AutoDock

How to start a calculation

1. Protein coordinates and charges
2. Ligand coordinates and charges
3. A starting structure of the ligand
4. A box including the expected binding site

1. Start with a pdb structure of the protein (a crystal structure or an equilibrated structure).

 
2. Remove alternative configurations (keep the one with the highest occupancy).

 
3. Add hydrogen atoms and charges to the structure.
This is done by the first 15 steps in the Amber-equilibration page.
If you have an equilibrated structure, this is already done, but then you should remove all water molecules instead.
 
4. Remove all water molecules.
   
   
5. Remove non-polar hydrogen atoms (adding their charge onto the carbon atom).
This is done automatically by changepdb, command dock .
 
6. Run addsol to add atomic solvation parameters.

addsol  protein.pdbq protein.pdbqs
 
7. Add hydrogens to the ligand (for example using Spartan) and write a pdb file.

 
8. Calculate ESP charges of the ligand at the AM1 level and add to the coordinate file (giving a ligand.pdbq file). Add the charges to the pdb file with the decimal point in column 58.

ATOM      1  A   INH     1       5.475   0.679  -3.852                  0.046
 
9. Run atmtobnd to get a connect list for the ligand:

pdbtoatm ligand.pdbq | atmtobnd >ligand.bnd
 
10. Define the torsions of the ligand you want to explor during the docking using autotors:

autotors -h -a -A ligand.bnd ligand.pdbq ligand.out.pdbq
Ensure that you have a picture with atom numbers or names of the ligand when running the program.
Answer the questions of this interactive program.
Define any ring system (or any other atom) as root (which is kept rigid).
Remove non-wanted torsions by writing d followed by a list of the numbers of the torsions you do not want to allow to rotate (one on each line; not that you will not get any further prompt).
Stop with a c.
 
11. Check the ligand charges by

check-qs ligand.out
(note that you should not give the .pdbq suffix of the  ligand .out.pdbq file). Possible errors are found in the file ligand .out.err.
 
12. Construct the grid parameter file (protein.gpf) by

mkgpf3 ligand.out.pdbq protein.pdbq
 
13. Construct the docking parameter file (ligand.proteindpf) by

mkdpf3 ligand.out.pdbq protein.pdbq
 
14. Edit the protein.gpf file.

You probably want to edit
npts: the number of points along each coordinate axis (60 60 60 will take about 10 min / conformation)
spacing: the spacing (in Å) between each grid (ensure that the whole active site fit in)
gridcentre: the centre of the grid (enter the centre of the active site)
You should ensure that you have a box that includes the whole protein or the active site. You get the size of the protein and the ligand by changepdb, command size. You can find the size of the active site by running rasmol and select atoms at the end of the active site.
 
15. Calculate the potential energy grid maps:

nice -19 autogrid3 -p protein.gpf -l protein.glg&
This may take a few minutes (e.g. 5 minutes for a 60*60*60 grid).
Check the result at the end of the file protein.glg:
Van der Waals energies should range between -10 and 1E+5 kcal/mole, whereas hydrogen bond energies should be -1000 to +1000 kcal/mole.
 
16. Edit the ligand.protein.dpf file.

You may want to edit:
ga_run: the number of final conformations; each conformation takes about one minute.
#do_local_only: uncomment this line (remove #) and comment out the ga_run line if you only want to minimise the structure
ga_pop_size;
ga_num_evals;
ga_num_generation: increase these variables if you want longer and more detailed calculations.


17. Run the docking with
nice -19 autodock3 -p ligand.out.protein.dpf -l ligand.protein.dlg &
This takes ~10 min for 10 conformations.
 
18. The interesting results are found in the ligand.protein .dlg file, especially after the title CLUSTERING HISTOGRAM. There you find the final structures ordered in groups (clusters) of similar structures. The RMSD TABLE give further information, especially the energies.

Next you find the pdb coordinates of the various structures; these can be directly copied into a pdb file to examine the result visually.
You may also copy all the final coordinates to a file by
cat protein.pdbq > file; grep ATOM ligand.protein.dlg  | grep -v DOCKED >> file
You can visualise this file by rasmol and select each docked conformation - they are numbered consecutively from 1 and upwards.
 
19. It may be useful to en the simulation by a simple energy minimisation, by uncommenting the line

#do_local_only 50
in the ligand.protein.dpf file.

Notes

• The protein is rigid; only the ligand is flexible.
• The can be at most 32 rotable bonds.

Default Lennard-Jones parameters

r(ij)=r(i)+r(j)
e(ij)=Sqrt(e(i)*e(j))

How to compile some utility programs

Autotors
cd /home/bio/Autodock/src/autotors
cc autotors.c -o autotors -lm
cd /home/bio/Autodock/bin/sgi4D.IRIX64_6.5_R10000
c ../../src/autotors/autotors autotors

Atmtobnd
cd /home/bio/Autodock/src/atmtobnd
cc atmtobnd.c -o atmtobnd -lm     (ignore three warnings)
cd /home/bio/Autodock/bin/sgi4D.IRIX64_6.5_R10000
cp ../../src/atmtobnd/atmtobnd .

Autodock
Follow the instructions in http://www.biochemistry.unimelb.edu.au/pscotney/autodock/AutoDock-HOWTO-3.html to change the Make file.
In addition we had to make the following two changes:
1. Comment out the following lines in file get_atom_type.cc
/*
    extern "C"
    {
*/
        #include <stdio.h>
        #include "get_atom_type.h"
/*
    }
*/

2. In Makefile, add -static on this row:
autodock3 : main.o $(ADLIB)
        echo $(EXE)'  on  '`date`', using '`hostname` >> LATEST_MAKE
        echo 'Flags: '$(CC) $(LINK) -DNOSQRT -L. -lad $(LIB) >> LATEST_MAKE
        $(CC) $(LINK) -DNOSQRT -static -o $@ main.o -L. -lad $(LIB)

Autogrid 
  Follow the instructions in http://www.biochemistry.unimelb.edu.au/pscotney/autodock/AutoDock-HOWTO-3.html to change the Make file.
In addition comment out the following lines in file get_atom_type.c
/*
    extern "C"
    {
*/
        #include <stdio.h>
        #include "get_atom_type.h"
/*
    }
*/