neurodocker 0.4.0

Create custom containers for neuroimaging

Neurodocker

Neurodocker is a command-line program that generates custom Dockerfiles and Singularity recipes for neuroimaging and minifies existing containers.

• Examples:
  • Examples gallery
  • Canonical examples
    • Docker
    • Singularity
  • Minimize existing Docker image
  • Example of minimizing Docker image for FreeSurfer recon-all
• Known issues

Installation

Use the Neurodocker Docker image (recommended):

$ docker run --rm kaczmarj/neurodocker:0.4.0 --help

Note: Do not use the -t/--tty flag with docker run or non-printable characters will be a part of the output (see moby/moby#8513 (comment)).

This project can also be installed with pip:

$ pip install neurodocker
$ neurodocker --help

If the pip install command above gives a permissions error, install as a non-root user:

$ pip install --user neurodocker

Note: it is not yet possible to minimize Docker containers using the Neurodocker Docker image.

Supported software

software	argument	description
AFNI	version*	latest
	method	binaries (default), source. Install pre-compiled binaries or build form source.
	install_path	Installation path. Default /opt/afni-{version}.
	install_r	If true, install R.
	install_r_pkgs	If true, install R and AFNI's R packages.
	install_python2	If true, install Python 2.
	install_python3	If true, install Python 3.
ANTs	version*	2.2.0, 2.1.0, 2.0.3, or 2.0.0. If method=source, version can be a git commit hash or branch.
	method	binaries (default), source.
	install_path	Installation path. Default /opt/ants-{version}.
	cmake_opts	If method=source, options for cmake.
	make_opts	If method=source, options for make.
Convert3D	version*	1.0.0 or nightly.
	method	binaries (default)
	install_path	Installation path. Default /opt/convert3d-{version}.
dcm2niix	version*	latest, git commit hash or branch.
	method	source (default)
	install_path	Installation path. Default /opt/dcm2niix-{version}.
	cmake_opts	If method=source, options for cmake.
	make_opts	If method=source, options for make.
FreeSurfer	version*	6.0.0-min
	method	binaries (default)
	install_path	Installation path. Default /opt/freesurfer-{version}.
	exclude_paths	Sequence of path(s) to exclude when inflating the tarball.
	license_path	Relative path to license file. If provided, this file will be copied into the Docker image. Must be within the build context.
FSL**	version*	5.0.11, 5.0.10, 5.0.9, 5.0.8
	method	binaries (default)
	install_path	Installation path. Default /opt/fsl-{version}.
Matlab Compiler Runtime	version*	2018a, 2012-17[a-b], 2010a
	method	binaries (default)
	install_path	Installation path. Default /opt/matlabmcr-{version}.
MINC	version*	1.9.15
	method	binaries (default)
	install_path	Installation path. Default /opt/minc-{version}.
Miniconda	version	latest (default), all other hosted versions.
	install_path	Installation path. Default /opt/miniconda-{version}.
	create_env	Name of conda environment to create.
	use_env	Name of previously installed environment.
	conda_install	Packages to install with conda. E.g., conda_install="python=3.6 numpy traits"
	pip_install	Packages to install with pip.
	activate	If true (default), activate this environment in container entrypoint.
MRtrix3	version*	3.0
	method	binaries (default)
	install_path	Installation path. Default /opt/mrtrix3-{version}.
NeuroDebian	os_codename*	Codename of the operating system (e.g., stretch, zesty).
	server*	Server to download NeuroDebian packages from. Choose the one closest to you. See neurodocker generate docker --help for the full list of servers.
	full	If true (default), use non-free sources. If false, use libre sources.
PETPVC	version*	1.2.2, 1.2.1, 1.2.0-b, 1.2.0-a, 1.1.0, 1.0.0
	method	binaries (default)
	install_path	Installation path. Default /opt/petpvc-{version}.
SPM12	version*	r7219, r6914, r6685, r6472, r6225
	install_path	Installation path. Default /opt/spm12-{version}.
		Note: Matlab Compiler Runtime is installed when SPM12 is installed.

* required argument.
** FSL is non-free. If you are considering commercial use of FSL, please consult the relevant license.

Generate Dockerfile

usage: neurodocker generate docker [-h] [-b BASE] [-p {apt,yum}]
                                   [--add-to-entrypoint ADD_TO_ENTRYPOINT]
                                   [--copy COPY [COPY ...]]
                                   [--install INSTALL [INSTALL ...]]
                                   [--entrypoint ENTRYPOINT]
                                   [-e ENV [ENV ...]] [-r RUN]
                                   [--run-bash RUN_BASH] [-u USER]
                                   [-w WORKDIR] [-f FILE] [-o OUTPUT]
                                   [--no-print] [--afni  [...]]
                                   [--ants  [...]] [--convert3d  [...]]
                                   [--dcm2niix  [...]] [--freesurfer  [...]]
                                   [--fsl  [...]] [--matlabmcr  [...]]
                                   [--minc  [...]] [--miniconda  [...]]
                                   [--mrtrix3 [[...]]] [--neurodebian  [...]]
                                   [--petpvc  [...]] [--spm12  [...]]
                                   [--add ADD [ADD ...]] [--arg ARG [ARG ...]]
                                   [--cmd CMD [CMD ...]]
                                   [--expose EXPOSE [EXPOSE ...]]
                                   [--label LABEL [LABEL ...]]
                                   [--volume VOLUME [VOLUME ...]]

optional arguments:
  -h, --help            show this help message and exit
  -b BASE, --base BASE  Base Docker image. Eg, ubuntu:17.04
  -p {apt,yum}, --pkg-manager {apt,yum}
                        Linux package manager.
  --add-to-entrypoint ADD_TO_ENTRYPOINT
                        Add a command to the file /neurodocker/startup.sh,
                        which is the container's default entrypoint.
  --copy COPY [COPY ...]
                        Copy files into container. Use format <src>... <dest>
  --install INSTALL [INSTALL ...]
                        Install system packages with apt-get or yum, depending
                        on the package manager specified.
  --entrypoint ENTRYPOINT
                        Set the container's entrypoint (Docker) / append to
                        runscript (Singularity)
  -e ENV [ENV ...], --env ENV [ENV ...]
                        Set environment variable(s). Use the format KEY=VALUE
  -r RUN, --run RUN     Run a command when building container
  --run-bash RUN_BASH   Run a command in bash
  -u USER, --user USER  Switch current user (creates user if necessary)
  -w WORKDIR, --workdir WORKDIR
                        Set working directory
  -f FILE, --file FILE  Generate file from JSON. Overrides other `generate`
                        arguments
  -o OUTPUT, --output OUTPUT
                        If specified, save Dockerfile to file with this name.
  --no-print            Do not print the generated file
  --add ADD [ADD ...]   Dockerfile ADD instruction. Use format <src>... <dest>
  --arg ARG [ARG ...]   Dockerfile ARG instruction. Use format
                        KEY[=DEFAULT_VALUE] ...
  --cmd CMD [CMD ...]   Dockerfile CMD instruction.
  --expose EXPOSE [EXPOSE ...]
                        Dockerfile EXPOSE instruction.
  --label LABEL [LABEL ...]
                        Dockerfile LABEL instruction.
  --volume VOLUME [VOLUME ...]
                        Dockerfile VOLUME instruction.

Generate Singularity recipe

usage: neurodocker generate singularity [-h] [-b BASE] [-p {apt,yum}]
                                        [--add-to-entrypoint ADD_TO_ENTRYPOINT]
                                        [--copy COPY [COPY ...]]
                                        [--install INSTALL [INSTALL ...]]
                                        [--entrypoint ENTRYPOINT]
                                        [-e ENV [ENV ...]] [-r RUN]
                                        [--run-bash RUN_BASH] [-u USER]
                                        [-w WORKDIR] [-f FILE] [-o OUTPUT]
                                        [--no-print] [--afni  [...]]
                                        [--ants  [...]] [--convert3d  [...]]
                                        [--dcm2niix  [...]]
                                        [--freesurfer  [...]] [--fsl  [...]]
                                        [--matlabmcr  [...]] [--minc  [...]]
                                        [--miniconda  [...]]
                                        [--mrtrix3 [[...]]]
                                        [--neurodebian  [...]]
                                        [--petpvc  [...]] [--spm12  [...]]

optional arguments:
  -h, --help            show this help message and exit
  -b BASE, --base BASE  Base Docker image. Eg, ubuntu:17.04
  -p {apt,yum}, --pkg-manager {apt,yum}
                        Linux package manager.
  --add-to-entrypoint ADD_TO_ENTRYPOINT
                        Add a command to the file /neurodocker/startup.sh,
                        which is the container's default entrypoint.
  --copy COPY [COPY ...]
                        Copy files into container. Use format <src>... <dest>
  --install INSTALL [INSTALL ...]
                        Install system packages with apt-get or yum, depending
                        on the package manager specified.
  --entrypoint ENTRYPOINT
                        Set the container's entrypoint (Docker) / append to
                        runscript (Singularity)
  -e ENV [ENV ...], --env ENV [ENV ...]
                        Set environment variable(s). Use the format KEY=VALUE
  -r RUN, --run RUN     Run a command when building container
  --run-bash RUN_BASH   Run a command in bash
  -u USER, --user USER  Switch current user (creates user if necessary)
  -w WORKDIR, --workdir WORKDIR
                        Set working directory
  -f FILE, --file FILE  Generate file from JSON. Overrides other `generate`
                        arguments
  -o OUTPUT, --output OUTPUT
                        If specified, save Dockerfile to file with this name.
  --no-print            Do not print the generated file

Examples

Please see the examples directory.

Canonical examples

The canonical examples install ANTs version 2.2.0 on Debian 9 (Stretch).

Note: Do not use the -t/--tty flag with docker run or non-printable characters will be a part of the output (see moby/moby#8513 (comment)).

Docker

$ docker run --rm kaczmarj/neurodocker:0.4.0 generate docker \
    --base debian:stretch --pkg-manager apt --ants version=2.2.0

# Build image by piping Dockerfile to `docker build`
$ docker run --rm kaczmarj/neurodocker:0.4.0 generate docker \
    --base debian:stretch --pkg-manager apt --ants version=2.2.0 | docker build -

Singularity

Install ANTs on Debian 9 (Stretch).

$ docker run --rm kaczmarj/neurodocker:0.4.0 generate singularity \
    --base debian:stretch --pkg-manager apt --ants version=2.2.0

Minimize existing Docker image

Neurodocker must be pip installed for container minimization.

In the following example, a Docker image is built with ANTs version 2.2.0 and a functional scan. The image is minified for running antsMotionCorr. The original ANTs Docker image is 1.85 GB, and the "minified" image is 365 MB.

# Create a Docker image with ANTs, and download a functional scan.
$ download_cmd="curl -sSL -o /home/func.nii.gz http://psydata.ovgu.de/studyforrest/phase2/sub-01/ses-movie/func/sub-01_ses-movie_task-movie_run-1_bold.nii.gz"
$ neurodocker generate docker -b centos:7 -p yum --ants version=2.2.0 --run="$download_cmd" | docker build -t ants:2.2.0 -

# Run the container in the background.
# The option --security-opt=seccomp:unconfined is important. Without this,
# the trace will not be able to run in the container.
$ docker run --rm -itd --name ants-container --security-opt=seccomp:unconfined ants:2.2.0

# Output a ReproZip pack file in the current directory with the files
# necessary to run antsMotionCorr.
$ cmd="antsMotionCorr -d 3 -a /home/func.nii.gz -o /home/func_avg.nii.gz"
$ neurodocker reprozip trace ants-container "$cmd"
# Create a Docker container with the contents of ReproZip's trace.
$ reprounzip docker setup neurodocker-reprozip.rpz test

Known issues

• Using the -t/--tty option in docker run produces non-printable characters in the generated Dockerfile or Singularity recipe (see moby/moby#8513 (comment)).
  • Solution: do not use the -t/--tty flag, unless using the container interactively.
• Attempting to rebuild into an existing Singularity image may raise an error.
  • Solution: remove the existing image or build a new image file.