Skip to content

Sage-Bionetworks-Challenges/model-to-data-challenge-workflow

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

37 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Model-to-Data Workflow

This repository will serve as a template for the CWL workflow and tools required to set up a model-to-data challenge infrastructure.

For more information about the tools, refer to ChallengeWorkflowTemplates.

Workflow Steps

Step Description
set_submitter_folder_permissions Grants admin user/team download permissions to the Docker submission log.
set_admin_folder_permissions Grants admin user/team download permissions to the predictions file.
get_docker_submission Downloads the submission's Docker image.
get_docker_config Extracts the Synapse credentials and format into Docker config.
download_goldstandard Downloads the goldstandard file.
validate_docker Ensures the Docker submission image exists and <1 terabyte in size.
email_docker_validation Sends an email notification to the partipant/team of the validation results. By default, an email will only be sent if there are errors.
annotate_docker_validation_with_output Updates the submission status (VALIDATED if valid, else INVALID).
check_docker_status Checks the submission status. If the status is INVALID, halt the workflow.
run_docker Runs the Docker submission model.
upload_results Uploads the predictions file.
annotate_docker_upload_results Adds the prediction_fileid and prediction_file_version annotations to the submission.
validate Validates the predictions file.
email_validation Sends an email notification to the partipant/team of the validation results. By default, an email will only be sent if there are errors.
annotate_validation_with_output Updates the submission status (VALIDATED if valid, else INVALID).
check_status Checks the submission status. If the status is INVALID, halt the workflow.
score Scores the predictions file.
email_score Sends an email notification to the participant/team of the scoring results. By default, all scores are sent.
annotate_submission_with_output Updates the submission status (SCORED if successful, else INVALID)

Usage

Requirements

  • pip3 install cwltool
  • A Synapse account/configuration file. Learn more here
  • A Synapse Submission object ID. Learn more here

Configurations

workflow.cwl

Step Description Required? Example
set_submitter_folder_permissions Provide the admin user ID or admin team ID for principalid Yes valueFrom: "3379097"
set_admin_folder_permissions Provide the admin user ID or admin team ID for principalid Yes valueFrom: "3379097"
download_goldstandard Update synapseid to the Synapse ID of the challenge's goldstandard Yes valueFrom: "syn12345678"
email_docker_validation Set errors_only to false if an email notification about a valid submission should also be sent No default: false
run_docker Set store to false if log files should be withheld from the participants No default: false
run_docker Provide the absolute path to the data directory for input_dir; this directory will be mounted during the Docker submission run. Yes valueFrom: "/challenge_data"
email_validation Set errors_only to false if an email notification about a valid submission should also be sent No default: false
email_score Add metrics and scores to private_annotations if they are to be withheld from the participants No default: [primary_metric, primary_metric_value]

validate.cwl

Line Description Required? Example
dockerPull: python:3.8.8-slim-buster Update the base image if the validation code is not Python If code is not Python, yes dockerPull: rocker/r-base:4.0.4
entry: | [validation code] Remove the sample validation code and replace with validation code for the Challenge Yes --
  • NOTE: expected annotations to write out are submission_status and submission_errors.

score.cwl

Line Description Required? Example
dockerPull: python:3.8.8-slim-buster Update the base image if the validation code is not Python If code is not Python, yes dockerPull: rocker/r-base:4.0.4
entry: | [scoring code] Remove the sample scoring code and replace with scoring code for the Challenge Yes --
  • NOTE: expected annotations to write out are primary_metric, primary_metric_value, and submission_status. If there is a secondary (tie-breaking) metric, include the secondary_metric and secondary_metric_value annotations as well.

Example Run

cwltool workflow.cwl --submissionId 12345 \
                     --adminUploadSynId syn123 \
                     --submitterUploadSynId syn456 \
                     --workflowSynapseId syn789 \
                     --synaspeConfig ~/.synapseConfig

where:

  • submissionId: Submission ID to run this workflow on
  • adminUploadSynId: Synapse ID of Folder accessible by admin user/team
  • submitterUploadSynId: Synapse ID of Folder accessible by submitter
  • workflowSynapseId: Synapse ID of File that links to workflow archive
  • synapseConfig: filepath to .synapseConfig file

About

This repository will contain model to data challenge workflows

Resources

License

Stars

Watchers

Forks