[gmx-users] bond, angle and dihedral restraints in Gromacs

[Mark Abraham]

Rebeca García Fandiño wrote:
> Thank you very much for the suggestion.
>  Actually, I am trying to reproduce the results from another publication 
> (JACS 2005, 127, 7166) http://pubs.acs.org/doi/full/10.1021/ja050044d 
> where they used Gromacs/amber to simulate a CNT. They apply: 
> "Carbon−carbon bond lengths of 0.14 nm and bond angles of 120° were 
> maintained by harmonic potentials with spring constants of 393 960 kJ 
> mol^- ^1 nm^- ^2 and 527 kJ mol^- ^1 deg^- ^2 before relaxation. In 
> addition, a weak dihedral angle potential was applied to bonded carbon 
> atoms.^"
> 
> So, as I understand, they apply RESTRAINTS to the bonds, angles and 
> dihedrals.

Well, they used potential forms similar to those used in restraints. One 
can use a harmonic "bond" potential (function type 6 IIRC) in GROMACS 
without using a distance restraint. See table 5.4.

> I have tried to construct the [distance_restraints] for my system (I 
> show you here a little part):
>
> [ distance_restraints ]
> ; atom1  atom2  type  index  type2  low   up1   up2   fac
> 1 310 1 1 1 0.142 0.146 0.150 1.0
> 2 290 1 2 1 0.142 0.146 0.150 1.0
> 3 291 1 3 1 0.142 0.146 0.150 1.0
> (...)
> 
> but  I don' t understand very well which value should I write for "low   
> up1   up2". The manual says that "the columns low, up1 and up2 hold the 
> values of r0 , r1 and r2 from eqn. 4.76", but looking at this equation I 
> still don' t understand which should be the better values for my case. I 
> have looked into the mail archive, and some other people asked for that, 
> but I did not find any proper solution.
> How could I choose the best values for "low   up1   up2" to restrain my 
> bond values to 1.42 Amstrong?
> 
> The other question is about angle restraints, I have found indications 
> in the wiki of how to treat the dihedral restraints 
> (http://wiki.gromacs.org/index.php/Dihedral_Restraints), however I did 
> not find any similar for the angle restraints. I have seen in the mail 
> archive that 2 pairs of atoms should be used, but I did not find 
> anything to trust in. Could you please give me some indications of how 
> to treat angle restraints?

Section 4.3.2 and Table 5.5 seem to cover it, I think.

Mark