Convolutional Neural Networks

Required files for repeating "Investigating the impact of convolutional neural networks through distance-weighted atomic contact features on binding affinity prediction" paper.

Steps:

1. Download PDBbind 2016 dataset from this site.
2. Use delete_excessive_files.py to delete .sdf an _pocket.pdb files from PDBbind 2016 (both refined and general sets).
3. Use generate_features.py script to generate features for your data. Output is saved in .pkl format.
4. 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.

Attribution

The graphical abstract has been designed using images from Flaticon.

Contact

Milad Rayka, milad.rayka@yahoo.com

Copyright

Copyright (c) 2023-2024, Milad Rayka