AutoDock Vina Tutorial for Beginners: Dock Your First Ligand From Scratch (2026)

Download the COX-2 structure directly into your receptor folder:

cd ~/docking/cox2_ibuprofen/receptor
wget https://files.rcsb.org/download/5KIR.pdb

Now clean it using PyMOL — remove waters, the co-crystallized ligand, and crystallization artifacts:

# Load the raw structure
load 5KIR.pdb

# Remove water molecules
remove resn HOH

# Check what ligands are present
select hetatms, hetatm
iterate hetatms, print(resn)

# Remove the co-crystallized ligand (residue name varies — check your output above)
remove resn IMN

# Remove common crystallization artifacts
remove resn SO4+GOL+EDO+PEG

# Save the cleaned structure
save 5KIR_clean.pdb

Now use the AutoDockTools preparation script to add hydrogens, assign Gasteiger charges, and generate the PDBQT file in one command:

prepare_receptor4.py \
  -r 5KIR_clean.pdb \
  -o 5KIR_receptor.pdbqt \
  -A hydrogens \
  -U nphs_lps_waters_deleteAltB

Verify it worked — you should see your receptor PDBQT file:

ls -lh 5KIR_receptor.pdbqt
# Should show a file of several hundred KB

grep -c "^ATOM" 5KIR_receptor.pdbqt
# Should return several thousand — the number of protein atoms

grep " \? " 5KIR_receptor.pdbqt
# Should return nothing — no unknown atom types

Go to pubchem.ncbi.nlm.nih.gov and search for ibuprofen. On the compound page, click Download → SDF → 3D Conformer. Save the file as ibuprofen.sdf in your ligand/ folder.

Or download it directly from the command line using the PubChem API (compound ID for ibuprofen is 3672):

cd ~/docking/cox2_ibuprofen/ligand
wget "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/CID/3672/SDF?record_type=3d" \
  -O ibuprofen.sdf

Now convert the SDF to PDBQT format using Open Babel. This handles 3D conformer generation, charge assignment, and format conversion in one step:

obabel ibuprofen.sdf \
  -O ibuprofen.pdbqt \
  --gen3d \
  -p 7.4 \
  --partialcharge gasteiger

The flags here: --gen3d generates or refines the 3D conformation, -p 7.4 sets the protonation state at pH 7.4 (physiological), --partialcharge gasteiger assigns Gasteiger charges compatible with AutoDock’s scoring function.

Inspect the output to confirm it looks right:

cat ibuprofen.pdbqt

You should see ATOM/HETATM records with two extra columns at the end (the charge and atom type), and a TORSDOF line near the top indicating the number of rotatable bonds. For ibuprofen this should be 4.

The grid box tells Vina where to search. It needs to be centered on the binding site and large enough to accommodate the ligand with room to explore — but not so large that the search becomes unfocused and slow.

Finding the binding site center

The easiest method when a co-crystallized ligand exists (as in 5KIR) is to use the centroid of that ligand as the box center. Load both the cleaned receptor and the original PDB in PyMOL to find it:

# Load original structure to see where the ligand was
load 5KIR.pdb, original
load 5KIR_clean.pdb, receptor

# Select the co-crystallized ligand in the original
select coligand, original and resn IMN

# Get the centroid coordinates
centerofmass coligand

PyMOL will print the XYZ centroid coordinates to the command line — something like [15.234, -8.441, 22.178]. Write these down: they become your center_x, center_y, center_z in the config file.

Setting the box size

For a typical drug-like molecule, a box of 20×20×20 Å is a safe starting point. This gives the ligand enough room to explore the binding pocket and find its optimal pose. If your target has a very large or elongated binding site, increase the box dimensions accordingly.

AutoDock Vina takes all its settings from a plain text config file. Create config.txt in your working directory (~/docking/cox2_ibuprofen/):

Use your actual centroid coordinates from Step 3 in place of the example values above. Everything else can stay as written for this tutorial.

Navigate to your working directory and run Vina pointing at your config file:

cd ~/docking/cox2_ibuprofen
vina --config config.txt

Vina will print its progress to the terminal. A typical run with exhaustiveness = 8 takes 30–90 seconds on a modern laptop. Here is what the output looks like:

Example Vina output for ibuprofen docked into COX-2. Your exact scores will differ slightly due to the random seed.

The three columns are: mode (pose number, ranked best to worst), affinity (predicted binding free energy in kcal/mol — more negative is better), rmsd l.b. / u.b. (how different this pose is from the top-ranked pose, lower and upper bound estimates).

If Vina exits immediately with no output, check that your file paths in config.txt are correct and that both PDBQT files exist where specified.

Understanding the binding affinity scores

Our top score of −8.3 kcal/mol for ibuprofen in COX-2 is in the moderate-to-strong range and consistent with its known experimental affinity — a good sign our protocol is working correctly.

Visualizing poses in PyMOL

Load the receptor and all docked poses into PyMOL to inspect them visually:

# Load receptor and docked poses
load receptor/5KIR_receptor.pdbqt, receptor
load output/docked.pdbqt, poses

# Show receptor as cartoon with binding site as sticks
hide everything
show cartoon, receptor
show sticks, receptor and (byres poses expand 5)

# Show all poses as sticks, colored by pose number
show sticks, poses
spectrum count, rainbow, poses

# Zoom to the binding site
zoom poses

The output PDBQT contains all 9 poses as separate MODEL entries. PyMOL loads them as individual states. Use the arrow keys or the state slider at the bottom to cycle through poses.

What to look for when inspecting poses