🧬 Just Molecular Docking (AutoDock Vina 1.2.7 on Google Colab)

This repository provides a hands-on, beginner-friendly workflow for molecular docking using AutoDock Vina in Google Colab.

Designed especially for:

• 🎓 Undergraduate workshops
• 🧪 Beginners in computer-aided drug design (CADD)
• 💻 Users who want zero local installation

Link to 📄 View PDF for more information.

---

🖼 Workflow Overview

Pipeline:

Ligand Prep → Receptor Prep → Docking → Pose Extraction → Score Analysis → Best Pose → (Optional MD) → Visualization

This notebook walks students from raw structures to analyzed docking results.

---

🔬 What Is Molecular Docking?

Molecular docking is a computational method used to:

Predict how a small molecule (ligand) binds to a protein.

It provides:

• 📌 A binding pose (geometry)
• 📊 A binding score (predicted strength, kcal/mol)

Docking is widely used in early-stage drug discovery to:

• Rank candidate molecules
• Understand protein–ligand interactions
• Support experimental design

⚠ Important: Docking predicts binding — it does not prove it.

---

🧪 What Determines Binding?

Docking scores are based on physical interactions between protein and ligand.

Main interaction types:

• Van der Waals interactions
  Shape complementarity and optimal atomic contact.

• Electrostatic interactions
  Attraction between opposite charges.

• Hydrogen bonds
  Strong dipole interactions (optimal distance ~2.5–3.2 Å).

• π–π interactions
  Aromatic stacking (Phe, Tyr, Trp, His).

• Halogen bonds (if present)
  Directional interaction from Cl, Br, I atoms.

These interactions together determine the predicted Vina score.

---

⚙️ Step-by-Step Tutorial Structure

1️⃣ Ligand Preparation

• Load .sdf / .mol2 / .pdb
• Generate 3D conformers
• Assign charges
• Convert to PDBQT

---

2️⃣ Receptor Preparation

• Remove water and unwanted ligands
• Add hydrogens
• Convert to PDBQT format

---

3️⃣ Docking Box Setup

Define:

• center_x, center_y, center_z
• size_x, size_y, size_z

The box must cover the binding pocket.

---

4️⃣ Run AutoDock Vina

Vina:

• Samples multiple poses
• Scores each pose
• Outputs ranked binding modes

Example:

vina --receptor receptor.pdbqt \
     --ligand ligand.pdbqt \
     --center_x 10 --center_y 15 --center_z 20 \
     --size_x 20 --size_y 20 --size_z 20 \
     --out output.pdbqt

---

5️⃣ Pose Extraction

• Split output.pdbqt
• Convert poses to PDB or SDF
• Save each pose separately

---

6️⃣ Vina Score Analysis

The notebook:

• Extracts vina_score
• Extracts rmsd_lb and rmsd_ub
• Saves results as CSV
• Generates bar plots inside Colab

Score interpretation (general guideline):

Vina Score (kcal/mol)	Interpretation
< -10	Strong
-8 to -10	Moderate
-6 to -8	Weak
> -6	Very weak

Always compare scores within the same system only.

---

7️⃣ Best Pose Selection

• Choose the most negative score
• Inspect geometry
• Check hydrogen bonds and key residues

Lowest score ≠ automatically correct pose
Always inspect visually.

---

8️⃣ (Optional) Short MD Relaxation

• Run a brief OpenMM simulation
• Relax steric clashes
• Observe short-term stability

This step helps check basic structural reasonableness.

---

9️⃣ Interactive Visualization

Upload:

• rec.pdb
• Selected docked pose

To:

👉 https://proteins.plus/

Use PoseView for:

• 2D interaction diagrams
• Hydrogen bond detection
• Residue mapping

Great for:

• Reports
• Class presentations
• Publications

---

🎯 Why This Tutorial?

• ✅ No installation required (runs entirely in browser)
• ✅ Fully commented Colab notebook
• ✅ Visual plots included
• ✅ Designed for teaching
• ✅ Easy to extend later

Interested in going beyond docking?

You can explore enhanced sampling molecular dynamics (MD) simulations directly in the cloud using DFDD:

---

🚀 Further Analysis

You can:

• Upload your own receptor
• Dock multiple ligands
• Download all outputs as ZIP
• Compare poses visually
• Perform batch docking

Batch version:

---

📌 Good Practice for Beginners

Always:

• Redock the crystal ligand first
• Check RMSD (< 2 Å is ideal)
• Use a consistent grid box
• Compare within the same protocol
• Do not overinterpret the score

---

👨‍🔬 Author

Kowit Hengphasatporn
Center for Computational Sciences (CCS)
University of Tsukuba

---

Citation

If you use DFDD or pKaNET Cloud in your research, please cite:

Hengphasatporn, K.; Duan, L.; Harada, R.; Shigeta, Y.
DFDD: A Cloud-Ready Tool for Distance-Guided Fully Dynamic Docking in Host–Guest Complexation.
Journal of Chemical Information and Modeling, 2026.
DOI: https://doi.org/10.1021/acs.jcim.5c02852

---

Happy Docking 🚀