Docker Torque

Introduction

This image runs the Torque scheduler and a single worker on a Ubuntu host. One user is provided for submission of jobs. --> This is from https://github.com/neilav/docker-torque

I have included Andre Marquand work for nispat --> https://github.com/amarquand/nispat

Build

docker build -t torque .

Run the docker with a link to you machine where the data is placed (-v argument). The "data" folder must contain files: covariates_allpatients.txt, covariates_HC.txt, features_allpatients.txt, features_HC.txt

docker run -v /path/to/the/data/dir:/mnt/data -h master --privileged -it torque bash

Once within the docker

change to user "batchuser"

su batchuser

Run the script "run_normative_parallel_test.py" pointing to your data (mounted) with "processing_dir" argument specifiying testrespfile_path and testcovfile_path (without CV)

/opt/conda/bin/python nispat/test_normative_modeling/run_normative_parallel_test.py --processing_dir /mnt/data/

Run the script "run_normative_parallel_test.py" pointing to your data (mounted) with "processing_dir" argument and CV specifying the number of folds

/opt/conda/bin/python nispat/test_normative_modeling/run_normative_parallel_test.py --processing_dir /mnt/data/ --cv_folds #folds

Include the parameter --cpu_cores to specify the number of cores to use. By default the script employ all the cores in the machines - 2

/opt/conda/bin/python nispat/test_normative_modeling/run_normative_parallel_test.py --processing_dir /mnt/data/ --cv_folds #folds --cpu_cores #n_cores

Include duration and memory parameters

/opt/conda/bin/python nispat/test_normative_modeling/run_normative_parallel_test.py --processing_dir /mnt/data/ --cv_folds #folds --cpu_cores #n_cores --memory '4gb' --duration '01:00:00'