This repo contains the code from DLPacker paper DLPacker: Deep Learning for Prediction of Amino Acid Side Chain Conformations in Proteins.

• Restore full-atom protein structure from backbone (packing)
• Generate structures of point mutants (assumes the backbone has not changed)
• Pack or refine parts of protein structure (e.g. after you modelled backbone of a missing loop)
• Restore partially of fully missing side chains (to be implemented)
• probably more

Input may contain any protein/protein complex/RNA/DNA/small molecules etc. Only water molecules are removed by default and MSE residues are renamed into MET, the rest will stay the same (except side chains of course).

pip install dlpacker

Or

pip install git+https://github.com/nekitmm/DLPacker

Alternatively

git clone https://github.com/nekitmm/DLPacker
cd DLPacker
pip install .

As easy as three lines of code:

from dlpacker import DLPacker
dlp = DLPacker('my_structure.pdb')
dlp.reconstruct_protein(order = 'sequence', output_filename = 'my_structure_repacked.pdb')

Input stricture might or might not contain side chains, existing side chains, if present, will be ignored.

You can find more examples with explanations in the jupyter notebook DLPacker.ipynb.

The table below shows validation RMSD (Å) for DLPacker as well as two other state of the art algorithms, SCWRL4 and Rosetta Packer (fixbb):

If you use our code in your work, please cite the DLPacker paper:

@article{misiura2022dlpacker,
title={DLPacker: deep learning for prediction of amino acid side chain conformations in proteins},
author={Misiura, Mikita and Shroff, Raghav and Thyer, Ross and Kolomeisky, Anatoly B},
journal={Proteins: Structure, Function, and Bioinformatics},
volume={90},
number={6},
pages={1278--1290},
year={2022},
publisher={Wiley Online Library}
}