Re: AMBER: Identification of improper torsions in small molecules parameterized with GAFF by antechamber

From: David A. Case <>
Date: Tue, 28 Oct 2008 08:57:22 -0400

On Mon, Oct 27, 2008, John Chodera wrote:
> I have a question about how precisely improper torsions that are to be
> applied to small molecules that have been parmeterized for GAFF by
> antechamber are generated.
> As far as I can tell, there are at least THREE programs that identify
> all of the improper torsions in the AmberTools distribution:
> * prepgen
> * parmchk
> * leap
> The codebases for these three programs are entirely separate, and the
> algorithms they use for identifying impropers appear to all be
> different.
> LEaP appears to have the 'final word', in that it actually searches
> for torsion parameters to apply.

This is correct: leap makes the torsions that are actually used in the energy
evaluations. (I think prepgen is irrelevant here: any torsions in a prep file
are ignored by leap anyway [except for some debug printout]. We are trying to
wean people away from using prep-format files, in favor of mol2-format.)

> There are anecdotal reports (on the
> mailing list) of users finding that parmchk fails to generate all the
> parameters expected by LEaP, which will still complain about missing
> impropers until they are added to the gaff.dat file.

I think this is pretty rare(?). Junmei has reported that some very large
percentage of the NCI small molecules get all of the values they need from
either gaff.dat or parmchk. The virtual screening in DOCK (using amberscore)
also uses this route -- people on the list may be able to comment on how often
this really occurs.
> In converting small molecules over to Desmond, I need to identify all
> of the improper torsion angles in the small molecule. Would running
> 'prepgen' (or some other antechamber-based tool) be sufficient to do
> this, or should I run leap and examine the output if I want to
> generate something that will be exactly the same as what would be
> generated for Amber?

Safest is to run leap and examine the resulting prmtop file; the rdparm
program formats prmtop info in a little nicer fashion.

...good luck...dac

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
Received on Wed Oct 29 2008 - 05:10:33 PDT
Custom Search