Skip to content

Home

Jump to bottom
Yntiglei edited this page Jan 15, 2024 · 25 revisions

Welcome to the launchcontainers wiki! logo launchcontainers

Launchcontainers

Launchcontainers is a Python-based program built to automatically launch containerlized analysis. Launchcontainers takes config.ymal, container.json, and subject-session-list.txt as inputs. Using 1 line of bash command, it will automatically send jobs to HPC clusters regarding your computing demands.

Launchcontainers is well-suited for multi-subject, multi-session datasets. It will save a lot of time if you need analysis your entire dataset with different parameters multiple times.

In future versions, we will add more functionalities so that you can launch all the MRI data analysis pipelines such as Heudiconv, fMRIprep, pRF pipelines and etc. Please keep track of this repo and if you have any questions or suggestions, don't hesitate to contact Gari: garikoitz@gmail.com and Tiger: t.lei@bcbl.eu

Workflow of Launchcontainers

workflow_lc

Launchcontainers has two modes :

  • Prepare mode:
    • Prepare folder structures that obey BIDS specifications.
    • Prepare input files under /BIDS/derivatives/$container_name/analysis-$analysis_name(through symbolic links to save spaces).
    • Backup the command link inputs under /BIDS/derivatives/$container_name/analysis-$analysis_name and under /BIDS/derivatives/$container_name/analysis-$analysis_name/sub-$sub/ses-$ses/output/log.
    • Generate job scripts for Run mode.
  • Run mode:
    • Launch the job scripts to cluster through dask.
    • Backup computing logs and errors under /BIDS/derivatives/$container_name/analysis-$analysis_name/sub-$sub/ses-$ses/output/log.

More documentation

Clone this wiki locally