Re: [AMBER] Selection of forcefield for protein-ligand simulations

From: Thomas Cheatham <>
Date: Sat, 27 Mar 2021 00:52:47 +0000

Multiple simulations is always better than a single run. A nice investigation is:

As far as united atom vs. all-atom, with gas phase or implicit solvent united atom force fields reduced the number of atoms and sped up the calculations - however now with routine use of explicit solvation and particle mesh Ewald, there is little speed-up through application of united atom force fields vs. atomistic and development of united atom force fields has slowed. Coarse-grained models are another possibility, but I doubt the reliability of their application to protein-ligand systems. Short simulations can allow comparisons between ligands; longer ones provide greater sampling (including more movement of protein and ligand) and can make simple comparisons harder or muddier until you get to sufficient time scales (microseconds - ms) to sample the full distribution. Our group has done this for a variety of nucleic acid systems, including those with bound ligands.


p.s. with 3 independent simulations use different random seeds (ig= -1) and our group also randomizes the ion distribution (to be different in each independent copy).
From: Prasanth G, Research Scholar <>
Sent: Friday, March 26, 2021 6:24:26 PM
To: AMBER Mailing List
Subject: [AMBER] Selection of forcefield for protein-ligand simulations

Dear sir,
Thank you for the detailed response. It was very insightful and helped me
get a better understanding on the reason behind choosing all atom
forcefields for protein-ligand simulations.
I have a follow up question on the same lines..
I wanted to know your opinion on the need to carry out protein-ligand MD
simulations in triplicates. From the literature, I could gather that
running Monte Carlo simulations in triplicates will make more sense, as it
is stochastic. In my humble opinion the classical MD simulations are fairly
deterministic and also as you had clearly pointed out, Protein-ligand
simulations are run for shorter durations during which, the probability of
the complex moving towards the nearest stable configuration on the free
energy landscape is quiet high. In other words, the complex gets more or
less stuck in a minima.
Do you feel this explanation could justify running the MD simulations as a
single run?

Thank you,
Prasanth G.
AMBER mailing list

AMBER mailing list
Received on Fri Mar 26 2021 - 18:00:02 PDT
Custom Search