[gmx-users] COMPASS force field revisited Bond Dihedral and Angle Dihedral cross terms

David van der Spoel spoel at xray.bmc.uu.se
Wed Feb 22 12:00:46 CET 2012

On 2012-02-21 23:26, Broadbent, Richard wrote:
> On 21/02/2012 19:02, "David van der Spoel"<spoel at xray.bmc.uu.se>  wrote:
>> On 2/21/12 6:17 PM, Richard Broadbent wrote:
>>> Dear All,
>>> I am considering conducting a simulation of a polymeric system in
>>> gromacs. I would like to use the COMPASS forcefield as it has a complete
>>> parameter set for my molecule.
>>> I believe the majority of the implementation is simple (though long and
>>> fiddly). However, it has Bond-Dihedral and Angle-Dihedral cross terms.
>>> These terms are not listed in either Chapter 4 or 5 of the Manual. Which
>>> leads me to believe that they are not implemented. However CMAP
>>> dihedrals are not documented in chapters 4&   5 and they are implemented
>>> in the code. I therefore had a quick look in src/gmxlib/ifunc.c and
>>> couldn't see those terms. However, my knowledge of the layout of the
>>> source code is extremely limited so I could be looking in the wrong
>>> place, and thought it best to ask.
>>> Are Bond-Dihedral and Angle-Dihedral cross terms available in gromacs
>>> 4.5.5?
>> What do the functions look like?
> Bond dihedral:
> (b-b0)*[k_1*cos(phi) +k_2*cos(2*phi)+k_3*cos(3*phi)]
> Angle Dihedral:
> (theta -theta_0)*[k_1*cos(phi) +k_2*cos(2*phi)+k_3*cos(3*phi)]
> There is also angle-angle-dihedral:
> k*(theta -theta_0)*(theta' -theta_0')*cos(phi)

For this to work, the best would be to implement a new dihedral 
function. Due to the way this works the function would need quite a few 
parameters though.

atoms i,j,k,l
3 b0 for i,j and j,k and k,l
2 theta_0 for i,j,k and j,k,l
k_1, k_2, k_3 (shouldn't there be a k_0)?

In all 8-9 parameters! What also is unfortunate is that a number of 
these parameters are most likely repeats (b0, theta_0) from other lists. 
One could of course combine bonds, angles and dihedrals into one monster 
routine, but then dealing with constraints becomes somewhat harder again.

By the way this thread now should be moved to the developer list I guess.
>> Most likely they are not implemented though if they're not in the manual.
>>> Thanks,
>>> Richard

David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se

More information about the gromacs.org_gmx-users mailing list