Re: [AMBER] Combining GAFF and GLYCAM force fields to describe a non-natural sugar

From: David Cerutti <dscerutti.gmail.com>
Date: Wed, 28 Nov 2018 13:53:58 -0500

This email is in no way intended to impede workplace productivity, create
environmental hazards, or disseminate political disinformation. If you
believe you have received this email in error, please understand it is
intended specifically for you. You. The contents of this mail have been
checked by the author, but no guarantees are made as to its authenticity
and no assurances are given as to the physical well being of any subjects
discussed herein. Any grammatical errors are the responsibility of the
author, except for instances of purple prose or the passive voice being
used which are purely the products of the whimsically delightful Grammarly
program with extreme settings in Preferences >> Advanced.

Center for Force Field Research Excellence, Rutgers University
* * * * *

The way I'd approach this sugars problem is to take the base parameters
from GLYCAM and try to graft in the fluorine chemistry. GAFF isn't the
tool here, I think--it is for making something from scratch, whereas you
want something that looks like an established parameter set for related
molecules. What I'd do is make your residue, assign charges based on
GLYCAM manually (use your best judgment), then take the mdgx program in
AmberTools. Look at these tutorials:

http://ambermd.org/tutorials/advanced/tutorial28/index.php
http://ambermd.org/tutorials/advanced/tutorial32/index.php

The first one is about "IPolQ" charges which you don't want if you are
trying to mimic GLYCAM--you want HF/6-31G* or HF/CC-PVTZ. The thing to do
is run mdgx -FITQ to see a list of options for charge fitting, then make a
bunch of configurations and run HF/(your basis set) on them to get charge
grids. mdgx will be able to take those charge grids and run a REsP-like
fit, and you can specify that it hold charges other than the fluorine and
proximal atoms to values from the original GLYCAM for the sugars that
you've decided are the closest match. Then, you can make a bunch of other
configurations by manipulating the fluorine group to sample dihedral
configurations, and refit parameters that impinge on the motion of that
group (I'd recommend angles as well as dihedrals).

This is the "hard core" way to do it, but I made mdgx to make the hardcore
stuff as straightforward as possible. It's mechanistically easier to take
mdgx and make a completely new set of parameters for your molecule, but
it'll take more computational power and the details of telling mdgx to
respect certain parameters from GLYCAM are not that difficult.

Dave
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 28 2018 - 11:00:03 PST
Custom Search