Molecular Docking Pipeline

An automated pipeline for batch molecular docking using AutoDock Vina, enabling docking simulations starting from UniProt IDs for receptors and PubChem IDs for ligands.

Features

• Automated docking: Performs molecular docking using AutoDock Vina
• Automatic preparation: Downloads and prepares protein structures from AlphaFold and ligands from PubChem automatically
• Batch processing: Supports processing of multiple ligand-receptor pairs
• Meeko integration: Uses Meeko for optimal ligand and receptor preparation
• Predefined database: Includes 10 common drug-target pairs for quick testing

System Requirements

• Operating System: Linux or macOS (limited Windows support)
• Python: ==3.10
• Main dependencies:
  • AutoDock Vina
  • RDKit
  • Meeko
  • NumPy

Installation

1. Create Conda Environment

conda create -n vina python=3.10
conda activate vina
conda config --env --add channels conda-forge
conda install -c conda-forge numpy swig boost-cpp libboost sphinx sphinx_rtd_theme

2. Install Meeko

pip install git+https://github.com/forlilab/Meeko.git

3. Install Additional Dependencies

pip install rdkit vina

Project Structure

molecular-docking-pipeline/
├── autodock.py          # Main script
├── batch_docking.py        # Batch processing script
├── meeko_converter.py       # Module for Meeko conversion
├── get_pdbs.py             # Module for structure downloads
├── ligand_structures/       # Directory for ligand structures
├── receptor_structures/     # Directory for receptor structures
└── docking_results/        # Directory for docking results

Usage

Basic Execution

python autodock.py

Execution with Custom Parameters

python autodock.py --receptor P23219 --ligand 60961

Available Parameters

• --receptor: UniProt ID of target receptor (default: P23219)
• --ligand: PubChem ID of ligand (default: 2244)

Predefined Database

The pipeline includes 10 predefined drug-target pairs:

Workflow

1. Automatic download: Downloads protein structures from AlphaFold and ligands from PubChem
2. Preparation: Converts structures to required PDBQT formats using Meeko
3. Configuration: Defines search space for binding site
4. Docking: Executes molecular docking simulation
5. Results: Saves best poses and generates detailed reports

Output

The pipeline generates the following output files:

• Docked structures: docking_results_[UniProt]_[PubChem].pdbqt
• Energy report: docking_report_[UniProt]_[PubChem].txt
• Console log: Binding energies for each generated pose

Example Output

Binding energies (kcal/mol):
Pose 1: -8.245
Pose 2: -7.891
Pose 3: -7.634
Pose 4: -7.289
Pose 5: -6.957

Batch Processing

For processing multiple receptor-ligand pairs, use the batch docking script:

Pairwise Mode (1:1)

python batch_docking.py --receptors P23219 P21728 P29274 --ligands 60961 2244 12345

All vs All Mode (Cartesian)

python batch_docking.py --receptors P23219 P21728 --ligands 60961 2244 --mode all_vs_all

Batch Parameters

• --receptors: Space-separated list of UniProt IDs
• --ligands: Space-separated list of PubChem IDs
• --mode: pairwise (default) or all_vs_all
• --output: Results summary file (default: batch_results.txt)

The batch script automatically runs autodock.py for each receptor-ligand combination and generates a comprehensive results summary.

Advanced Configuration

Docking Parameters

You can modify docking parameters in the code:

# Search space parameters
docker.define_search_space(
    center_x=15.0, center_y=10.0, center_z=5.0,
    size_x=20, size_y=20, size_z=20
)

# Docking parameters
energies = docker.run_docking(exhaustiveness=16, n_poses=5)

Database Customization

To add new drug-target pairs, modify the ligands_drugs and drugs_receptor dictionaries in the main code.

Troubleshooting

Common Errors

• FileNotFoundError: Check internet connection for structure downloads
• Import Error: Ensure all dependencies are correctly installed
• Meeko Error: Verify Meeko installation from GitHub repository

Debugging

For detailed debugging, check:

• Existence of downloaded PDB files
• Correct conversion to PDBQT format
• Search space parameters

License

This project is distributed under the MIT License. See the LICENSE file for details.

Citations

If you use this pipeline in your research, please consider citing:

• AutoDock Vina
• Meeko
• AlphaFold Database
• PubChem Database

Class Documentation

AutoDockVinaRunner

Main class for handling molecular docking operations:

Methods:

• prepare_receptor(receptor_pdbqt): Loads and prepares the receptor protein
• prepare_ligand(ligand_pdbqt): Loads and prepares the ligand
• define_search_space(center_x, center_y, center_z, size_x, size_y, size_z): Defines the binding site search space
• run_docking(exhaustiveness, n_poses): Executes the docking simulation
• save_results(output_file): Saves docking results to file

Key Functions

• parse_arguments(): Parses command line arguments for receptor and ligand IDs
• main(): Main function that orchestrates the entire docking workflow
• is_jupyter_notebook(): Detects if running in Jupyter environment for proper argument handling