ADL: Best way to assign partial charges

Douglas Houston dhouston at
Fri Apr 17 01:35:14 PDT 2009

Thanks for all the info Sargis, that's great.

I've just read in Ruth's 2007 paper that the calibration was indeed  
performed using "charges added using the Gasteiger PEOE method."

It also says that "The Gasteiger method was chosen for its fast and  
easy operation and ready availability as part of Babel."

So it would appear that this calibration did NOT use the script. It is worrying that you found differences  
between the Gasteiger charges computed by the scripts and those  
computed by OpenBabel.

It seems to me that even if the official advice is to use the .py  
scripts, using Babel to assign charges would be superior since that's  
how AD4's force field was calibrated.

Since all my work so far has been done using the .py scripts, I need  
to be sure that this conclusion is correct before I go back and redo  
everything. Of course, it may well be that I'm worrying about nothing  

Please advise if this is wrong,


Quoting Sargis Dallakyan <sargis at>:

> Douglas Houston wrote:
>> I've tried to find out as much as I can about the best way to   
>> assign partial charges to the receptor, but I've only succeeded in   
>> making myself more confused.
>> According to the manual, "appropriate partial atomic charges must   
>> be assigned. This can be achieved by
>> the user’s preferred method, e.g. using InsightII, Quanta, Sybyl,   
>> AMBER or CHARMm. Alter-
>> natively, one of the shell scripts described in the Appendix can be used."
>> According to the FAQ on the website, "the AD4 scoring function was   
>> calibrated using Gasteiger partial charges on both the ligand and   
>> the macromolecule."
>> But according to the front page of the website "AutoDock 4.0 now   
>> has a free-energy scoring function that is based on ... the AMBER   
>> force field"
>> So can anyone answer the following questions:
>> 1. Does the scoring function use Gasteiger charges or the AMBER force field?
> It is my understanding that AutoDock 4.0 is using Gasteiger charges and
> the AMBER force field. If you look at the AutoDock force field [1] and
> AMBER force field [2] equations: ΔGvdW, ΔGH-bond and ΔGelec are similar
> up to a weight (W) factor.
> [1]
> [2]
> As you probably know, previous version of AutoDock (3.0) was calibrated
> using Kollman charges for protein, which I believe is the same as AMBER
> charges. From the computational point of view Kollman/AMBER charges are
> taken from a lookup table whereas Gasteiger-Marsili charges are
> computed  using Partial Equalization of Orbital Electronegativities
> (PEOE) algorithm.
>> 2. Was AD4 calibrated using Gasteiger charges assigned by the   
>> script? If not, how exactly were they assigned?
> I think so.
>> 3, Is there an "official" recommended way of assigning charges to   
>> the receptor before calculating grids, and if so, what is it?
> I've recently done virtual screenings with PyRx
> (, where I used Gasteiger charges
> computed by PyBable module which is a re-implementation of Babel 1.6 in
> Python and which is used by and
> I then computed Gasteiger-Marsili partial charges
> using openbabel 2.2.0 ( and
> used that charges in pdbqt files for the ligands. There are noticeable
> changes in the numerical values between these two methods, however, the
> final top 10 ranked compounds were almost identical.
> The bottom line is that it is "official" recommended to use
> and to assign partial atomic
> charges, however users are free to chose any other packages they want.
> Hope this clarifies things a bit.
> Sargis
> -- 
> Sargis Dallakyan, Ph.D. - Research Programmer III
> The Scripps Research Institute,
> Dept. Molecular Biology,  MB-5,
> 10550  North Torrey Pines Road,
> La Jolla,  CA 92037-1000
> Tel: (858) 784-9559
> ________________________________________________
> --- ADL: AutoDock List  --- ---

Dr. Douglas R. Houston
Room 3.23
Institute of Structural and Molecular Biology
Michael Swann Building
King's Buildings
University of Edinburgh
Edinburgh, EH9 3JR, UK
Tel. 0131 650 7358

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

More information about the autodock mailing list