Written and maintained by CASRAI Editorial Board
Last updated
Molecular docking is a computational method that predicts how a small molecule (a ligand) fits into a target such as a protein, and how strongly it binds. This guide covers the software side: what a docking program does, how to install the free AutoDock Vina program from its official sources, and a command-line walkthrough with the exact commands from the current Vina documentation, including how to read the output table and why scores from different scoring functions cannot be compared.
What is molecular docking, and what is it used for in research?
Molecular docking is used to predict the bound conformation and binding free energy of small molecules to a target, according to the Scripps Research description of the AutoDock project. In practice a researcher supplies a three-dimensional receptor structure and one or more ligands, defines a search box around the site of interest, and the program returns ranked ligand poses with a predicted score.
The same approach is run on many ligands in a row for virtual screening, a topic that sits within cheminformatics alongside pharmacology and bioinformatics (see also bioinformatics vs cheminformatics). Receptor coordinates typically come from a deposited experimental structure such as an entry in the Protein Data Bank, or from a predicted model such as those in the AlphaFold Protein Structure Database.
What is a molecular docking simulation?
A docking run is a search plus a score. The program samples ligand positions and conformations inside the search box, then scores each one with a scoring function. AutoDock Vina is described by Scripps as a turnkey docking program based on a simple scoring function; AutoDock4 is based on an empirical free energy force field. The result is a set of poses ranked by predicted binding affinity, not a measured binding value.
Which software is used for molecular docking?
The AutoDock suite from Scripps Research is a suite of free open-source software and includes AutoDock Vina and AutoDock4. This guide documents that suite because its installation and command-line steps are published in the official Vina documentation. AutoDock Vina is released under the Apache-2.0 licence, was created by Oleg Trott and is maintained by the Forli Lab at Scripps Research.
Install or access AutoDock Vina (official sources only)
The official installation page offers three routes. Pick one; the Python package and the command-line binary are installed separately.
| Route | How | Platforms |
|---|---|---|
| Pre-built binary | Download from the AutoDock-Vina releases page on GitHub (https://github.com/ccsb-scripps/AutoDock-Vina/releases), then check it runs with ./vina_<version>_<OS>_<arch> --help |
Per the release assets |
| pip | pip install -U numpy vina (a virtual environment is recommended) |
Linux and macOS only |
| Conda | Create an environment with conda create -n vina python=3, activate it, add the conda-forge channel, install numpy, swig, boost-cpp, libboost, sphinx and sphinx_rtd_theme, then run pip install vina |
Per the installation page |
Building from source is, in the documentation’s words, not meant for regular users. The basic tutorial runs on macOS, Linux and Windows Subsystem for Linux (WSL); it also needs the Python package Meeko for input preparation.
Step-by-step: dock one ligand with AutoDock Vina
These steps follow the “Basic docking” tutorial in the AutoDock Vina documentation, which uses a protein-ligand example with files named 1iep_*. Replace file names and box coordinates with your own.
- Prepare the receptor. With Meeko, this command also writes the search box file:
mk_prepare_receptor.py -i 1iep_receptorH.pdb -o 1iep_receptor -p -v --box_size 20 20 20 --box_center 15.190 53.903 16.917. The documented alternative from the ADFR Suite isprepare_receptor -r 1iep_receptorH.pdb -o 1iep_receptor.pdbqt, for which the receptor coordinate file must contain all hydrogen atoms. - Prepare the ligand. Start from an SDF file:
mk_prepare_ligand.py -i 1iep_ligand.sdf -o 1iep_ligand.pdbqt. The documentation advises against PDB format for small molecules because it carries no bond connectivity. - Define the search box. The box file uses the keys
center_x,center_y,center_zandsize_x,size_y,size_z. The tutorial example uses a centre of 15.190, 53.903, 16.917 and a size of 20.0 on each axis. - Run Vina with the Vina scoring function.
vina --receptor 1iep_receptor.pdbqt --ligand 1iep_ligand.pdbqt --config 1iep_receptor.box.txt --exhaustiveness=32 --out 1iep_ligand_vina_out.pdbqt - Optional: use the AutoDock4 scoring function. This route needs precomputed affinity maps (the tutorial uses autogrid4 for them):
vina --ligand 1iep_ligand.pdbqt --maps 1iep_receptor --scoring ad4 --exhaustiveness 32 --out 1iep_ligand_ad4_out.pdbqt. Adding--scoring vinardoselects the Vinardo scoring function instead. - Export poses to a common format.
mk_export.py 1iep_ligand_vina_out.pdbqt -s 1iep_ligand_vina_out.sdf
On exhaustiveness, the documentation states: “The default exhaustiveness value is 8; increasing this to 32 will give a more consistent docking result.” A higher value therefore trades run time for consistency between repeated runs.
Output and how to read it
Vina writes all poses to a PDBQT output file and prints a table to the terminal with three kinds of column: mode (the pose number), affinity (kcal/mol), and dist from best mode, which is given as two RMSD values (rmsd l.b. and rmsd u.b.). Mode 1 is the top-ranked pose.
The key caution is about comparing scores. In the tutorial example the AutoDock4 scoring function gave a best score of -14.72 kcal/mol and the Vina scoring function gave -13.23 kcal/mol for the same ligand; those numbers are for that example only. The documentation warns that energy scores given by the AutoDock and Vina force fields “are not comparable between each other.” Compare ligands only with scores from the same scoring function, and treat any score as a predicted ranking aid rather than an experimental measurement.
Troubleshooting
- ADFR receptor preparation fails. The documentation says the receptor coordinate file given to
prepare_receptormust contain all hydrogen atoms; the tutorial starts from a file named1iep_receptorH.pdb. It also advises checking the protonation state of your molecules before docking. - Ligand loses bond information. Do not use a PDB file for the small molecule; supply SDF, as the documentation advises, and convert with
mk_prepare_ligand.py. - Repeated runs give different poses. Raise
--exhaustivenessabove the default of 8; the documentation uses 32 for a more consistent result. pip install vinais unavailable on Windows. The pip route is documented for Linux and macOS only. Use WSL or the pre-built binary from the releases page.- AD4 scoring cannot find maps. The
--mapsoption needs affinity maps generated beforehand (the tutorial’s AutoDock4 route uses autogrid4).
Free alternatives within the AutoDock suite
Everything below is free and open source and is documented on the official Scripps and GitHub pages. They differ in the scoring function, not in cost.
| Option | What the official sources say | How it is selected |
|---|---|---|
| AutoDock Vina (Vina scoring) | A turnkey docking program based on a simple scoring function; Apache-2.0 licence | Default when you run vina with a receptor and ligand |
| AutoDock4 scoring | Based on an empirical free energy force field | --scoring ad4 with --maps |
| Vinardo scoring | Listed as an alternative scoring option in the Vina tutorial | --scoring vinardo |
The Vina repository also lists support for multiple ligands and a batch mode for virtual screening, macrocycle docking, hydrated docking, external AutoDock maps, and Python 3 bindings on Linux and macOS.
Frequently asked questions
What is molecular docking?
It is a computational method that predicts how a small molecule binds to a target and estimates the binding free energy. See the first section above for how it is used in research.
Which software is used for molecular docking?
AutoDock Vina and AutoDock4 from the Scripps Research AutoDock suite are free, open-source options with official documentation, covered step by step above.
What is a molecular docking simulation?
A run in which the program samples ligand poses in a defined search box and scores each one, returning ranked poses with predicted affinities in kcal/mol.
Can I compare AutoDock4 and Vina scores directly?
No. The Vina documentation states that the energy scores of the AutoDock and Vina force fields are not comparable between each other.
Sources (official documentation)
- AutoDock Vina documentation, “Basic docking” tutorial: https://autodock-vina.readthedocs.io/en/latest/docking_basic.html (checked 2026-10-29)
- AutoDock Vina documentation, “Installation”: https://autodock-vina.readthedocs.io/en/latest/installation.html (checked 2026-10-29)
- Scripps Research Center for Computational Structural Biology, AutoDock project page: https://ccsb.scripps.edu/projects/docking (checked 2026-10-29)
- AutoDock-Vina repository, ccsb-scripps on GitHub: https://github.com/ccsb-scripps/AutoDock-Vina (checked 2026-10-29)
