Before doing anything with GraphINVENT, you will need to configure the GraphINVENT dependencies, as the code is dependent on very specific versions of packages, like PyTorch and RDKit, to run. You can use one of two approaches for this.
The first approach to installing GraphINVENT dependencies involves using Docker. To build the GraphINVENT container using Docker, run:
apptainer build docker/graphinvent.sif docker/graphinvent.def
Update the PYTHON_PATH variable in submit.py to run the GraphINVENT container:
PYTHON_PATH = "apptainer exec docker/graphinvent.sif /opt/conda/envs/graphinvent/bin/python"
And that's it! To learn how to start training models, go to 1_introduction.
The second approach to installing all GraphINVENT dependencies uses conda for this. If you do not have conda installed, please check out the installation instructions at this link. A virtual environment can be easily created using conda by typing into the terminal:
module load Anaconda3 // you may need to load conda first, e.g., if you are creating the environment on the cluster
conda create -n graphinvent python=3.9 pytorch::pytorch torchvision torchaudio -c pytorch
The above command might be Mac specific, in which case you might need to change the syntax a bit to install PyTorch on a different OS (see the PyTorch documentation).
Then, to activate the environment:
conda activate graphinvent
To install additional packages to the virtual environment, like RDKit, tqdm, h5py, scikit-learn, matplotlib, and tensorboard, use:
conda install -n graphinvent conda-forge::rdkit conda-forge::tqdm anaconda::h5py anaconda::scikit-learn matplotlib tensorboard
To install additional packages to the virtual environment, should the need arise, use:
conda install -n graphinvent {package_name}
To learn how to start training models, go to 1_introduction.
If you are getting the following error message when you try to run the submission script:
Fatal Python error: config_get_locale_encoding: failed to get the locale encoding: nl_langinfo(CODESET) failed
try setting LANG and LC_CTYPE with a locale:
export LANG="en_US.UTF-8"
export LC_CTYPE="en_US.UTF-8"