Installation

UNIQUE is currently compatible with Python 3.8 through 3.12.1. To install the latest release, run the following in a compatible environment of choice:

pip

pip install unique-uncertainty

conda

conda install -c conda-forge unique-uncertainty

mamba

mamba install -c conda-forge unique-uncertainty

Tip

To create a dedicated virtual environment for UNIQUE using conda/mamba with all the required and compatible dependencies, check out: For Developers.

For Developers

See also

If you wish to work on the codebase itself, check first how to best contribute to UNIQUE.

Warning

The following steps are recommended only for expert/power-users.

First, clone the repository and check into the project folder.

git clone https://github.com/Novartis/UNIQUE.git ./unique
cd unique

The project uses conda and/or mamba for dependencies management.

Install first the conda dependencies and the Jupyter kernel needed to run the examples:

# Install conda environment and jupyter kernel locally
make env && make jupyter-kernel
conda activate .conda/unique

Next, enable the pre-commit hooks for automatic code formatting/linting:

# Setup precommit hooks
make pre-commit

Lastly, install UNIQUE from source:

pip install -e .
# Use `pip install -e .[dev]` to also install optional dependencies

In this way, you will have access to the UNIQUE codebase and be able to make local modifications to the source code, within the ./.conda/unique local environment that contains all the required dependencies.

Additionally, if you use Jupyter Notebooks, the unique kernel will be available in the “Select kernel” menu of the JupyterLab/JupyterNotebook UI.

Finally, when using git for code versioning, the predefined pre-commit hooks will be run against the commited files for automatic formatting and syntax checks.

You can find out more about custom Jupyter kernels here and pre-commit hooks here.