Re: [AMBER] Which force field is "the best"

From: David A Case <>
Date: Sun, 9 Nov 2014 17:29:46 -0500

On Sun, Nov 09, 2014, Gonzalo Jimenez wrote:
> Sorry for this probably very naive question, but since more and more
> force fields are being released and now that AMBER14 is available, I
> was wondering if the question “which is the best force field for
> proteins?” can be easily answered.

> I have been using ff99SB and ff99SBildn for years, with very good
> results, but since ff02 and ff14 are now available, I wonder if there is
> one modern favorite or default force field to use these days.

If I could jump in where angels fear to tread:

The consensus of Amber developers (a fairly limited community) is that the
parameters in ff14SB are likely in general to give better results than earlier
versions of Amber force fields. The issues addressed with the "ildn"
additions have also been addressed (in generally similar ways) in ff14SB.
My own work, which is far from exhaustive, seems to favor ff14SB over ff99SB
in virtually every circumstance.

The parameters in ff14ipq represent an alternative approach to
parameterization; this represents a more marked departure from other "Amber"
ff's, and should certainly be treated with caution until we have more
experience with it. (Note that the paper describing ff14ipq is now available:
JCTC 10: 4515, 2014.)

The ff02 force field is not generally recommended, and has been moved to the
"oldff" folder to indicate this. Amber's ff03 force field is still widely
used, and has a number of adherents: be sure to use the "r1" revision (in

Of course, results can depend in important ways on the types of simulations
you are doing. Force field requirements for looking a fluctuations about
the native state for well-folded proteins are presumably different than those
for looking at protein folding events, or at conformational preferences in
peptides; both in turn are probably different than what is need to look at
membrane proteins; and so on....

The code in Amber14 makes it much easier to consider non-Amber force fields,
the CHARMM c36 ff is perhaps worth noting in particular.


AMBER mailing list
Received on Sun Nov 09 2014 - 15:00:02 PST
Custom Search