🇭🇺 Magyar

AutoDock Vina

AutoDock Vina is a widely used molecular docking program designed to model the interactions between small molecules (ligands) and macromolecules (receptors). The example presented in this documentation demonstrates the docking of imatinib (Gleevec) into the c-Abl tyrosine kinase protein, an important drug target in the treatment of chronic myeloid leukemia.

Required Software

• AutoDock Vina 1.2.7

• Meeko

• Python 3

• AutoGrid4

• ChimeraX

Workflow Overview

1. Receptor preparation

2. Ligand preparation

3. Affinity map generation (optional)

4. Running the docking calculation

Receptor Preparation

The selected structure can be downloaded from

https://www.rcsb.org/3d-view/1IEP.

However, the downloaded structure contains two chains and two ligand molecules, among other components, which makes it unsuitable as a starting structure for our purposes. We require a receptor structure containing hydrogen atoms.

To prepare the input .pdb file, we use ChimeraX. Open the downloaded 1IEP.pdb structure. The receptor can then be prepared using the following commands in the ChimeraX command line.

First, delete chain B:

delete #1/B

Next, remove everything from chain A that is not part of the protein:

delete ~protein

Then add hydrogen atoms to the resulting receptor:

addh

The resulting structure is now suitable for docking.

The following command can be used to save the prepared receptor (alternatively, the graphical interface can be used):

save 1iep_receptorH.pdb

The file 1iep_receptorH.pdb will serve as the receptor input structure.

With this file prepared, the receptor can be converted into the format required for docking using the following command:

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

This command generates the following files:

• 1iep_receptor.pdbqt

• 1iep_receptor.box.pdb: a file that can be visualized in ChimeraX to display the docking search space

• 1iep_receptor.box.txt: a file containing the search space parameters that can be used as part of the docking configuration

Ligand Preparation

The ligand input file in .sdf format can be downloaded directly from the Small Molecules section of the

https://www.rcsb.org/structure/1IEP

entry. There, the ligand is listed under the name STI, and the corresponding structure for each chain can be downloaded directly in the desired format.

Hydrogen atoms must be added to the available 1iep_G_STI.sdf ligand before docking. This can be accomplished in ChimeraX using the same approach described previously:

addh

The resulting structure is then saved as 1iep_ligand.mol2.

The ligand preparation step can then be performed using the following command:

mk_prepare_ligand.py -i 1iep_ligand.mol2 -o 1iep_ligand.pdbqt

This generates the file:

• 1iep_ligand.pdbqt

The official documentation does not recommend the use of the PDB format for small molecules because bond connectivity information may be lost. Instead, SDF or MOL2 formats are preferred.

Generating AutoGrid4 Maps (Optional)

If the AutoDock4 scoring function is to be used, affinity maps must first be generated.

Prepare the receptor together with the GPF file:

mk_prepare_receptor.py -i 1iep_receptorH.pdb -o 1iep_receptor -p -v -g \
--box_size 20 20 20 --box_center 15.190 53.903 16.917

This generates the file

1iep_receptor.gpf,

which serves as the input file for AutoGrid4.

autogrid4 -p 1iep_receptor.gpf -l 1iep_receptor.glg

After the calculation, the following files will be created:

• 1iep_receptor.maps.fld: the main grid file containing references to all affinity maps and a description of the grid geometry

• 1iep_receptor.A.map: affinity map for aromatic/carbon-like atoms

• 1iep_receptor.C.map: interaction-energy map for aliphatic carbon atoms

• 1iep_receptor.N.map: affinity map for neutral nitrogen atoms

• 1iep_receptor.NA.map: affinity map for nitrogen atoms acting as hydrogen-bond acceptors

• 1iep_receptor.OA.map: affinity map for oxygen atoms acting as hydrogen-bond acceptors

• 1iep_receptor.HD.map: affinity map generated for hydrogen-bond donor atoms

• 1iep_receptor.d.map: desolvation map

• 1iep_receptor.e.map: electrostatic map

Docking with the AutoDock Vina Scoring Function

By default, AutoDock Vina uses its own scoring function (Vina scoring function). In this case, precomputed affinity maps are not required; only the receptor PDBQT file and the definition of the search space (grid box) are needed.

vina --receptor 1iep_receptor.pdbqt --ligand 1iep_ligand.pdbqt \
   --config 1iep_receptor.box.txt \
   --exhaustiveness=32 --out 1iep_ligand_vina_out.pdbqt

Important parameters:

• --receptor: receptor PDBQT file

• --ligand: ligand PDBQT file

• --center_x/y/z: center coordinates of the search space

• --size_x/y/z: dimensions of the search space in Å

• --exhaustiveness: thoroughness of the search

• --out: output file containing the docking results

Larger values of exhaustiveness result in a more extensive search but also increase the computation time. According to the documentation, imatinib is a particularly challenging molecule to dock, therefore it is often beneficial to use a higher value than the default setting of 8.

At the end of the run, the following output is obtained:

mode |   affinity | dist from best mode
     | (kcal/mol) | rmsd l.b.| rmsd u.b.
-----+------------+----------+----------
    1       -13.24          0          0
    2       -11.28      2.996      12.41
    3       -11.14      3.799      12.25
    4        -10.7      1.106      1.622
    5       -10.61      2.572      12.62
    6       -10.27      1.741      13.59
    7       -9.613      3.042      12.57
    8       -9.454      2.363       12.8
    9       -8.981      3.877      12.66

The affinity is reported in kcal/mol. More negative values indicate stronger predicted binding.

Note

The rmsd l.b. and rmsd u.b. values reported by AutoDock Vina do not represent the RMSD relative to the crystal structure. Instead, they describe the deviation of each docked pose from the highest-ranked pose (the pose listed first in the results). Comparison with the experimental crystal structure requires a separate RMSD calculation.

Docking with the AutoDock4 Scoring Function

One of the new features introduced in AutoDock Vina 1.2.x is the ability to use the AutoDock4 scoring function. In this case, the AutoGrid4 affinity maps generated previously are required.

vina --ligand 1iep_ligand.pdbqt --maps 1iep_receptor --scoring ad4 \
     --exhaustiveness 32 --out 1iep_ligand_ad4_out.pdbqt

Important parameters:

• --ligand: ligand PDBQT file

• --maps: prefix of the maps generated by AutoGrid4

• --scoring ad4: use the AutoDock4 scoring function

• --exhaustiveness: search thoroughness

• --out: output file

When --maps 1iep_receptor is specified, the program automatically loads the corresponding .maps.fld and .map files.

At the end of the calculation, the following result is reported:

mode |   affinity | dist from best mode
     | (kcal/mol) | rmsd l.b.| rmsd u.b.
-----+------------+----------+----------
    1       -15.21          0          0
    2       -15.07     0.8504      1.045
    3       -13.71      1.291      1.808
    4       -12.72      1.374       2.01
    5       -12.65      3.986      6.062
    6       -11.75      1.629      2.555
    7       -11.25      4.993      11.35
    8       -10.87      3.587      11.47
    9       -10.86       3.55      5.707

Comparison of the Vina and AutoDock4 Force Fields

Vina Scoring Function

Advantages:

• simple to use

• no need to run AutoGrid4

• faster preparation workflow

• often a suitable choice for virtual screening

Disadvantages:

• may be less accurate for certain ligand-receptor systems

• not fully compatible with classical AutoDock4 protocols

AutoDock4 Scoring Function

Advantages:

• uses more detailed atom-type-specific energy maps

• many older publications and validated protocols are based on it

• may provide better reproduction of experimental binding poses for some systems

Disadvantages:

• requires additional preparation steps

• AutoGrid4 calculations are mandatory

• involves managing a larger number of files

It is often worthwhile to dock the same ligand using both scoring functions and compare:

• the predicted binding energies,

• the resulting binding poses,

• key molecular interactions,

• RMSD values relative to the crystal structure.

Example:

Method	Affinity (kcal/mol)
Vina	-13.24
AD4	-15.21

The energies should be compared with caution because the two scoring functions are based on different mathematical models. Comparison of the resulting binding poses and key interactions is generally more informative.

The highest-affinity docking pose obtained using the AutoDock4 force field is shown in red, while the crystal structure is shown in green.

The highest-affinity docking pose obtained using the AutoDock4 force field is shown in red, while the highest-affinity pose obtained using the Vina scoring function is shown in orange.

Generally accepted RMSD values relative to the crystal structure:

RMSD (Å)	Evaluation
<2	Excellent reproduction
2-3	Acceptable
>3	Significant deviation

Note

The RMSD between the crystal structure and the docked ligand can only be considered reliable if the atom mapping between the two structures is correct.

Source of the example: https://autodock-vina.readthedocs.io/en/latest/docking_basic.html