This repo has migrated to GitLab

bids-pre-curation

The bids pre-curation gear will help in the renaming of acquisitions and classification of data based on user-input on a bulk scale.

Workflow

This gear can run at the project, or subject level. It should initially be run at the project level to prepare the whole project for BIDS curation. After that, changes can be made to specific subjects or sessions by running at the

To run at the project level, select the "Analyses" tab (at the top, between "Information" and "Data Views") then click on the "Run Analysis Gear" button and in the drop-down menu, look for the "BIDS Apps" section and select "BIDS Pre-Curate".

Stage 1

Users will run pre-curate on their project data, this will generate csv files that will be populated with a unique list of container labels and the information we have about the files within, as well as slots for the information that we need (classification, task, etc.).

This will be known as Stage 1, in Stage 1, suggested names can be automatically populated in the new_[type]_label field by selecting the 'suggest' configuration option. These suggested names come from the allows set in the configuration.

Important notes for Stage 1:

• At this point, bids-pre-curate DOESN'T DO ANYTHING WITH THE COLUMNS TASK, MODALITY, AND RUN

  • It is important to note that these columns are also datatype specific. For example, anat files can't contain the task entity. Additionally, modality can only be a certain set of values that is again specific to the datatype, see the BIDS Specification for details on this (particularly The Bids Specification > Modality Specific Files)

  • Future versions of this gear will respect the task and modality columns, but for now the BIDS name you want needs to be populated in the new_[container]_label

      e.g. acq-highres_run-01_T1w
    

• The ignore column in the acquisitions spreadsheet is a place to ignore acquisitions from BIDS curation. If yes or true (ignoring case) are put in this column, it will append _ignore-BIDS to the end of the acquisition label which will cause the bids-curate gear to set the ignore field in the acquisition metadata.

Stage 2

The user will then download and modify this csv-file (outside of Flywheel) to provide the missing/corrected information.

The corrected csv file will then be uploaded to the project and provided as input to a run of this gear to do on-the-fly mappings to properly update metadata.

This will be known as Stage 2. In Stage 2, three CSV inputs will be provided (see Inputs section). Stage 2 can be performed without actually making any changes by selecting the dry_run configuration option. This will print what would have been changed in the gear logs without actually changing it.

Inputs

Stage 1:

NONE

Stage 2:

• 3 CSV files

  1. session_table
  2. subject_table
  3. acquisition_table

These files should be uploaded at the project level: In the main UI select the project, choose the information tab and click Upload Attachment under Attachments

Outputs

Stage 1:

• CSV files that will be populated with a unique list of container labels and the information we have about the files within, as well as slots for the information that we need (classification, task, etc. for acquisitions)

• These files will be attached to the project by the customer after the customer fills them out if providing input.

Stage 2:

NONE

Config

• dry_run (boolean): default: false, Perform a dry-run for stage 2. Instead of actually renaming the sessions, subjects, and acquisitions, log what they would be renamed to.

dry_run only applies to stage 2

• suggest (boolean): default: true, Whether or not to suggest new names by removing all spaces and special characters.

suggest only applies to stage 1

• allows (string): default: '_-', String of characters to allow in names when suggesting new names. Defaults to underscores and hyphens. Spaces can be allowed by adding a space in this config option.

allows only applies to stage 1 when suggest is checked.

Testing

Full integration testing

From the root directory of the project run ./tests/integration_tests/test.sh run full --all. This script will do the following:

1. Copy the project directory to /tmp/gear
2. Craete a docker container with the correct dependencies for testing
3. Inside this container it will:
   1. Run unit tests
   2. Remove and re-populate the project '$GROUP/$PROJECT' with dummy data in need of pre-curation
   3. Run stage 1 to export CSV
   4. Edit the CSVs to rename files trivially
   5. Run stage 2 (renaming the files and updating metadata)

After running the script, '$GROUP/$PROJECT' should be named correctly and have metadata uploaded.

Fine grained testing

The test.sh script has a number of options for running, its full usage is as follows:

   Usage: test.sh {run,shell,test}

  -h, --help    Print this message

  -------------------------------------------------
  ... run {full,pdb} [opts]
    Run with either full run or drop into PDB

  opts:
    -a, --all         Run all tasks
    -c, --clean       Clean project and populate with new data
    -o, --stage-one   Run stage 1 test
    -s, --csv         Populate csv for stage 2
    -t. --stage-two   Run stage 2 test

  -------------------------------------------------
  ... test {unit_test,stage1,stage2,all} [opts]
    Use pytest to perform integration test on stage 1, stage 2 or full.

  opts:
    -d, --with-debug  Run pytest and drop to pdb on error
    -c. --with-cov    Run pytest and print coverage report

  -------------------------------------------------
  ... shell
    Enter bash shell in container

In general, running stage2 without having first run clean will lead to an API error, since the subjects will have already been moved. The recommended workflow is to run either ./tests/integration_tests/test.sh run full -a or running in order

1. ./tests/integration_tests/test.sh run full -c
2. ./tests/integration_tests/test.sh run full -o
3. ./tests/integration_tests/test.sh run full -s
4. ./tests/integration_tests/test.sh run full -t

And similarly for testing instead of running, either test all, or run clean and then test stage 1, then run csv and then stage 2