,extract_variables_outcome_2_control,extract_variables_outcome_2_case]
    outputs:
      highly_sensitive:
       # cohort1: output/matched_outcome1.csv
        cohort2: output/matched_outcome2.csv
        cohort3: output/matched_outcome3.csv

  check_matching_2:
    run: r:latest -e 'rmarkdown::render("analysis/check_matching_2.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [merge_outcome2_3]
    outputs:
      moderately_sensitive:
        html: output/check_matching_2.html

  check_matching_3:
    run: r:latest -e 'rmarkdown::render("analysis/check_matching_3.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [merge_outcome2_3]
    outputs:
      moderately_sensitive:
        html: output/check_matching_3.html

  check_exposures_outcome2:
    run: r:latest -e 'rmarkdown::render("analysis/check_exposures_outcome2.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [merge_outcome2_3]
    outputs:
      moderately_sensitive:
        html: output/check_exposures_outcome2.html 

  check_exposures_outcome3:
    run: r:latest -e 'rmarkdown::render("analysis/check_exposures_outcome3.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [merge_outcome2_3]
    outputs:
      moderately_sensitive:
        html: output/check_exposures_outcome3.html 

  model_univariable_outcome2:
    run: r:latest -e 'rmarkdown::render("analysis/model_univariable_outcome2.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [merge_outcome2_3]
    outputs:
      moderately_sensitive:
        html: output/model_univariable_outcome2.html 
  
  model_univariable_outcome3:
    run: r:latest -e 'rmarkdown::render("analysis/model_univariable_outcome3.Rmd", knit_root_dir = "/workspace", output_dir="/workspace/output")'
    needs: [merge_outcome2_3]
    outputs:
      moderately_sensitive:
        html: output/model_univariable_outcome3.html 
  

 

# ## monthly matching datasets
#   generate_study_population_month:
#     run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population --index-date-range "2020-02-01 to 2021-12-01 by month" --skip-existing --output-dir=output/measures --output-format=csv.gz
#     outputs:
#       highly_sensitive:
#         cohort: output/measures/input_general_population_*.csv.gz
  
#   exclusion:
#     run: r:latest analysis/exclusion.R
#     needs: [process, generate_study_population_month]
#     outputs:
#        highly_sensitive:
#         cohort4: output/control_general_population_*.csv
  

  # generate_study_population_noCOVID:
  #   run: cohortextractor:latest generate_cohort --study-definition study_definition_noCOVID
  #   outputs:
  #     highly_sensitive:
  #       cohort: output/input_noCOVID.csv 
  
  # check_input_noCOVID: 
  #   run: r:latest -e 'rmarkdown::render("analysis/check_input_noCOVID.Rmd", knit_root_dir = "/workspace", output_dir = "output")'
  #   needs: [generate_study_population_noCOVID]
  #   outputs:
  #     moderately_sensitive:
  #       html: output/check_input_noCOVID.html
  
  # matching_1.2:
  #   run: python:latest python analysis/matching_outcome_1_noCOVID.py
  #   needs: [generate_study_population_noCOVID, process]
  #   outputs:
  #     moderately_sensitive: 
  #       matching_report1: output/matching_report_noCOVID_infection.txt
  #     highly_sensitive: 
  #       combined1: output/matched_combined_noCOVID_infection.csv
  
  # extract_variables_outcome_1.2:
  #   run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_noCOVID
  #   needs: [matching_1.2]
  #   outputs:
  #     highly_sensitive:
  #       cohort: output/input_outcome_1_noCOVID.csv
        


## general population
#2020
  generate_study_population_2002:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2002
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2002.csv

  generate_study_population_2003:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2003
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2003.csv
  
  generate_study_population_2004:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2004
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2004.csv
  
  generate_study_population_2005:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2005
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2005.csv

  generate_study_population_2006:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2006
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2006.csv
  
  generate_study_population_2007:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2007
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2007.csv
  
  generate_study_population_2008:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2008
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2008.csv

  generate_study_population_2009:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2009
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2009.csv
  
  generate_study_population_2010:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2010
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2010.csv
  
  generate_study_population_2011:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2011
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2011.csv

  generate_study_population_2012:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2012
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2012.csv
  
  #2021
  generate_study_population_2101:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2101
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2101.csv 

  generate_study_population_2102:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2102
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2102.csv

  generate_study_population_2103:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2103
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2103.csv
  
  generate_study_population_2104:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2104
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2104.csv
  
  generate_study_population_2105:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2105
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2105.csv

  generate_study_population_2106:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2106
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2106.csv
  
  generate_study_population_2107:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2107
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2107.csv
  
  generate_study_population_2108:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2108
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2108.csv

  generate_study_population_2109:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2109
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2109.csv
  
  generate_study_population_2110:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2110
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2110.csv
  
  generate_study_population_2111:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2111
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2111.csv

  generate_study_population_2111_ab:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2111_ab
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2111_ab.csv
  
  generate_study_population_2111_confounding:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2111_confounding
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2111_confounding.csv
  
  generate_study_population_2111_CCI:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2111_CCI
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2111_CCI.csv

  generate_study_population_2112:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_general_population_2112
    outputs:
      highly_sensitive:
        cohort: output/input_general_population_2112.csv

  
  # exclude COVID case
  process_0:
    run: r:latest analysis/process_0.R
    needs: [generate_study_population_2002,generate_study_population_2003,generate_study_population_2004,generate_study_population_2005,generate_study_population_2006,generate_study_population_2007,generate_study_population_2008,generate_study_population_2009,generate_study_population_2010,generate_study_population_2011,generate_study_population_2012,
    generate_study_population_2101,generate_study_population_2102,generate_study_population_2103,generate_study_population_2104,generate_study_population_2105,generate_study_population_2106,generate_study_population_2107,generate_study_population_2108,generate_study_population_2109,generate_study_population_2110,generate_study_population_2111,generate_study_population_2112]
    outputs:
      highly_sensitive:
        cohort1: output/general_population_2002.csv
        cohort2: output/general_population_2003.csv
        cohort3: output/general_population_2004.csv
        cohort4: output/general_population_2005.csv
        cohort5: output/general_population_2006.csv
        cohort6: output/general_population_2007.csv
        cohort7: output/general_population_2008.csv
        cohort8: output/general_population_2009.csv
        cohort9: output/general_population_2010.csv
        cohort10: output/general_population_2011.csv
        cohort11: output/general_population_2012.csv
        cohort12: output/general_population_2101.csv
        cohort13: output/general_population_2102.csv
        cohort14: output/general_population_2103.csv
        cohort15: output/general_population_2104.csv
        cohort16: output/general_population_2105.csv
        cohort17: output/general_population_2106.csv
        cohort18: output/general_population_2107.csv
        cohort19: output/general_population_2108.csv
        cohort20: output/general_population_2109.csv
        cohort21: output/general_population_2110.csv
        cohort22: output/general_population_2111.csv
        cohort23: output/general_population_2112.csv


  # matching by month

  matching_1_2002:
    run: python:latest python analysis/matching_outcome_1_2002.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2002.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2002.csv
  
  matching_1_2003:
    run: python:latest python analysis/matching_outcome_1_2003.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2003.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2003.csv

  matching_1_2004:
    run: python:latest python analysis/matching_outcome_1_2004.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2004.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2004.csv
  
  matching_1_2005:
    run: python:latest python analysis/matching_outcome_1_2005.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2005.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2005.csv

  matching_1_2006:
    run: python:latest python analysis/matching_outcome_1_2006.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2006.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2006.csv
  
  matching_1_2007:
    run: python:latest python analysis/matching_outcome_1_2007.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2007.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2007.csv

  matching_1_2008:
    run: python:latest python analysis/matching_outcome_1_2008.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2008.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2008.csv
  
  matching_1_2009:
    run: python:latest python analysis/matching_outcome_1_2009.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2009.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2009.csv

  matching_1_2010:
    run: python:latest python analysis/matching_outcome_1_2010.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2010.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2010.csv
  
  matching_1_2011:
    run: python:latest python analysis/matching_outcome_1_2011.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2011.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2011.csv

  matching_1_2012:
    run: python:latest python analysis/matching_outcome_1_2012.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2012.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2012.csv
  
  matching_1_2101:
    run: python:latest python analysis/matching_outcome_1_2101.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2101.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2101.csv

  matching_1_2102:
    run: python:latest python analysis/matching_outcome_1_2102.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2102.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2102.csv
  
  matching_1_2103:
    run: python:latest python analysis/matching_outcome_1_2103.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2103.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2103.csv

  matching_1_2104:
    run: python:latest python analysis/matching_outcome_1_2104.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2104.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2104.csv
  
  matching_1_2105:
    run: python:latest python analysis/matching_outcome_1_2105.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2105.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2105.csv

  matching_1_2106:
    run: python:latest python analysis/matching_outcome_1_2106.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2106.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2106.csv
  
  matching_1_2107:
    run: python:latest python analysis/matching_outcome_1_2107.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2107.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2107.csv

  matching_1_2108:
    run: python:latest python analysis/matching_outcome_1_2108.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2108.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2108.csv
  
  matching_1_2109:
    run: python:latest python analysis/matching_outcome_1_2109.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2109.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2109.csv
  
  matching_1_2110:
    run: python:latest python analysis/matching_outcome_1_2110.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2110.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2110.csv
  
  matching_1_2111:
    run: python:latest python analysis/matching_outcome_1_2111.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2111.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2111.csv
  
  matching_1_2112:
    run: python:latest python analysis/matching_outcome_1_2112.py
    needs: [process_0,process_1.2]
    outputs:
      moderately_sensitive: 
        matching_report1: output/matching_report_general_population_infection_2112.txt
      highly_sensitive: 
        combined1: output/matched_combined_general_population_infection_2112.csv


    ## seperate matched by case and contorl
  pre_extract_variables_outcome_1:
    run: r:latest analysis/pre_extract_variables_outcome_1.R
    needs: [matching_1_2112,matching_1_2111,matching_1_2110,matching_1_2109,matching_1_2108,matching_1_2107,matching_1_2106,matching_1_2105,matching_1_2104,matching_1_2103,matching_1_2102,matching_1_2101,
    matching_1_2012,matching_1_2011,matching_1_2010,matching_1_2009,matching_1_2008,matching_1_2007,matching_1_2006,matching_1_2005,matching_1_2004,matching_1_2003]
    outputs:
      highly_sensitive:
        cohort1: output/matched_outcome_1_case.csv
        #cohort1.1: output/matched_outcome_1_case_2003.csv
        cohort1.2: output/matched_outcome_1_control_2003.csv
        #cohort2.1: output/matched_outcome_1_case_2004.csv
        cohort2.2: output/matched_outcome_1_control_2004.csv
        #cohort3.1: output/matched_outcome_1_case_2005.csv
        cohort3.2: output/matched_outcome_1_control_2005.csv
        #cohort4.1: output/matched_outcome_1_case_2006.csv
        cohort4.2: output/matched_outcome_1_control_2006.csv
        #cohort5.1: output/matched_outcome_1_case_2007.csv
        cohort5.2: output/matched_outcome_1_control_2007.csv
        #cohort6.1: output/matched_outcome_1_case_2008.csv
        cohort6.2: output/matched_outcome_1_control_2008.csv  
        #cohort7.1: output/matched_outcome_1_case_2009.csv
        cohort7.2: output/matched_outcome_1_control_2009.csv  
        #cohort8.1: output/matched_outcome_1_case_2010.csv
        cohort8.2: output/matched_outcome_1_control_2010.csv 
        #cohort9.1: output/matched_outcome_1_case_2011.csv
        cohort9.2: output/matched_outcome_1_control_2011.csv   
        #cohort10.1: output/matched_outcome_1_case_2012.csv
        cohort10.2: output/matched_outcome_1_control_2012.csv   
        #cohort11.1: output/matched_outcome_1_case_2101.csv
        cohort11.2: output/matched_outcome_1_control_2101.csv
        #cohort12.1: output/matched_outcome_1_case_2102.csv
        cohort12.2: output/matched_outcome_1_control_2102.csv  
        #cohort13.1: output/matched_outcome_1_case_2103.csv
        cohort13.2: output/matched_outcome_1_control_2103.csv
        #cohort14.1: output/matched_outcome_1_case_2104.csv
        cohort14.2: output/matched_outcome_1_control_2104.csv
        #cohort15.1: output/matched_outcome_1_case_2105.csv
        cohort15.2: output/matched_outcome_1_control_2105.csv
        #cohort16.1: output/matched_outcome_1_case_2106.csv
        cohort16.2: output/matched_outcome_1_control_2106.csv
        #cohort17.1: output/matched_outcome_1_case_2107.csv
        cohort17.2: output/matched_outcome_1_control_2107.csv
        #cohort18.1: output/matched_outcome_1_case_2108.csv
        cohort18.2: output/matched_outcome_1_control_2108.csv  
        #cohort19.1: output/matched_outcome_1_case_2109.csv
        cohort19.2: output/matched_outcome_1_control_2109.csv  
        #cohort20.1: output/matched_outcome_1_case_2110.csv
        cohort20.2: output/matched_outcome_1_control_2110.csv 
        #cohort21.1: output/matched_outcome_1_case_2111.csv
        cohort21.2: output/matched_outcome_1_control_2111.csv   
        #cohort22.1: output/matched_outcome_1_case_2112.csv
        cohort22.2: output/matched_outcome_1_control_2112.csv       

# 2020
  extract_variables_outcome_1_case: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_case
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_case.csv
  
  extract_variables_outcome_1_control_2003: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2003
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2003.csv

  extract_variables_outcome_1_control_2004: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2004
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2004.csv
        
  extract_variables_outcome_1_control_2005: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2005
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2005.csv

  extract_variables_outcome_1_control_2006: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2006
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2006.csv

  extract_variables_outcome_1_control_2007: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2007
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2007.csv

  extract_variables_outcome_1_control_2008: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2008
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2008.csv

  extract_variables_outcome_1_control_2009: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2009
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2009.csv

  extract_variables_outcome_1_control_2010: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2010
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2010.csv
  
  extract_variables_outcome_1_control_2011: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2011
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2011.csv
  
  extract_variables_outcome_1_control_2012: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2012
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2012.csv
 # 2021            
  extract_variables_outcome_1_control_2101: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2101
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2101.csv

  extract_variables_outcome_1_control_2102: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2102
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2102.csv 

  extract_variables_outcome_1_control_2103: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2103
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2103.csv

  extract_variables_outcome_1_control_2104: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2104
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2104.
        
  extract_variables_outcome_1_control_2105: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2105
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2105.csv

  extract_variables_outcome_1_control_2106: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2106
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2106.csv

  extract_variables_outcome_1_control_2107: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2107
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2107.csv

  extract_variables_outcome_1_control_2108: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2108
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2108.csv

  extract_variables_outcome_1_control_2109: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2109
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2109.csv

  extract_variables_outcome_1_control_2110: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2110
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2110.csv
  
  extract_variables_outcome_1_control_2111: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2111
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2111.csv
  
  extract_variables_outcome_1_control_2112: 
    run: cohortextractor:latest generate_cohort --study-definition study_definition_outcome_1_control_2112
    needs: [pre_extract_variables_outcome_1]
    outputs:
      highly_sensitive:
        cohort: output/input_outcome_1_control_2112.csv