Re: [AMBER-Developers] Chi torsions in the ff10 force field

From: Thomas Cheatham III <>
Date: Mon, 28 Mar 2011 12:23:44 -0600 (Mountain Daylight Time)

To follow up on the "chi" force field discussion for those not present at
the recent AMBER developers meeting (where we discussed to some degree
user's experiences with the RNA force fields and talked about what to
include in ff10) I'll summarize...

There have been three reparameterizations of the chi torsional potential
for nucleic acids relevant to AMBER ff9X (that I am aware of).


(1) Ode et al. JCC 29, 2541-2542 (2008) - MP2/aug-cc-pVTZ//HF/6-31+G**
chi surfaces at 30 deg intervals for base models, tests with ff99 +

(2) Yildirim et al. JCTC 6, 1520-1531 (2010) - MP2/6-31G*//HF/6-31G* chi
surfaces for full nucleotides at 5 deg intervals, both C3'-endo and
C2'-endo puckers, fit to ff99.

(3) Banas et al. JCTC 6, 3836 (2010) - MP2/CBS//PBE/6-311++G(3df,3pd)/
COSMO chi surfaces at 10 deg intervals, both c3'-endo and C2'-endo puckers
(deoxy and ribo), fit/tested with ff94, ff99 + ff99bsc0.

The most recent work includes higher levels of theory, inclusion of
solvation, and separate optimizations at MM and QM. Although this latter
force field has already been made available in the Banas et al. paper, a
full discussion of the OL parameterization has not yet been published as
the paper is tied up in review. The "OL3" force field currently within
ff10 refers to the "OL3" variant which was renamed to "OL" in the Banas et
al. paper. I agree that this is confusing.

For ff10, inclusion of an updated chi potential for RNA is desirable as
pointed out by Ilyas. As the Ode et al. and Yildirim et al. papers
provided only very limited testing on RNA with ff99, we (i.e.
independently Sponer's group and mine among others) began testing all of
these force fields in long and varied RNA simulations with ff94, ff99, and
ff99bsc0 starting in ~May 2010; this testing slowed the release of the OL
force field (and submission of the papers). A key goal of the OL force
field was to avoid the "ladder-like" underwound A-RNA helical structure
observed in longer simulations of RNA helices (see Mlynsky et al. JPCB
114, 6642-6652 (2010)) due to population of chi angles in the high-anti

Although our simulations suggest that all of the chi modifications improve
the results of RNA simulation by providing a more realistic description of
chi, particularly with the syn regions in guanine, the Ode force field is
not recommended because the high-anti "ladder-like" RNA structure is
actually promoted. Also, the chi modifications do not seem useful for
B-DNA. This is why ff10 for DNA recommends ff99+bsc0.

For the RNA tetraloops, "better" behavior is seen with both the Yildirim
and OL (OL3) force fields (as discussed in the Banas et al. paper).
However, the Yildirim force field may over-correct chi in the high-anti
region leading to too low roll / inclination / propeller and a wider
groove. Moreover, both of these force fields appear to have issues with
bulges and do not maintain all tetraloop structures in long (> 200 ns)
simulations (unpublished) and therefore I believe that none of these fully
solve the problems with the RNA force fields. We are continuing to test
out both sets of force fields and we and other AMBER developers are
continuing work to improve the RNA force fields. As for the AmberTools
1.5 snapshot we can state that the chi modifications improve the treatment
of RNA with ff99+bsc0... Based on our combined experiences and
discussions, for now we recommend for RNA ff99+bsc0+OL noting that this
recommendation will continue to evolve.

I have not yet looked at what is in the manual but will attempt to update
it to add discussion of the chi modifications pointing out that the Ode et
al. parameters should be avoided and that the Yildirim are a reasonable
alternative to OL and that further tests with each are on-going.


\-/ Thomas E. Cheatham, III (Associate Professor) College of Pharmacy
-/- Departments of Med. Chem. and of Pharmaceutics and Pharm. Chem.
/-\ Adjunct Asst Prof of Bioeng.; Center for High Performance Computing
\-/ University of Utah, 30 S 2000 E, SH 201, Salt Lake City, UT 84112
/-\ (801) 587-9652; FAX: (801) 585-9119
\-/ BPRP295A

AMBER-Developers mailing list
Received on Mon Mar 28 2011 - 11:30:03 PDT
Custom Search