Install

To get started with the pipeline, please install the following requirements:

• Singularity BIDS apps (HeuDiConv, MRIQC, fMRIPrep, xcpEngine, QSIprep)

• Singularity image of Matlab R2019a (how to: https://github.com/mathworks-ref-arch/matlab-dockerfile)

• Singularity image of Ubuntu with JQ installed Singularity image of Python3 (Based on Docker python/3.9.0)

• Singularity image of Docker HTML to PDF (https://github.com/pinkeen/docker-html-to-pdf)

• Brain Connectivity Toolbox for Matlab (https://sites.google.com/site/bctnet/Home/functions/BCT.zip?attredirects=0)

• xcpEngine dsn files (https://github.com/PennBBL/xcpEngine/tree/master/designs)

• ASHS (https://sites.google.com/site/hipposubfields/)

• bidsphysio (https://github.com/cbinyu/bidsphysio)

You will need to install git, to clone the build recipe files from this repository. Please cite the images used and retrieve all relevant licenses (you will need your own Freesurfer license for some containers).

This following commands can be used to build these required images for the pipeline. Run on a machine which has sufficient permissions for Singularity and Docker building!

# Generates Singularity images used in pipeline
mkdir ./singularity_images
chmod 777 -R ./singularity_images
cd ./singularity_images

singularity build mriqc-0.16.1.sif docker://poldracklab/mriqc:0.16.1
singularity build heudiconv-0.9.sif docker://nipy/heudiconv:0.9.0
singularity build fmriprep-v21.0.1.sif docker://nipreps/fmriprep:21.0.1
singularity build xcpengine-1.2.4.sif docker://pennbbl/xcpengine:1.2.4
singularity build qsiprep-v0.15.1.sif docker://pennbbl/qsiprep:0.15.1

# See README.md for more information on
# provided def files for ubuntu-jq, python3
SINGULARITY_NOHTTPS=1 singularity build ubuntu-jqjo.sif jqjo.def
SINGULARITY_NOHTTPS=1 singularity build python3.sif defpy3
SINGULARITY_NOHTTPS=1 singularity build ashs-1.0.0.sif ashsdef
SINGULARITY_NOHTTPS=1 singularity build pylearn.sif pylearn.def

#Start Docker registry for localhost
docker run -d -p 5000:5000 --restart=always --name registry registry:2
# Follow directions to build Docker images for the following:
# https://github.com/pinkeen/docker-html-to-pdf
git clone https://github.com/pinkeen/docker-html-to-pdf.git
# You will want to set the port used in html2pdf to an unused port on the HPC
cd ./docker-html-to-pdf
docker build -t html2pdf:79.1 -t localhost:5000/html2pdf:79.1 .
docker push localhost:5000/html2pdf:79.1
cd ../
SINGULARITY_NOHTTPS=1 singularity build html2pdf.sif docker://localhost:5000/html2pdf:79.1

#build jq and jo image for new project_doc.sh
cd ./ubuntu-jqjo
docker build -t localhost:5000/ubuntu-jqjo:0.2 .
docker push localhost:5000/ubuntu-jqjo:0.2
cd ../
SINGULARITY_NOHTTPS=1 singularity build ubuntu-jqjo-v0.2.sif docker://localhost:5000/ubuntu-jqjo:0.2

# Follow directions to build Docker images for the following:
git clone https://github.com/cbinyu/bidsphysio.git
cd ./bidsphysio
docker build -t bidsphysio:latest -t localhost:5000/bidsphysio:latest .
docker push localhost:5000/bidsphysio:latest
cd ../
SINGULARITY_NOHTTPS=1 singularity build bidsphysio.sif docker://localhost:5000/bidsphysio:latest

## Prerequisites
#The following examples use the CUDA 10.2 toolkit and runtime (loaded via module or native install)
git clone https://github.com/mrfil/scfsl.git
cd ./scfsl
docker build -t scfsl_gpu:0.3.2 -t localhost:5000/scfsl_gpu:0.3.2 .
cd ../
docker push localhost:5000/scfsl_gpu:0.3.2
SINGULARITY_NOHTTPS=1 singularity build scfsl_gpu-v0.3.2.sif docker://localhost:5000/scfsl_gpu:0.3.2

# https://github.com/mathworks-ref-arch/matlab-dockerfile
git clone https://github.com/mathworks-ref-arch/matlab-dockerfile.git
mkdir ./matlab-dockerfile/matlab-install
# place install files and modified license in correct directories per guide in repo
cd ./matlab-dockerfile
docker build -f Dockerfile.R2019a -t matlab:r2019a -t localhost:5000/matlab:r2019a --build-arg MATLAB_RELEASE=R2019a .
cd ../
docker push localhost:5000/matlab:r2019b
SINGULARITY_NOHTTPS=1 singularity build matlab-R2019a.sif docker://localhost:5000/matlab:r2019a

The process for creating the MATLAB container has changed! You can build more recent MATLAB containers using the official Docker Hub images from MathWorks. You will still need to provide license information as denoted in the repository listed above.