From: Mayne, Christopher G (cmayne2_at_illinois.edu)
Date: Mon Apr 04 2016 - 10:09:06 CDT

Mohan,

It is best to focus on the PES below ~10 kcal/mol, as this is the region that will be sampled under most simulation conditions; anything above that is considered “high energy” and likely irrelevant. As I said before, without any additional information I would say that your fit looks pretty good overall. I would pay special attention to the scan around frame 250 to try to fix the shape of that potential using the tools I suggested in VMD 1.9.3 alpha.

Regards,
Christopher Mayne


On Mar 31, 2016, at 10:55 PM, Mohan maruthi sena <maruthi.sena_at_gmail.com<mailto:maruthi.sena_at_gmail.com>> wrote:

Thank you, sir, for a reply. I will try with VMD 1.9.3 alpha release and try for a better fit. Please find the attached plot with y-axis visible. Can you please comment on it.


Thanks & Regards,
Mohan

On Thu, Mar 31, 2016 at 8:35 PM, Mayne, Christopher G <cmayne2_at_illinois.edu<mailto:cmayne2_at_illinois.edu>> wrote:
Mohan,

By in large, your fit looks quite good. Without the y-axis, it’s a little hard to tell how much the inconsistencies matter. You shouldn’t expect the MM PES to match the QM *exactly*, and depending on the complexity of your molecular system, might actually be a sign of overfitting. The only part the concerns me a bit is the second to last set of peaks, in which the shape appears to be different.

In response to your specific questions:

1) Yes! I came across this same problem and built a module that allows you to load in the Gaussian log files, which are parsed for both the energy and the coordinates. The plugin then shows an interactive plot that is color coded by the dihedral scan. As you move through the coordinates in the VMD window, and indicator will show you where you are on the PES, and the dihedral that was scanned is shown in a color-coded licorice representation. This allows you to analyze the QM PES with respect to the structure.

This tools is in the 1.9.3 alpha releases, under the Scan Torsion tab just below the Gaussian Settings via the “Open Torsion Explorer” button.

2) By default ffTK excludes torsions terminating in hydrogens from the scan. If you want to explicitly scan these torsions, e.g., for polar hydrogens, then you can add them to the “Dihedrals to Scan” box. During the fitting, however, all dihedrals provided in the “Dihedral Parameter Settings” will be fit, and should include dihedrals that terminate with hydrogens.


Regards,
Christopher Mayne

> On Mar 30, 2016, at 9:37 AM, Mohan maruthi sena <maruthi.sena_at_gmail.com<mailto:maruthi.sena_at_gmail.com>> wrote:
>
> Hi all,
> I am trying to generate force field parameters for a molecule which contains 41 atoms. Except dihedral parameters, I am able to generate charges, bonded parameters. I am trying to fit dihedrals and tried to refine it but still the fit is not perfect. Please find the attached torsional profiles. My queries are as follows:
>
> 1) Is there any way to identify individual dihedrals contribution by looking at entire dihedral fitting profile?
> 2) FFTK does not consider dihedrals containing hydrogens, then how does it generate dihedral parameters for those torsionals?
>
>
> Your help will be highly appreciated.
>
>
>
>
> --
> Thanks and Regards,
> Mohan Maruthi
> <test_ml.png>





--
Thanks and Regards,
Mohan Maruthi
<test1.png>