AMBER Archive (2009)

Subject: [AMBER] extra torsion terms in

From: Astrid Maaß (
Date: Thu Feb 26 2009 - 08:32:30 CST

Dear Amber developers,

Recently I encountered some unexpected behaviour in tleap. I am
interested in understanding torsion rotations and for my test compound
ethanediol I noticed that in the topology file created by tleap the
oh-c3-c3-h1 torsions are differently assigned despite their chemical

In detail, I am using a prep file created by antechamber and using GAFF
to model the molecule (via AmberTools 1.2 with applied bugfixes.all from
the date of yesterday). Tleap does the processing without any
complaints. In the topology file the first two times that the
oh-c3-c3-h1 patterns show up, the force constant, periodicity and phase
shift exactly correspond to the entry in gaff.dat - i.e. 0.25 kcal/mol,
0 and 1. The third and fourth time however, the torsion is described not
only by the above parameters, but also by additional numbers: 0.1555
kcal/mol, 0 and 3.

I suspect these values were somehow inherited from the torsion visited
immediately before (H-C1-C2-H5), which were taken from a wildcard entry
in gaff.dat (X -c3-c3-X). When modifying gaff.dat by adding a single
line that provides the required pattern of atom types explicitly
(h1-c3-c3-h1) with the same numbers as the wildcart entry, the topology
file happens not to show the extra torsional terms.

Attached to this mail is all the input I used to create the different
topology files together with the output by running
"~/tools/amber10/exe/tleap -f input/leapin_orig_gaff " and
"~/tools/amber10/exe/tleap -f input/leapin_modif_gaff "; maybe someone
could check, please?

I would appreciate any help in this matter -- many thanks in advance.

Best regards
Astrid Maaß

AMBER mailing list