Miniconda Environments (Python)🔗

DCAN Labs maintains lab-wide shared miniconda environments which are configured for ease of access to several of our commonly used Python tools such as Dcm2bids. Using the shared environment is in most cases preferred to individual users doing their own Python package installations and/or setting up environments in their MSI home directory.

IMPORTANT: Environments and installed packages within the lab-wide environment must have group read and write permissions open to ensure access for all group users! Visit the directory and file permissions section for info on using a umask to set your default permissions to be group-accessible. If you need to fix the permissions of an existing environment and/or package in the lab-wide environment, those can be found at /projects/standard/faird/shared/code/external/envs/miniconda3/mini3/envs/ and /projects/standard/faird/shared/code/external/envs/miniconda3/mini3/pkgs/.

Conda Usage🔗

To load the DCAN labwide miniconda3 environment, first run the following command: 

    source /projects/standard/faird/shared/code/external/envs/miniconda3/load_miniconda3.sh

  • Note: It is advised to create your own environment within this miniconda3 for whatever you need miniconda3 for. That way we are not constantly changing the versioning of packages on the base environment.

  • The base environment has its own set of commonly used packages installed in it. However, when you activate a new environment within the base environment, it will NOT inherit the packages installed in the base environment.

  • This conda environment activates on any share. You do not have be working on faird to activate this conda environment.

To list the available environments within the miniconda environment:

 conda info --envs

To activate a specific environment: 

 conda activate environment_name

To list the packages within an environment (if the environment is activated you do not need to specify the name):

 conda list -n environment_name

Creating Environments🔗

To create a new conda environment, follow these instructions:

  1. Load the miniconda3 base environment - as seen above
  2. Check to make sure your environment does not already exist by running conda info --envs
  3. Run conda create --name your_env_name
  4. Run conda activate your_env_name
  5. If you need to install with pip, run conda install pip

  6. For installs on things included with miniconda3 run conda install package_name1 package_name2

    Note: It is a list without commas. You can also install one by one.

More information about creating and using conda environments within VS code can be found on the VSCode page

You can also create an environment with a YAML file of requirements.

  1. Load the miniconda3 base environment - as seen above
  2. Check to make sure your environment does not already exist by using conda info --envs
  3. Run conda env create -f /path/to/yaml/for/build.yml

  4. Run conda activate the_env_name

    Check out the conda user documentation and cheat sheet for further information.

Available Environments🔗

There are quite a few environments that exist in the lab-wide conda environment for commonly used pipelines/processes/code etc. The CDNI Google Drive has a miniconda spreadsheet that lists all of the available environments and their purpose. If you create a new conda environment, please update this sheet!

The base environment also includes some commonly used packages that can be helpful for running code that imports special libraries. These include bids-validator, dcm2bids, dcm2niix, debugpy, h5py, matplotlib, mkdocs, nipype, nibabel, numpy, pandas, pybids, pyyaml, scipy, seaborn, and more! If you have a script that requires a special library, try activating the base environment to see if that package is in there.

For questions, suggestions, or to note any errors, post an issue on our Github.