The base recipe was generated using
neurodocker from the script
create_container_recipe.sh
.
If possible do not modify the Dockerfile directly but modify it via this script.
Creating the Dockerfile, building it and testing it can be done with:
make docker_test
The test only run examples 1, 2 and 3 from the examples
folder.
The conda-nighres.yml
specifies the version of the minimum dependencies.
If you want to freeze all your dependencies you can do so with
conda env export > conda-nighres.yml
.
Note that if you do this you will need to remove somes lines (for example that concerns nighres as it is a local pacakage.)
make clean_env_file
If you need to update the following packages, do it in the
setup.py
file:
numpy
nibabel
psutil
antspyx
Then run the following to update conda-nighres.yml
pip install .
conda env export > conda-nighres.yml
make clean_env_file
The following pacakages that are necessary for setting up the environment building nighres JAVA or for running examples can be updated directly.
pip
jcc
Nilearn
dipy
gcc_linux-64
gxx_linux-64
This can be done with:
conda update -n nighres pip jcc Nilearn dipy gcc_linux-64 gxx_linux-64
pip install .
conda env export > conda-nighres.yml
make clean_env_file
The versions of the Java dependencies for nighres are in dependencies_sha.sh
.
They contain the commit id (SHA1) from the cbstools and imcntk repo.
For example.
cbstools_sha=7a34255
imcntk_sha=ea901d8
Those values are then used by the build process (build.sh
).
To update those to the latest commit on the master branch simply use
make update_dep_shasum
.