Required files for repeating "Investigating the impact of convolutional neural networks through distance-weighted atomic contact features on binding affinity prediction" paper.
Steps:
- Download PDBbind 2016 dataset from this site.
- Use
delete_excessive_files.py
to delete.sdf
an_pocket.pdb
files from PDBbind 2016 (both refined and general sets). - Use
generate_features.py
script to generate features for your data. Output is saved in.pkl
format. - Finally, use
train_and_analysis.ipynb
for training and analyzing your results.
Caution: general_set_binding_data.csv
, refined_minus_core_set_binding_data.csv
, and core_set_binding_data.py
files contain binding affinity data which are used during training process.
The graphical abstract has been designed using images from Flaticon.
Milad Rayka, milad.rayka@yahoo.com
Copyright (c) 2023-2024, Milad Rayka