[gmx-users] request for dihedral PMF test system or complete alanine dipeptide topology file

[David Mobley]

Chris,

> I have included my .top file. The relevant section is in [ dihedral_restraints ]
> This file is a template that I use like this:
> sed 's/PHI/${phi_value}/' template.top | sed 's/PSI/${psi_value}/' > phi${phi_value}_psi${psi_value}.top

Just out of curiousity, which GROMACS version are you using? The
dihedral restraints were sort of broken in 3.3 originally; this is
fixed in 3.3.1.

> I read in the gromacs manual 3.3 page 53 "Note that, since it is harmonic, periodicity is not taken into account,
> so it is best to define improper dihedrals to have a E0 as far away from ±180deg as you can manage."

Right. I remember looking into this at one point, and I don't think
the manual is correct; it looks to me like dihres.c handles
periodicity properly. In particular, there is the followign comment in
dihres.c:

/* phi can jump if phi0 is close to Pi/-Pi, which will cause huge
     * force changes if we just apply a normal harmonic.
     * Instead, we first calculate phi-phi0 and take it modulo (-Pi,Pi).
     * This means we will never have the periodicity problem, unless
     * the dihedral is Pi away from phiO, which is very unlikely due to
     * the potential.
     */

Consequently, I have done PMF's with phi0=180 before with no problems,
leading me to suspect this is indeed working fine. :) Of course, it's
worth taking note of the comment there about the dihedral being pi
away from Phi0. This means that your restraint forces need to be
reasonably large or you may have potential periodicity problems. This
has never been an issue for me.

> ; Include topologies
> #include "/home/cneale/gromacs/proteintop/dialanine.itp"
> [ dihedral_restraints ]
> ; ai   aj    ak    al  type  label  phi  dphi  kfac  power
> ; phi C'(n-1) - N - CA - C'
>    5    7     9    15     1      1  PHI     0     1      2
> ; psi N - CA - C' - N(n+1)
>    7    9    15    17     1      1  PSI     0     1      2

I assume you're clear on the fact that the phi and dphi are in degrees
but the force constants aren't? And your units are otherwise correct?
(i.e. you're remembering output is in kJ/mol?)

Other than those two things, I'm not too sure what to say. The system
I've done this in is a sidechain dihedral in a big protein system. I
can send you topologies if you want, but it takes rather a lot of
sampling to get the PMF right there so it's less than ideal as a test
system.

David