Autodock

See discussions, stats, and author profiles for this publication at:
https://www.researchgate.net/publication/341830560
AutoDock Tutorial
Method · June 2020
DOI: 10.13140/RG.2.2.27339.21289
CITATIONS
0
READS
6,460
1 author:
Some of the authors of this publication are also working on these related projects:
The discovery of novel class IIa HDACs inhibitors View project
Molecular Genetics View project
Ammar Elmezayen
University of Texas Southwestern Medical Centre

At the start of this tutorial, all you need is a directory (e.g., “Docking”) which consists of two
files;
protein.pdb and ligand.mol2.

1. Open ADT tool of AutoDock.

2. File -> Preferences -> Set: change the startup directory.
3. Ligand -> Input -> Open: open your ligand mol2 file.
4. Ligand -> Torsion Tree -> Choose Torsions: decide on the rotatable bonds to be
considered.
5. Ligand -> Output -> Save as PDBQT: save your ligand file with “pdbqt” extension (e.g.
ligand.pdbqt). Type in the extension manually.
6. Grid -> Macromolecule -> Open: Open your protein pdb file (e.g. protein.pdb), and
ADT tool will prompt a new window to save your protein into pdbqt file.
7. Grid -> Set Map Types -> Choose Ligand: select the ligand molecule from the list
shown.
8. Grid -> Grid Box: Select the number of grid points in x, y and z directions. Usually, the
default 40 is taken. However, you can increase or decrease that number in any direction.
From “Center” menu, select “Center on ligand”. Leave other parameters on default.
(Observe your grid box. Make sure that it covers the whole ligand as well as the binding
site).
9. File -> Close saving current: save your settings before closing. Otherwise, your settings
will
be lost if the program is closed.
10. Grid -> Output -> Save GPF: save your gpf file (e.g. docking.gpf). Type in the
extension
manually.
11. Check out the content of Grid Parameter File, docking.gpf (by Notepad).
12. Docking -> Macromolecule -> Set Rigid Filename: Select your macromolecule protein
file
(e.g. protein.pdbqt).
13. Docking -> Ligand -> Choose: select your ligand. Leave everything on default and
select
Accept.
14. Docking -> Search Parameters -> Genetic Algorithm: set the number of GA Runs as
20.
Also, set the maximum number of “evals” to
long = 25000000 depending on the number
rotatable bonds. Leave other parameters on default and select Accept.
15. Docking -> Docking Parameters: Leave on default.
16. Docking -> Output -> Lamarckian GA: save your dpf file (e.g. docking.dpf). Type in
the extension manually.
17. Close your ADT tool.
In your working directory you should now have the following four files:
• ligand.pdbqt • receptor.pdbqt • docking.gpf • docking.dpf

A. Identifying potential binding sites for Ligands

1. Open model.pdb in Discovery Studio:-
• Type ds model.pdb in terminal.
• Open lig1.pdb by:- File ---> Insert From ---> file ---> choose file location from
dialogue box ---> choose lig1.pdb ---> Open.
• Click on Receptor-Ligand Interactions on top.
• From the side panel generated in the LHS, the last tab will be of Define and Edit
Binding site.
◦ It will take us to Define Site.
◦ Under Define Site, one will find ‘From Receptor Cavities’. A dialog box as
shown in the image will appear ---> click OK
◦ Another dialog box will appear. Click on ‘OK’
◦ A sphere appears for site 1 on the protein molecule, which depicts all the
possible binding sites. Number of binding sites can also be seen on the left-hand
side panel. Large number of ions are shown, where the ligand atoms can bind.
◦ Selected all the potential Binding Sites for Ligand 1.

• Now, open the other two ligands in Discovery Studio.

◦ Find the Potential Ligand Binding Sites for each of the Ligands

B. Structure Based Docking

We perform Blind Docking to check interaction of the Ligand at the Binding Site

Steps:-

1. To mutate non-standard Amino Acids:-

• Load the protein model.pdb in Chimera
• Select --> Select All
• Select --> Residue → ‘MSE’. The Amino Acid ‘MSE’ will get selected.
• Tools → Structure Editing → Rotamers → long press ‘Rotamer Type’ and select
‘MET’. Then click ‘OK’. Then a dialog box will appear. Click on the first option
showing the first Rotamer
 • Save PDB as model_fin.pdb
2. Open model_fin.pdb in Autodock:- adt model_fin.pdb
• View the Protein as a Ribbon by clicking on the ‘radio button’ under R in the Left
Side panel and deselecting the ‘radio button’ under L. ‘L’ stands for ‘Lines’ depiction
of the protein

3. Preparing the Protein molecule:-

• Deleting water:- Edit ---> Delete Water
• Adding Hydrogens:- Edit ---> Hydrogens ---> Add ---> Polar only ---> Ok
• Adding Charges:- Edit ---> Charges ---> Add Kollman charges
The added Hydrogens can be seen in the Lines depiction of the protein as:-
• Checking for Missing atoms:-
◦ Edit ---> Misc ---> check for Missing atoms ---> Select All Residues --->
Dismiss.
◦ Again go to- Edit ---> Misc ---> Repair for Missing atoms ---> Dismiss
• Grid -> Macromolecule ---> choose ---> click on model_fin ---> Select Molecule
---> Save as pdbqt
◦ We now have a file named as model_fin.pdbqt
4. Now, to load Ligand files, we:-
◦ File ---> Read Molecule ---> directory where the Ligand files are present –->
choose the ligand file/s one by one. Remember to add Polar Hydrogens.
◦ Perform the same for all the Ligand files
5. Now, select the Ligand for Autodocking:-
• Grid ---> Macromolecule ---> Choose –-> Then choose the ‘lig1’ shown in the
dialog box
• Then click on ‘OK’ in the dialog box that appears
• A file called- lig1.pdbqt has been saved
6. Now, delete both the pdb molecules. This is done by selecting the protein/ligand in the
Left Hand panel, and then Right Click ---> delete

7. Load the pdbqt files of both the Protein and the Ligand.
• File –-> Read Molecule ---> model_fin.pdbqt
• For Gridding:-
◦ Select your molecule by:-
▪ Grid ---> Macromolecule ---> Choose ---> select the molecule:- ‘model_fin’
◦ Then choose your Ligand by:-
▪ Grid ---> Set map Types ---> Choose Ligand ---> select Ligand:- ‘lig1’
◦ Then make a box by:-
 ▪ Grid --- > Grid Box :- Will show a 3D differently coloured box around the
molecule. The adjustments to the box dimensions will be available in a menu
• Save the grid as a ‘Grid Parameter File’.
◦ Grid --- > Output --- > Save GPF --- > set the directory you want to save it in
and set the name:- model_fin_Grid
◦ Will create a file named model_fin_Grid.gpf
8. Now, we do Autodock:-
• Close the Grid adjustment menu after saving the gpf file.
• Run AutoGrid:-
◦ Run --- > Run Autogrid --- > Choose parameter filename:- ‘Browse’ next to
‘parameter filename’ --- > Select model_fin_Grid.gpf file as your parameter
file --- > A log file- .glg file, i.e. Grid Log file will automatically be saved --- >
‘Launch’
• Prepare files for docking:-
◦ For Protein:- Docking --- > Macromolecule --- > Set Rigid Filename --- >
select model_fin.pdbqt
◦ For Ligand:- Docking --- > Ligand --- > Choose --- > lig1
▪ Select ‘Ok’ on all the dialog boxes that pop up
◦ Docking --- > Search parameters --- > Genetic Algorithms --- > Accept
◦ Docking --- > Docking Parameters --- > Accept
◦ Docking --- > Output --- > Lamarckian (GA 4,2) --- > Enter the output filename
as:- model_lig1_parameter_output.dpf
◦ Run --- > Autodock --- > set Log Filename as :- model_lig1_docking.dlg --- >
set Cmd from :- autodock4 -p model_lig1_parameter_output.dpf -l & to
autodock4 -p model_lig1_parameter_output.dpf -l model_lig1_docking.dlg &
• Visualising all the models generated of the Ligand, by loading the protein --- >
Analyse --- > Dockings --- > Open --- > open the .dlg file . Then, we do:- Analyse
--- > conformations --- > ‘Play’ or ‘Play, ranked by energy’
• Open the model_lig1_docking.dlg file ---> look for the lowest energy. In the image
below, lowest energy is –6.6.8 of the Run 5, i.e docking model 5. So, we searched
‘Run=5’ copied the coordinates of MODEL 5 to text editor and saved it as a .pdb
file. (file named: model_fin_run5.pdb)

9. Now, we grid the Ligand around the lowest energy binding site, i.e. ‘run=5’.
• Load the Protein molecule. File --- > Read Molecule --- > choose model_fin.pdbqt
• Open the model_lig1_docking.dlg file.
◦ Analyse --- > Dockings --- > Open --- > choose model_lig1_docking.dlg
◦ Anaylse --- > Conformations --- > ‘Play, ranked by energy’
◦ Place the Ligand at Run=5, and then draw a grid around it:-
▪ Choose the Protein first:- Grid --- >Macromolecule --- > ‘model_fin’
 ▪ Choose the Ligand:- Grid --- > Set Map Types –- > Choose Ligand --- >
‘lig1’
10. To get the pdb output of the best Ligand conformation run Vina:
• Make a config.txt file to put all the grid box parameters in it and run command:
◦ vina –config config.txt –log vina_lig1.log –ligand lig1.pdbqt
▪ A log file and a lig1_out.pdbqt is generated
◦ Now get the pdb coordinates of the conformations using openbabel:
▪ obabel -ipdbqt lig1_out.pdbqt -opdb -O lig1_vina.pdb

◦ Only 1 model was made in which the first one is the best as it has the least
energy.

Each Ligand has 9 conformations. For each Ligand, the first conformation is the
best one

11. Now, open the Ligand and Protein in Discovery Studio

• ds model_fin.pdb
• File --- > Insert From --- > Files --- > select the best model of the ligand
• Now, to see the interactions:-
◦ View Interactions --- > under ‘Show Receptor-Ligand Interactions in 2D”,
select:- Show 2D Diagram
◦ Do for the other 2 Ligands
• The ligand was modified:-
• And the autodock vina was run to test the affinity of the ligand to the molecule. The
molecule was initially at -9.6 kcal/mol;, but reduced to -9.7kcal/mol

Homology Modelling
Predict the secondary structure of the given sequence using Chou & Fasman, GOR
& Neural Network methods.
Sequence

>query
MAGHIVLNTGAKMPMLGLGTWKSSPGQVTEAVKTAIDVGYRHIDCAHVYQNEQEVGLAIQ
QKLKEQVVTRKELFIVSKLWCTYHEKSMVKEACKKTLRDLQLDYLDLYLIHWPTGFKAGK
EFFPLDEAGNVIPSDTDVLDTWTAMEELVDEGLVAAIGISNFNHLQIERILNKPGLKYKPAV
NQIECHPYLTQEKLVSYCQSKGIVVTAYSPLGSPDRPWAKPDDPSLLEDPRIKEIAAKHNK
TSAQVLIRFPMQRNLVVIPKSVTPERIAENFKVFDFELSSQDVTTLLSYNRNWRVCALMSC
ANHKDYPFNEEF
1. We went to this website:- http://cib.cf.ocha.ac.jp/bitool/MIX/
2. The inputted the protein Sequence
Structure Prediction by Chou Fasman
Structure Predition by GOR method

Structure prediction by
NN
BLAST the given sequence against PDB and find a suitable template from any plant
species for homology modelling.
Report the query coverage, percent identity, PDB code, resolution of the data and
the chain ID of the chosen template.

1. The Query Protein Sequence was taken, and then BLAST was performed

2. Then among all the BLAST results, the template chosen was from the plant species:-
Oryza sativa
Query Coverage= 88%
Percent Identity= 49.11%
PDB Code:- 6KBL
Resolution:- 1.7 A
Chain ID:-
Ensure that you fix the template for alternate conformers, anisotropic thermal
parameters and non-standard residues, if any.

We do the normal file cleaning Protocol

1. CCP4i
2. Then put in command line:-
• grep -v "HOH" 6kbl_pdbset3.pdb | grep -v "HETATM" | grep -v "CONECT" | grep -iv
"ANISOU" | sed '/CRYST1/,$!d' >6kbl_final_cleaned.pdb
◦ 6kbl_final_cleaned.pdb is the final cleaned file

Construct 3 homology models of the given protein using MODELLER (5).

1. We first take out the sequence of the protein from CHIMERA
• Open it in Chimera
• Then Select--->‘Select All’
• Then Tools--->Sequence--->Sequence

2. Then we saved the sequence as:- 6kbl_sequence.fasta

3. We first go to CLUSTALW to do MSA of both the query and the 6kbl protein

4. Then we download the MSA file:-

The file was saved as:- query_6kbl.pir

5. Then appropriate changes were made to the modeller script

6. Then we make changes in the MSA file:-

7. Then we run the modeller script

• python3 model_default.py

8. This generated 3 Homology Models:-

• query.B99990001.pdb
• query.B99990002.pdb
• query.B99990003.pdb

Based on energy, Ramachandran outliers and RMSD with the template, identify the
best model

1. Based on Energy:-

The 3rd Model seems to be the best because it has the least energy

2. Based on Ramachandran Outliers

• Open in COOT
• coot
• File--->Open Coordinates--->select your file
• Validate--->Ramachandran Plot
Model 1 has 5 outliers

Now,
Model 2 has 7 outliers

Now,

Model 3 has only 3 outliers

According to this, Model3 seems best so far as it has least Outliers

3. Based on RMSD values

We will be comparing the RMSD of each model with the 6kbl parent protein we took as
template

• Comparison between 6kbl and query.B99990001

◦ ExecutiveAlign: 1811 atoms aligned.
Executive: RMSD = 2.799 (1811 to 1811 atoms)
• Comparison between 6kbl and query.B99990002
◦ ExecutiveAlign: 1811 atoms aligned.
Executive: RMSD = 2.699 (1811 to 1811 atoms)
• Comparison between 6kbl and query.B99990003
◦ ExecutiveAlign: 1811 atoms aligned.
Executive: RMSD = 2.482 (1811 to 1811 atoms)
According to this Model3 is the best
THUS WE CONCLUDE THAT MODEL3 IS THE BEST MODEL BASED ON ENERGY,
RAMACHANDRAN OUTLIERS, AND RMSD VALUES
Generate the cartoon, molecular surface and electrostatic surface of the
constructed model
Pymol was used
1. Cartoon generation:- Show--> As---> cartoon

2. Molecular Surface generation:- Show---> As---> surface

3. Generate Electrostatic surface:- Action---> generate---> Vaccum Electrostatics

Construct a superposition diagram of template and the best model.

This was done in Chimera
1. Both the PDB files were opened in Chimera
2. Tools---> Structure Comparison---> ‘Matchmaker’:- then choose the reference and
the structure to be compared

The RMSD was

found to be 0.234 A
between 300
pruned atom pairs