bcpaff - Exploring protein-ligand binding affinity prediction with electron density-based geometric deep learning

Details are described in our paper. Please cite if you use this work.

To setup the conda environment, install Multiwfn, and download the datasets, just run the following in your $CWD:

cd bcpaff
make

(this step uses mamba, you can change it to conda by using make with_conda instead).

Structure preparation and training (remove --test_run to run on all structures; remove --cluster_options no_cluster to run via Slurm):

make data_processing

(basically running bcpaff.data_processing.data_processing)

ML model training:

make ml_experiments

To interactively visualize BCPs in Jupyter Notebook:

from bcpaff.qtaim.qtaim_viewer import QtaimViewer 
from bcpaff.qtaim.qtaim_reader import QtaimProps 

qp = QtaimProps(basepath="PATH_TO_COMPOUND_FOLDER")
v = QtaimViewer(qp)
v.show()