Re: Force field design

From: Kenno Vanommeslaeghe (
Date: Wed Jul 16 2014 - 12:06:23 CDT

On 07/15/2014 10:58 PM, Aron Broom wrote:
> 1) get the partial charges for your atoms based on at least one structure
> (you can use more than one to get a more accurate derivation, but might be
> simplest to stick with one). I'm not sure if CGenFF or any of the VMD
> tools allow for this, but I do know the RESP server
> is designed for this.

Nope, you don't start with the charges, you start with choosing a force
field, and then obtain charges that are _compatible_ with that force
field. For example, RESP charges have not been validated with the CHARMM
force field. And the MMFF94 charges assigned by SwissParam are even (far)
worse in this respect.

> 2) find a forcefield which has some general atom types such that it can
> handle all the bond, angle, and dihedral terms. I think there is such a
> thing for CHARMM, others on the list would have a better sense, but at
> least there is also the general amber forcefield. Then you either look at
> the atom types in some documentation of the forcefield, and assign yours
> based on their bonding/chemical environment, or perhaps let some automated
> script attempt this (I've only ever done it manually).
> 3) now you have a molecule that has bond, angle, and dihedral terms, and
> you add in (or it might get automatically done depending on the tools you
> are using) your partial charges, and viola! You have a classical
> CHARMM-like forcefield parameter for your molecule.

Using the General Amber Force Field (GAFF) will give you an *AMBER*-like
force field. To get a CHARMM-like force field, you need to use the CHARMM
General Force Field (CGenFF).

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:36 CST