Re: [AMBER] Forcefield_PTM

From: Jason Swails <>
Date: Tue, 21 Oct 2014 09:02:30 -0400

On Tue, 2014-10-21 at 17:55 +0530, Sourav Purohit wrote:
> Hi all.
> The forcefield_PTM paper says that it is compatible with ff03. I tried to
> use it with ff10 just to check what happens. While preparing the system in
> xleap, it works fine.

In general, Amber force fields can be broken down into 3 different
"families" based on the charge derivation philosophy.

The original Cornell, et al. force fields (starting with ff94) uses
HF/6-31G* Merz-Kollman RESP charges under the assumption that lack of
electron correlation overpolarizes the charges consistent with the
effect aqueous solvent has on the solute. Newer members of this
"family" of force fields kept the same charges, but modified certain
torsion parameters to improve agreement with either experimental or
high-level QM measurements. These include ff96, ff99, ff99SB, ff10 (same
as 99SB for proteins), ff12SB, and ff14SB.

The other two families have, as far as I know, only 1 member each. The
ff03 family still uses RESP to derive charges, but uses B3LYP/cc-pVTZ
rather than HF/6-31G* to derive charges. This DFT method accounts for
exchange and correlation, and so yields less-polarized charges than the
original Merz-Kollman approach.

The last family is quite new -- ff14ipq, which uses an iterative
approach to compute self-consistent charges based on conformational
ensembles of MP2/cc-pVTZ RESP fits. (Iterative because as the charges
change, so too does the force field and the resulting conformational
ensemble, which would change the charges...)

You can't assume that parameters derived with different charge
philosophies are compatible with each other. In fact, prevailing
opinion dictates that they are considered *incompatible* until proven

So yes, it "works", but you will have to do more work to prove it's any
good if you want to satisfy would-be reviewers.

> There is only one unusual warning about improper
> torsions:
> *Building improper torsion parameters.old PREP-specified impropers: <ALY
> 16>: CA H N -M <ALY 16>: CH CE NZ HZ <ALY 16>: CH3
> NZ CH OH <ALY 16>: CA O C +M total 74 improper torsions
> applied 4 improper torsions in old prep form*
> Building H-Bond parameters.
> Incorporating Non-Bonded adjustments.
> Not Marking per-residue atom chain types.
> Marking per-residue atom chain types.
> (Residues lacking connect0/connect1 -
> these don't have chain types marked:
> res total affected
> CASN 1
> NSER 1
> WAT 10351
> What does this warning mean??

I don't see any warnings, but maybe I'm missing something? There's some
information about improper torsion terms that have been applied, but
that's all...

> If I carry the simulation ahead with the
> topology file generated, will the simulation be fine??

You will have to determine that for yourself. To my knowledge, nobody
has ever tried combining this force field with ff99SB or ff10 before.
As a result, you can't point to any previous work as evidence that this
combination will give valid scientific results. So if you insist on
using ff10 instead of ff03, you will need to justify your decision and
provide evidence that your combination is valid.

However, the impropers here are not a problem.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Tue Oct 21 2014 - 06:30:02 PDT
Custom Search