[AMBER] AmberTools23 changes antechamber behavior for GAFF2

From: John Chodera via AMBER <amber.ambermd.org>
Date: Tue, 4 Jul 2023 10:37:04 -0500

Hi all!

I help maintain the OpenMM openmmforcefields [
https://github.com/openmm/openmmforcefields] package, which aims to enable
users in the OpenMM ecosystem to use Amber force fields. Part of the
functionality we support is to enable users to assign small molecule GAFF
parameters via antechamber using AmberTools under the hood.

Since there have been multiple releases (GAFF 1.4, 1.7, 1.8, 1.81, 2.1,
2.11, and now 2.2.20), we aim to enable users to specify the GAFF version
they want to assign, and ensure they can generate reproducible parameters
with the latest version of AmberTools.

Previously, I had recovered all versions of gaff.dat that had been
distributed with previous AmberTools releases (confirming with Junmei that
my list was complete), and enabled users to select whichever version they
wanted to use (gaff-1.4, 1.7, 1.8, 1.81, 2.1, 2.11). Under the hood, we
drove antechamber and prmchk2 (from the latest installed AmberTools) with
pointers to the specific version of gaff.dat that was selected:

antechamber -i molecule.sdf -fi sdf -o molecule.mol2 -fo mol2 -at
gaff${GAFF_MAJOR_VERSION}
parmchk2 -i molecule.mol2 -f mol2 -p gaff-${GAFF_VERSION}.dat -o
molecule.frcmod

For AmberTools 19.11 through 22.5, this provided stable parameters. Now,
with the release of AmberTools23, these commands generate *different
*parameters
even when fed gaff-2.11.

This behavior appears to be due to the new GAFF 2.2.20 including an updated
ATOMTYPE_GFF2.DEF which includes new atom types.

The good news is that parmchk2 allows *both *the -p <parmfile> and -c
<PARMCHK.DAT> to be specified to enable its behavior to be defined in terms
of paths to specific versions of these files. We can simply point to the
appropriate historical version in this case.

The *bad* news is that antechamber does not appear to have the same
flexibility: the argument -at <atom type> only seems to accept a token like
gaff2 instead of a file path to an atom type definition file, and sets
hard-coded paths based on this token.

Am I missing a hidden option within ambertools that would allow us to
specify the specific paths to files that define the atom types it should
use? If not, would it be possible to release a bugfix version that includes
the ability to override the specific file paths set by the -at parameter?
This would enable us to generate reproducible parameter sets even when the
default versions of atom types are updated with GAFF.

Thanks!

John

-- 
John D. Chodera
Member, Computational and Systems Biology Program
Memorial Sloan Kettering Cancer Center
email: john.chodera.choderalab.org
book a meeting: https://calendly.com/john-chodera
office: 646.888.3400
mobile: 415.867.7384
url: http://www.choderalab.org
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jul 04 2023 - 09:00:02 PDT
Custom Search