Installation

Please complete the following steps to install BONAFIDE and its dependencies.

BONAFIDE and its Python dependencies

  1. Clone the GitHub repository and change to its root directory.

$ git clone https://github.com/MolecularAI/atom-bond-featurizer.git
$ cd atom-bond-featurizer

  1. Create a new conda environment and activate it.

$ conda env create -n bonafide_env -f bonafide_env.yml python=3.12
$ conda activate bonafide_env

Note

bonafide_env.yml provides the necessary dependencies for calculating features with BONAFIDE, whereas bonafide_env_dev.yml contains additional dependencies for development purposes (e.g., testing, documentation building).

  1. Install the kallisto package (in case its usage is intended). This is done separately due to version conflicts between kallisto dependencies and other BONAFIDE dependencies.

$ pip install kallisto --no-deps

  1. Install BONAFIDE and run the post_install_setup.py script to complete the installation.

$ pwd
.../atom-bond-featurizer
$ pip install .
$ python post_install_setup.py

ALFABET

The environment for the machine learning model ALFABET is not compatible with BONAFIDE’s environment. It must therefore be installed in a separate environment if the model is intended to be used through BONAFIDE.

$ conda create -n alfabet_env -c conda-forge python=3.7 rdkit
$ conda activate alfabet_env
$ pip install tensorflow
$ pip install alfabet

To allow BONAFIDE to discover the environment in which ALFABET is installed, go through the following steps.

  1. Create an instance of BONAFIDE’s AtomBondFeaturizer class and inspect ALFABET’s configuration settings.

>>> from bonafide import AtomBondFeaturizer
>>> f = AtomBondFeaturizer()
>>> f.print_options("alfabet")
Default configuration settings at:
.../src/bonafide/_feature_config.toml

alfabet
   python_interpreter_path: ~/miniconda3/envs/alfabet_env/bin/python

  1. Go to the _feature_config.toml file and change the value of python_interpreter_path to point to the Python interpreter of alfabet_env. The path to the file is shown at the top of the output from the print_options() method. The path to the Python interpreter can be found out with the following commands.

$ conda activate alfabet_env
$ which python
.../alfabet_env/bin/python

Multiwfn

Multiwfn is not a Python dependency and therefore must be installed separately. Go through the following steps to install the program (if not already installed). Additional information on the installation can be found in section 2.1 of the Multiwfn manual. BONAFIDE depends on the version 3.8 of Multiwfn.

  1. Download the Multiwfn_3.8_bin_Linux_noGUI.zip file from this webpage.

  2. Unzip the folder and place it in an appropriate directory which can be accessed by BONAFIDE.

  3. Ensure that the downloaded Multiwfn_noGUI file has executable permission. If not, this can be added by changing the current working directory to where Multiwfn was saved followed by executing the following command.

$ chmod +x Multiwfn_noGUI

  1. Add the following lines to your .bashrc file. Replace <path to parent directory> with the correct path.

export Multiwfnpath=<path to parent directory>/Multiwfn_3.8_bin_Linux_noGUI
export PATH=$PATH:<path to parent directory>/Multiwfn_3.8_bin_Linux_noGUI

Finalization and checking

At any stage of the installation process, the check_bonafide_installation.py script can be executed to inspect the status of the installation. This script checks if all dependencies are installed correctly and whether BONAFIDE can access them. Ensure that BONAFIDE’s conda environment is activated before running the script.

$ conda activate bonafide_env
$ python check_bonafide_installation.py
...

If the output of this script ends with ==> Hurray, all dependencies are installed correctly!, the installation process was completed successfully.