PoseCheck: Benchmarking Generated Poses

Paper | Documentation

What is PoseCheck?

PoseCheck is a package for analysing the quality of generated protein-ligand complexes from 3D target-conditioned generative models.

<!-- # ## Why use PoseCheck? # # # # <p align="center"> # <a href="https://www.graphein.ai/#gh-light-mode-only"> # <img src="notebooks/figs/ours.png"/> # </a> # </p> -->

Installation

pip install posecheck

Example usage

We provide a simple top level API to easily interact with the whole of the benchmark. Just define the PoseCheck object once at the top of your existing testing code and test molecules by loading them in iteratively. You can also use all the testing functions manually as well (see Docs for more info).

from posecheck import PoseCheck

# Initialize the PoseCheck object
pc = PoseCheck()

# Load a protein from a PDB file (will run reduce in the background)
pc.load_protein_from_pdb("data/examples/1a2g.pdb")

# Load ligands from an SDF file
pc.load_ligands_from_sdf("data/examples/1a2g_ligand.sdf")
# Alternatively, load RDKit molecules directly
# pc.load_ligands_from_mols([rdmol])

# Check for clashes
clashes = pc.calculate_clashes()
print(f"Number of clashes in example molecule: {clashes[0]}")

# Check for strain
strain = pc.calculate_strain_energy()
print(f"Strain energy of example molecule: {strain[0]}")

# Check for interactions
interactions = pc.calculate_interactions()
print(f"Interactions of example molecule: {interactions}")

Tips

• Reading and processing all the PDB files using reduce can take a while for a large test set. If you are running PoseCheck frequently, it might be worth pre-processing all proteins yourself using prot = posecheck.utils.loading.load_protein_from_pdb(pdb_path) and setting this directly within PoseCheck using pc.protein = prot.

Data from the paper

The data for the paper can be found at the following Zenodo link and should be placed in the data directory.

Cite

@article{harris2023benchmarking,
  title={Benchmarking Generated Poses: How Rational is Structure-based Drug Design with Generative Models?},
  author={Harris, Charles and Didi, Kieran and Jamasb, Arian R and Joshi, Chaitanya K and Mathis, Simon V and Lio, Pietro and Blundell, Tom},
  journal={arXiv preprint arXiv:2308.07413},
  year={2023}
}

Acknowledgements

PoseCheck relies on several other codebases to function. Here are the links to them:

• RDKit: A collection of cheminformatics and machine learning tools.
• ProLIF: Protein-Ligand Interaction Fingerprints generator.
• Seaborn: Statistical data visualization library.
• NumPy: The fundamental package for scientific computing with Python.
• DataMol: A minimalist and practical chemoinformatics library for python.
• Pandas: Powerful data structures for data analysis, time series, and statistics.
• Reduce: A program for adding hydrogens to a Protein DataBank (PDB) molecular structure file.

There is also the similar package PoseBusters which provides additional tests to us and is recommended if you are benchmarking protein-ligand docking models.