Re: [AMBER] transforming Antechamber GAFF parameters into another force field

From: Nicholas Musolino <musolino.MIT.EDU>
Date: Fri, 26 Mar 2010 18:16:04 -0400

Dear Denny,

It seems like there are two questions you're asking: (1) is it possible to perform GAFF parameterization in an automated fashion? And (2) is it possible to convert GAFF parameters to some other force field?

To answer question (1), I have been able to successfully employ GAFF in an automated fashion, provided that the inputs to it aren't too bad. By that I mean, I supply antechamber with MOL2 files with atomtype and topology already established, and with reasonable structures. I found that when supplying PDBs and asking Antechamber to do atomtype perception, bondtype perception, and parameter assignment, it was too much work for antechamber to do reliably over hundreds of molecules.

In practical terms, to automate this process, you could use a shell (BASH) script, and program the script to do whatever you would do by hand for an individual case.

To answer question (2), it's not immediately clear what you want to do. Are you saying that you want to use the GAFF parameters alone in a simulation (e.g. a pure liquid of some novel molecule)? Or that you want to transform the GAFF parameters to work with some other FF? Remember, mixing and matching force fields in one simulation (or in one study of a molecule) is, generally speaking, not a good idea.

Don't forget that you can use the AMBER force field and appropriately formatted PARM7 files with other MD packages, such as NAMD. See

Hope this is helpful!

Nicholas Musolino

Nicholas Musolino
Ph.D. candidate, Department of Chemical Engineering, MIT | 617-253-6675 | Room E19-528

On Mar 26, 2010, at 6:00 PM, Denny Bromley wrote:

> I need to parameterize 100-200 small molecules into our force field and
> would like to avoid doing it all from scratch. Antechamber would be perfect
> except that we aren't using Amber (or GAFF). However, it does provide
> parameters for a self-coherent molecule, so I feel like I'm almost there if
> I can just find a transform. I've got some ideas, but I don't really want
> to re-invent the wheel, particularly if they've been shown to not work.
> Have there been any successful efforts in this direction so far?

AMBER mailing list
Received on Fri Mar 26 2010 - 15:30:05 PDT
Custom Search