Re: Some Basic queries for FEP and TI

From: Brian Radak (
Date: Sun Mar 21 2021 - 14:41:11 CDT

1) Using the whole protein is almost always technically more correct, but
makes sampling more expensive and harder to converge (potentially
unattainable on such large a system). Unless you think there are important
conformational changes involved, I would consider a specialized strategy
for your quantity of interest.

2) Charge changing perturbations do cause a shift in dG values due to a
change in the zero of energy arising from PME. However, the shift vanishes
when the periodic box is large and the mean dielectric is high (eg if the
system is mostly water). I would look up the work of Rocklin, et al. on
using continuum corrections, but other strategies may also work.

Also, divalent ion parameters in fixed charge force fields are notoriously
poor. You generally either get good structure or good energetics - rarely
both. I would validate your model in someway first, bc the results are
unlikely to just work out nicely on your first try.

Hope that helps,

On Sun, Mar 21, 2021, 12:21 PM Debadrita Basu <>

> Hello Everyone,
> My protein of interest has ~850 residues (quite big) where a Magnesium
> ion is intrinsically bound. Also some ligands are known to bind to some
> specific grooves of this protein. I wanted to see the role of Magnesium on
> the ligand binding (ie whether it has a facilitating or inhibiting role).
> So, I decided to carry out the FEP calculation where I discharge the
> Magnesium atom to a dummy Magnesium atom, (in two different systems: ligand
> bound and apo) followed by thermodynamic integration. I used CGENFF to
> parameterize.
> I decided to follow the FEP algorithm implemented in NAMD, but I have a
> couple of questions in my mind:
> 1. Whether I will get a true result by using the whole protein.
> 2. While discharging the Magnesium atom I will end up changing the total
> charge of the system, is that acceptable? If not, how can it be solved?
> Hope, I get something to learn from your answers.
> Thank You.
> --
> Debadrita Basu.
> Senior Research Fellow
> Bose Institute
> Kolkata, India.

This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:11 CST