Atom order for improper dihedrals.

[lohedges]

Atom order is important in the definition of improper dihedral parameters. In order to be able to convert between formats we need to be consistent in how they are loaded in to Sire, or provide a means of flagging the central atom (the atom that is planar, with the three atoms around it governing its planarity).

From what I can tell, atom ordering for the three main forcefields that we consider is as follows:

AMBER:
The central atom is third, with the other atoms appearing in alphabetical order. See top of page 7 here for details.

GROMACS:
The central atom is first. Unsure of the ordering of other atoms. See here for details.

CHARMM:
The central atom is first. Unsure of the ordering of other atoms. See here for details.

Obviously there will still be ambiguity in the ordering of the three other atoms, but it would be good to fix the position of the one atom that we do know about.

How does Sire deal with this issue internally?

(Some discussions of dealing with this problem for the CHARMM forcefield in OpenMM can be found here and here.)

Cheers.

[jmichel80]

Hi Lester, That's a good point. I'm not convinced the rules for ordering atoms around impropers are consistently applied in various third party tools. In FESetup I made no assumptions about the order in which atoms are listed around an improper. This leads to cumbersome code to find equivalent impropers between two molecules (since all possible orderings have to be checked). You can see how I dealt with this here between line 897-1005 https://github.com/halx/FESetup/blob/master/mutate/topol/pertfile.py I think Sire just check if two improper have atoms listed in the same order. There is code in SireMol::Dihedralid to check if a dihedral matches a reference dihedral or its mirror (swapping at0 and at3). Maybe you can add a more general 'equivalent' method to SireMol::ImproperID that is similar to what is done in FESetup to decide whether two impropers are equivalent. This would be useful to compare topologies with impropers defined with a different order. In principle the energy should be independent of the ordering so you could retain the ordering as found in an input topology on reading, but if writing a topology from scratch you could use a consistent convention across topologies. Best wishes, Julien

-------------------------------------------------------------- Dr. Julien Michel, Royal Society University Research Fellow Room 263, School of Chemistry Joseph Black Building, University of Edinburgh David Brewster Road Edinburgh EH9 3FJ United Kingdom phone: +44 (0)131 650 4797 http://www.julienmichel.net/ -------------------------------------------------------------

On Wed, Mar 7, 2018 at 10:13 AM, Lester Hedges ***@***.***> wrote: Atom order is important in the definition of improper dihedral parameters. In order to be able to convert between formats we need to be consistent in how they are loaded in to Sire, or provide a means of flagging the central atom (the atom that is planar, with the three atoms around it governing its planarity). From what I can tell, atom ordering for the three main forcefields that we consider is as follows: AMBER: The central atom is third, with the other atoms appearing in alphabetical order. See top of page 7 here <http://alma.karlov.mff.cuni.cz/bio/99_Studenti/00_Dalsi/ParamFit/2013_ParamFit_AmberTools13.pdf> for details. GROMACS: The central atom is first. Unsure of the ordering of other atoms. See here <https://mailman-1.sys.kth.se/pipermail/gromacs.org_gmx-users/2013-April/080820.html> for details. CHARMM: The central atom is first. Unsure of the ordering of other atoms. See here <https://www.charmm.org/ubbthreads/ubbthreads.php?ubb=showflat&Number=25342> for details. Obviously there will still be ambiguity in the ordering of the three other atoms, but it would be good to fix the position of the one atom that we do know about. How does Sire deal with this issue internally? (Some discussions of dealing with this problem for the CHARMM forcefield in OpenMM can be found here <openmm/openmm#220> and here <openmm/openmmforcefields#33>.) Cheers. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#193>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ALLd5N8NYa2G9m8wrYOWDlcrWBlN5gdGks5tb7LmgaJpZM4SgL0O> .

The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

[chryswoods]

Closing as no new updates