From: JC Gumbart (
Date: Mon Jul 06 2020 - 18:57:17 CDT

What was the initial penalty with CGenFF parameters? I assume higher? So why not use the FFTK ones?

I agree though that if it remains in the QM region, the force-field parameters shouldn’t matter (never having run QM/MM myself, that is).


> On Jul 6, 2020, at 12:34 PM, Ritu Arora <> wrote:
> Hi,
> I am trying to parameterize ff for s-nitrosocysteine using vmd-fftk. The purpose of doing this is to simulate the protein of my interest, which undergoes s-nitrosation at cysteine residue, to find an initial structure for QM/MM calculations. FYI, x-ray crystal structure of the modified protein is not available. I understand that I cannot generate accurate ff for this molecule because of the complexity in its electronic structure (I mean the multireference character associated with RSNO moiety which can be largely affected by its protein environment).
> CGenff provides me ff with really high penalty scores ~500. So, I thought of using fftk to gain atleast some accuracy. I am right now optimizing bond and angle parameters for EtSNO model, but Current Final Obj. (~70 to 80) value has not converged even after 10 iterations. So, wondered if it is because of the electronic structure of -SNO moiety. How should I proceed?
> Do you think that it will be fine to use CGenFF parameters (with such high penalty scores) because ultimately I will be considering that region in my QM part of QM/MM calculations?
> I look forward to your reply.
> Thank you.
> Ritu