Re: [AMBER] mdgx, improper terms and multifit

From: David Cerutti <>
Date: Thu, 22 Aug 2019 13:45:22 -0400

I do not have improper term fitting in mdgx. The major reason is that for
nearly every improper one can define a dihedral that covers the same motion.

mdgx is designed to perform fits with two or more libraries of
conformations (potential energy surfaces) and will correlate the parameters
between the two molecules--they'll be the same variables in the fit. The
key is to have the motions controlled by the same parameters in each
topology file. mdgx will parse all topologies, and require you to give it
a parm##.dat file (i.e. parm10.dat from ff14SB) so that it knows where the
parameters originally came from. This is to help it distinguish between
parameters that merely happen to have the same values in the inputs, and
parameters that are truly distinct. If your nitrogen has a different atom
type in each file it can't be controlled by the same parameters, so you'll
need to go back into antechamber and rederive your models if you want that


On Thu, Aug 22, 2019 at 1:23 PM viktor drobot <> wrote:

> Hello, community! I have two major questions related to mdgx.
> *Question 1.* Is it possible to fit improper terms with mdgx? If not,
> how should I properly handle them? If yes, what should I do in case if
> there are dihedral term with the same naming exists (quite common for
> big molecules)?
> *Question 2.* Also I would like to clarify the following. Let's imagine
> that we have two molecules, the second one is a derivative of the first
> one - for example, quite large amine and its amide derivative. I've
> already calculated quantum energies for both conformational libraries
> and want to use both simultaneously during bonded parameter fit.
> Although the most parts of both molecules are the same (hence
> corresponding atom types), there are some differences - at least atom
> type of acylated nitrogen in amide differs from atom type of amine and
> also there are extra acyl part in amide exist.
> So I want to know if it's possible to:
> 1) perform bonded parameter fit using two conformational libraries
> (amine and amide) simultaneously
> 2) fit bonded terms from amine and amide, which are the same (actually
> belongs to both molecules)
> 3) fit terms which involves amine/amide nitrogen (there will be
> different atom types!)
> 4) fit terms from acyl part of amide
> and do it within single config file for mdgx?
> Actually I can't get how mdgx would proper handle the common nitrogen
> for both molecules but of different atom type...
> Any advices/help would be very appreciable. Thank you!
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Thu Aug 22 2019 - 11:00:01 PDT
Custom Search