Skip to content

Job request: 274

Organisation:
The London School of Hygiene & Tropical Medicine
Workspace:
carehomes
ID:
q4sjhbfuzvorufpv

This page shows the technical details of what happened when authorised researcher Dave Evans requested one or more actions to be run against real patient data in the project, within a secure environment.

By cross-referencing the indicated Requested Actions with the Pipeline section below, you can infer what security level various outputs were written to. Outputs marked as highly_sensitive can never be viewed directly by a researcher; they can only request that code runs against them. Outputs marked as moderately_sensitive can be viewed by an approved researcher by logging into a highly secure environment. Only outputs marked as moderately_sensitive can be requested for release to the public, via a controlled output review service.

Jobs

  • Action:
    data_setup
    Status:
    Job identifier:
    yh3447oj55ktyjne-manually-set

Pipeline

Show project.yaml
version: '3.0'

expectations:
  population_size: 1000000

actions:
  generate_cohort:
    run: cohortextractor:latest generate_cohort --study-definition study_definition 
    outputs:
      highly_sensitive:
        cohort: input.csv

  generate_cohort_coverage:
    run: cohortextractor:latest generate_cohort --study-definition study_definition_coverage
    outputs:
      highly_sensitive:
        cohort: input_coverage.csv

  calc_coverage:
    run: r:latest analysis/calculate_tpp_coverage.R input_coverage.csv data/msoa_pop.csv
    needs: [generate_cohort_coverage]
    outputs:
      moderately_sensitive:
        log: coverage_log.txt
        data: tpp_msoa_coverage.rds
#        map: map_coverage_msoa.pdf

  data_check:
    run: r:latest analysis/data_check.R input.csv tpp_msoa_coverage.rds
    needs: [generate_cohort, calc_coverage]
    outputs:
      moderately_sensitive:
        log: data_checks.txt
        figure: tpp_coverage_carehomes.png

  data_setup:
    run: r:latest analysis/data_setup.R input.csv tpp_msoa_coverage.rds
    needs: [generate_cohort, calc_coverage]
    outputs:
      moderately_sensitive:
        log: data_setup_log.txt
        data: community_prevalence.rds
      highly_sensitive:
        analysisdata: analysisdata.rds
        input_clean: input_clean.rds
        ch_linelist: ch_linelist.rds
        ch_agg_long: ch_agg_long.rds

  descriptive:
    needs: [data_setup]
    run: r:latest analysis/descriptive.R input_clean.rds ch_linelist.rds ch_agg_long.rds community_prevalence.rds
    outputs:
      moderately_sensitive:
        report: descriptive.pdf
        log: log_descriptive.txt
        data: ch_gp_permsoa.csv


  run_models_50_cutoff:
    needs: [data_setup]
    run: r:latest analysis/run_models.R analysisdata.rds community_prevalence.rds 50
    outputs:
      moderately_sensitive:
        coeffs: coeffs_bestmod_50.csv
        output: output_model_run_50.txt
        log: log_model_run_50.txt
      highly_sensitive:
        fit: fit_opt_50.rds
        data: testdata_50.rds

  validate_models_50_cutoff:
    needs: [run_models_50_cutoff]
    run: r:latest analysis/validate_models.R fit_opt_50.rds testdata_50.rds 50
    outputs:
      moderately_sensitive:
        report: test_pred_figs_50.pdf


  run_models_90_cutoff:
    needs: [data_setup]
    run: r:latest analysis/run_models.R analysisdata.rds community_prevalence.rds 90
    outputs:
      moderately_sensitive:
        coeffs: coeffs_bestmod_90.csv
        output: output_model_run_90.txt
        log: log_model_run_90.txt
      highly_sensitive:
        fit: fit_opt_90.rds
        data: testdata_90.rds

  validate_models_90_cutoff:
    needs: [run_models_90_cutoff]
    run: r:latest analysis/validate_models.R fit_opt_90.rds testdata_90.rds 90
    outputs:
      moderately_sensitive:
        report: test_pred_figs_90.pdf


  run_all:
    needs: [validate_models_90_cutoff, validate_models_50_cutoff, descriptive]
    # In order to be valid this action needs to define a run commmand and
    # some output. We don't really care what these are but the below seems to
    # do the trick.
    run: cohortextractor:latest --version
    outputs:
      moderately_sensitive:
        whatever: project.yaml

Timeline

  • Created:

  • Started:

  • Runtime:

These timestamps are generated and stored using the UTC timezone on the TPP backend.

Job information

Status
Succeeded
Backend
TPP
Workspace
carehomes
Requested by
Dave Evans
Branch
master
Force run dependencies
No
Git commit hash
f142d48
Requested actions
  • data_setup

Code comparison

Compare the code used in this Job Request