Launchcontainers is a Python-based tool for automatically launching parallel computing tasks on HPC or local clusters. It was designed to:
- Prepare folder structures and input files automatically for Neuroimaging pipelines
- Backup the input configs for data provenance
- Deploy jobs in local HPC, SGE, or SLURM in parallel
Currently, launchcontainers works along with anatROIs, RTP-preproc, and RTP2-pipeline.
To use the newest version, please pip install launchcontainers==0.4.3
- Update to
0.4.3. Add feature QC rtp2pipelines, it was integrated in the codelaunchcontainers/launchcontainers /quality_control/, you can use it by doing python qc_rtp2pipeline_output.py analysis_dir, it will read if all the tracts is finished for you - Update to
0.4.2. Refactor the launchcontainer command-line interaction to make it more user-friendly- For prepare mode, the user will do:
lc --log-dir path/to/log/dir prepare -lcc path/to/lc_yaml -ssl path/to/subseslist -cc path/to/cc- The prepare mode will create symlink and prepare analysis folder structure. it will output analysis_dir in the commandline for run mode
- For run mode, the user will do:
lc --log-dir path/to/log/dir run -w path/to/analysis_dir --run_lc- The run mode will do a independent check on analysis dir to see if the configs there is correct
- Then it will summarize the config settings and folder structure (using the cli tree command) to the user and ask for user input
- Once user type y or yes, the program will launch
- Several helper functions also implemented in the cli
lc:lc --copy_configs -o path/to/working/diryou can type this to copy all the example configs to the working directorylc --create_bids -csc path/to/csc/yaml -ssl path/to/subseslistyou can use this to create a fake BIDS folder for testing
- for more info, pip install lc and type
lc -h
- For prepare mode, the user will do:
- The update
0.3.5will be capable work with heudiconv, Presurfer and NORDIC_raw, there will be a new derivatives folder called Processed_nifti, it will stored the processed .nii.gz by NORDIC_raw and Presurfer 0.3.5: The add_intended_for function from heudiconv will be used here to edit the fmap _epi.json- Add requests into pyproject.toml, remove version limit to common package such as nibabel and numpy
- Changed rtp/rtp2-preproc multishell option to separateed_shell_files
- Edited lc_config.yaml comment about dask_worker options
- Fixed error message by dask progress (0.3.18)
- launchcontainers --copy_configs "~/path/to/working_directory" will copy the corresponding config files to your specified directory!
- We updated the lc_config.yaml for RTP2-pipelines, please have a look!
check the How to use for more information