AutoDock-GPU Docking Pipeline

This repository contains a fully automated virtual screening pipeline using AutoDock-GPU, AutoGrid, MGLTools, and Meeko. The pipeline enables batch docking of ligands against receptor ensembles using GPU acceleration.

---

🔧 Features

• ✅ Automated receptor preparation using prepare_receptor4.py (MGLTools)
• ✅ Ligand preparation via Meeko, with fallback to prepare_ligand4.py (MGLTools)
• ✅ Grid map generation using prepare_gpf4.py + autogrid4
• ✅ Fast docking using autodock_gpu_128wi (CUDA)
• ✅ Customizable grid center and box size
• ✅ Supports batch ligand screening per target
• ✅ Compatible with PyMOL for visualizing poses

---

📁 Repository Structure

This modular structure supports scalable, resumable docking workflows, making it easy to manage thousands of ligands, protein conformations, and kinase targets.

virtual_screening/
├── input/
│   ├── kinase_list.txt
│   ├── ligands/
│   │   └── abl1/
│   │       ├── actives/
│   │       └── decoys/
│   └── receptors/
│       └── abl1/
├── preprocessed/
│   ├── ligands_pdbqt/
│   ├── receptors_pdbqt/
│   ├── aligned_ligands/
│   └── grid_centers/
├── grids/
├── docking_output/
├── results/
├── logs/
├── scripts/
│   ├── run_vs.py
│   ├── align_ligands.py
│   └── prepare_grids.py

---

⚙️ Dependencies

• Python 3.x
• AutoDock-GPU
• AutoGrid
• MGLTools 1.5.7
• RDKit
• Meeko

---

🚀 Quickstart: Virtual Screening Workflow

Follow these steps to run the full docking pipeline:

1. Prepare receptor structures Generate receptor .pdb files (e.g., from BioEmu or other sources), then:

   bash scripts/prepare_receptors.sh

2. Prepare ligand files Convert SDF ligands (actives/decoys) into .pdbqt using Meeko or MGLTools:

   bash scripts/prepare_dude_ligands.sh

3. Generate grid maps Compute grid centers (e.g., via PyMOL selection) and generate .gpf / .fld files:

   /usr/bin/python3 -m pymol -cq scripts/prepare_grids.py --kinase abl1

4. Run virtual screening Launch batch docking with AutoDock-GPU:

   python scripts/run_vs.py

Each step can be configured or modified via script arguments or by editing the corresponding files. Intermediate outputs (e.g., .pdbqt, .fld, .dlg) will be saved in the preprocessed/, grids/, and docking_output/ directories.

---

📊 Receptor & Grid Preparation

Receptor preprocessing and grid generation are handled by MGLTools and AutoGrid to ensure compatibility with AutoDock's docking engine. The pipeline uses .pdbqt, .gpf, and .fld files for docking setup.

---

🤝 Acknowledgments

This pipeline integrates open-source tools developed by:

• The Scripps Research Institute: AutoDock-GPU, AutoGrid, MGLTools
• The Forli Lab: Meeko