From: Kenno Vanommeslaeghe (kvanomme_at_rx.umaryland.edu)
Date: Fri Oct 17 2014 - 11:15:18 CDT
On 10/17/2014 04:12 AM, Francesco Pietra wrote:
> Now, I have to study the published notes about CGenFF as to how its output
> can be improved. Through ffTK only?
There do exist other options; it's just that ffTK is a product of this
(VMD/NAMD) community and happens to be the one I'm most familiar with.
GAAMP is an important alternative, being highly automated and aiming to be
CHARM-compatible. I haven't looked into it closely but what I can say is
that the group who developed it is a big authority on CHARMM ion
parametrization.
> Incidentally, how to get the a correct
> residue name instead of
> RESI /scratch 0.000 !
>
> By just manually replacing /scrat?
Yes, that would work. The other way to correct it is by having the correct
molecule name written to the mol2 files. Some softwares allow the user to
set this through their GUI; if not, editing the line after
"@<TRIPOS>MOLECULE" in the mol2 file will do the trick. But the only
advantage of fixing it at the level of the mol2 file is that the comments
with the parameters will be correct, which can also be accomplished with a
find-and-replace, and which doesn't change any aspect of the simulation.
> The problems I alluded to come with, say, non
> heme iron proteins, where even problems of antiferromagnetic coupling
> between irons might be involved.
There you're getting into territory that is not well covered by the
underlying the Class I additive potential energy function. Within that
potential energy function, it is often difficult and sometimes impossible
to get a good model for organometallic species. QM/MM is often a safer
bet. Even the heme model we're talking about is not the most refined
example of force field development, although that hasn't stopped it from
serving many people well.
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:56 CST