Introduction
In this exercise, we will perform a molecular docking simulation using AutoDock Vina to predict the binding orientation of a ligand within the binding site of a receptor protein.
Molecular docking explores possible ligand conformations and orientations within a defined region of the receptor. AutoDock Vina evaluates these binding poses using a scoring function that estimates binding affinity.
We will use the Python interface of AutoDock Vina to perform docking, examine predicted binding affinities, and save the resulting ligand conformations for subsequent structural analysis using PyMOL.
1. Download the input files
The receptor and ligand structures have already been prepared in PDBQT format.
Download the following files and place them in your Jupyter notebook working directory.
| File | Description |
|---|---|
receptor.pdbqt | Prepared protein receptor containing atomic coordinates, atom types, and partial charges |
ligand.pdbqt | Prepared ligand containing atomic coordinates, atom types, partial charges, and rotatable bond information |
Preparation of files for docking simulation
The procedure to prepare docking can be found:
Prepare docking simulation for autodock vina
2. Perform molecular docking
Execute the following Python code in a Jupyter notebook.
from vina import Vina
# Load Vina object
v = Vina(sf_name='vina')
# Set the receptor and ligand
v.set_receptor('receptor.pdbqt')
v.set_ligand_from_file('ligand.pdbqt')
# Define the docking box
v.compute_vina_maps(center=[63.051, 22.770, 15.879], box_size=[20, 20, 20])
# Perform docking
v.dock(exhaustiveness=32, n_poses=50)
# Save the results
v.write_poses('docked_ligand.pdbqt', n_poses=50, overwrite=True)
Explanation of the parameters
| Parameter | Description |
|---|---|
sf_name='vina' | Selects the AutoDock Vina scoring function |
set_receptor() | Loads the prepared receptor structure |
set_ligand_from_file() | Loads the prepared ligand structure |
center | Specifies the center of the docking search box in Å |
box_size | Specifies the dimensions of the search box in Å |
exhaustiveness=32 | Controls the thoroughness of the conformational search |
n_poses=50 | Requests up to 50 predicted binding poses |
The docking box defines the three-dimensional region in which AutoDock Vina searches for possible ligand-binding conformations. In this example, a cubic search box of 20 × 20 × 20 Å is centered at the specified coordinates.
The exhaustiveness parameter controls the computational effort devoted to searching for binding poses. Higher values generally increase computational time and may improve the thoroughness of the search.
3. Examine the docking results
After docking, the predicted binding poses can be evaluated using their estimated binding affinities.
# Retrieve predicted binding energies and RMSD values
energies = v.energies(n_poses=50)
# Display the results
print(energies)The output contains the estimated binding energy (kcal/mol) and RMSD-related values for each predicted pose.
Interpretation
- More negative binding energies indicate more favorable predicted binding according to the Vina scoring function.
- RMSD values describe the structural differences between predicted ligand poses.
- The lowest-energy pose is generally ranked as the most favorable predicted binding conformation.
Note that docking scores are approximate and should not be interpreted as experimentally measured binding affinities.
4. Save and visualize the docking results
The predicted ligand conformations are saved in:
docked_ligand.pdbqt
This file can be converted into PDB format using Open Babel for visualization in PyMOL.
!obabel docked_ligand.pdbqt -O docked_ligand.pdbOpen the receptor structure and the predicted ligand poses in PyMOL to examine their spatial relationships.
For detailed analysis, students should compare the predicted ligand-binding orientations, identify amino acid residues near the ligand, and examine possible interactions such as hydrogen bonds and hydrophobic contacts.
5. Questions for discussion
- What is the predicted binding affinity of the highest-ranked ligand pose?
- How do the predicted ligand conformations differ from one another?
- Which amino acid residues are located near the predicted ligand-binding site?
- How does changing the docking search box or exhaustiveness parameter affect the docking results?
- What are the major limitations of molecular docking in predicting protein–ligand interactions?