[gmx-users] Torsional potential parametrization

[Mark Abraham]

Jarvist Frost wrote:
> Dear Gromacs users,
> 
> I'm interested the parametrisation of a torsional angle between successive monomer units - as briefly described on page 13 of Berk Hess' "Molecular dynamics simulations & Gromacs introduction" slides from the 2007 Gromacs Workshop:
> 
> https://extras.csc.fi/chem/courses/gmx2007/Berk_talks/forcef.pdf
> 
> As such, we have completed energy scans of a dimer in 'Gaussian' at a suitable level of theory.
> 
> Berk's slide then says to use 'mdrun -rerun with the optimized quantum structures' to get the 'Forcefield' view of the total energies.
> 
> At this point I'm unsure of what is suggested:
> Do I first do an MD run of a single dimer in vacuum (or very large box with pbc) and then use that trajectory, sampling total energies at the torsional angles which it goes through?
> Or do I input coordinates of each Gaussian scan structure individually and somehow position restrain the torsional value, and then do a single-point energy calculation somehow?

I surmise that Berk's suggestion is a shortcut version of the latter. 
'mdrun -rerun' takes an existing trajectory and recalculates energies 
and forces and writes new output, but it doesn't do the integration 
necessary to actually follow the time evolution - it just takes the next 
structure from the trajectory. You can readily concatenate PDB files 
into a combined file that gromacs tools will treat as a trajectory, so 
you take your Gaussian structures, convert to PDB, concatenate and do 
Berk's suggestion to get the "'Forcefield' view of the total energies.

Mark