Re: [AMBER] Side chain and backbone-specific parameters.

From: Lee-Ping Wang <>
Date: Tue, 10 Jun 2014 10:50:35 -0700

Hi Jason,

Thank you. I was mainly concerned that the frcmod format appears to support
atom classes with a maximum of two letters (which would make hundreds of
atom classes unwieldy), but your response sounds like this isn't a problem.

I'm curious, how is this handled? Can I have multiple atom classes that use
the same two letters (they just get used for different residues)?


- Lee-Ping

-----Original Message-----
From: Jason Swails []
Sent: Tuesday, June 10, 2014 10:47 AM
Subject: Re: [AMBER] Side chain and backbone-specific parameters.

On Tue, 2014-06-10 at 10:25 -0700, Lee-Ping Wang wrote:
> Hi everyone,
> I'm working on some new protein force fields which have distinct
> backbone / sidechain parameters for different amino acids. My
> parameterizations are done using Gromacs, which allows atom-specific
bonded interactions (i.e.
> overriding the values that come from atom classes).
> I don't see this functionality in AMBER, and it looks like the
> ff99sb-ildn and ff12sb models have extra atom classes to accommodate the
extra torsions.
> Since I plan to port my force fields to AMBER, I'd like your advice on
> how to do this most effectively.
> What is the best way to implement residue-specific backbone and side
> chain parameters? Should this be done by adding many atom classes?


> Is it even
> possible to add on the order of 100 atom classes?

Yes. tleap should create a condensed set of vdW atom classes by combining
atoms that have the same vdW parameters into a single class, but should have
no problems handling a very large number of atom classes for parameter

Indeed, loading the ff14SB force field with GLYCAM, lipid14, and GAFF force
fields loads ~200 atom types already.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
AMBER mailing list
Received on Tue Jun 10 2014 - 11:00:04 PDT
Custom Search