Installation (development version)#

This is the installation guide for the development version of the Enerzyme package. You can get the package from the GitHub repository:

git clone https://github.com/Benzoin96485/Enerzyme.git
cd Enerzyme

Checkout the development branch:

git checkout devel

We recommend creating a conda environment with a yaml file requirements.yaml for the dependencies:

conda env create -f requirements.yaml

which includes the following contents:

name: enerzyme
channels:
- conda-forge
- defaults
dependencies:
    # Base depends
- python
- pip
    # Pip-only installs
- pip:
    - numpy             # for numerical computing
    - h5py              # for HDF5 file support
    - tqdm              # for progress bars
    - ase               # for simmulation environment
    - joblib            # for checkpointing
    - addict            # for passing parameters to submodules
    - pandas            # for saving prediction results
    - torch             # for deep neural networks
    - scikit-learn      # for data splitting
    - transformers      # for training schedulers
    - torch-ema         # for EMA training
    - pyyaml            # for parsing configuration files
    - torch_geometric   # for graph neural networks
    - rdkit             # for chemoinformatics
    - e3nn              # for equivariant neural networks
    - lightning         # for multi-GPU training

Then activate the environment:

conda activate enerzyme

and go to https://data.pyg.org/whl/ and find the latest wheel file for torch-scatter that matches your PyTorch version, CUDA version, Python version, and platform. For example, if you are using PyTorch 2.5.1, CUDA 12.4, Python 3.12, and Linux x86_64 platform, you can click on the torch-2.5.1+cu124 link and find the link to the wheel file torch_scatter-2.1.2+pt25cu124-cp312-cp312-linux_x86_64.whl.

Then install the wheel file:

pip install https://data.pyg.org/whl/torch-2.5.0%2Bcu124/torch_scatter-2.1.2%2Bpt25cu124-cp312-cp312-linux_x86_64.whl

Finally, install the package in the repository root directory:

pip install -e .

Check the library installation:

python -c "import enerzyme"

and the command line interface:

enerzyme -h

This Page