This project contains Jupyter Notebooks designed to identify binding sites in protein-ligand complexes using dipolar EPR distance distributions. The methodology is based on the approach described in the upcoming publication "Enhanced Binding Site Identification in Protein–Ligand Complexes with a Combined Blind Docking and Dipolar Electron Paramagnetic Resonance Approach" published in Journal of American Chemical Society and available at https://doi.org/10.1021/jacs.5c01274.

This version of the repository contains an example script of the HSA-TCPP docking (Examples/TCPP). Two notebooks describe both blind and focused docking with the comparison with experimental EPR data.

Protein-Ligand Binding Site Identification Based on Dipolar EPR Experiments

Repository Contents

This repository includes Jupyter Notebooks and helper scripts used for the binding site identification using dipolar EPR data.

Required Software

This approach relies on AutoDock-GPU for both blind and focused docking. You can find AutoDock-GPU at: https://github.com/ccsb-scripps/AutoDock-GPU

After installing AutoDock-GPU, make sure to specify the path to its executable file within the Jupyter Notebook for docking.

Required Python Packages

For Docking and Spin Label Modeling:

• ADFR Suite (https://ccsb.scripps.edu/adfr/downloads/)
• MDAnalysis
• Antechamber
• Acpype - this workflow uses modified Acpype package that allows choosing the type of atom/bond type prediction. Until this feature is implemented in the main repository you can use my modified fork at (https://github.com/Mishakolok/acpype).
• RDKit
• Numpy
• ChiLife (https://github.com/StollLab/chiLife)
• SciPy
• Scikit-learn (SkLearn)
• Meeko (https://github.com/forlilab/Meeko)
• OpenBabel

This workflow uses GAFF2 parametrization of ligands, which require Amber force field installed for your MD package. You can find AmberFF14SB and other force field parameters for GROMACS at https://github.com/intbio/gromacs_ff

Installation

This workflow is designed to run on Linux

1. Install AutoDock-GPU (either download pre-compiled binaries or build yourself) and put the path to the adgpu executable in the notebook
2. Install ADFRSuite
3. Get the following scripts from Autodock-Vina repository (https://github.com/ccsb-scripps/AutoDock-Vina) and put them in the vina_scipts directory

mapwater.py (for hydrated docking)
prepare_gpf.py
prepare_gpfzn.py (for docking with Zn)
prepare_flexreceptor.py (for flexible docking)

4. Install acpype from above-mentioned fork
5. Create conda environment with required packages using conda create or mamba create

conda create --name epr_bindsite numpy pandas matplotlib scipy scikit-learn meeko=0.6.1 ipykernel openbabel pymol-open-source mdanalysis -c conda-forge 

It is critical to use the latest Meeko version (0.6.1) due to RDKit compatibility.

6. Install additional packages

pip install chilife
pip install kneed

Citations

If you find this approach or software useful in your work, we kindly request that you cite both the original paper describing the method (https://doi.org/10.1021/jacs.5c01274, proper citation will be added when available) as well as all the software used.