Re: [AMBER] Ligand is unstable in simulations

From: David A Case <>
Date: Mon, 4 Oct 2021 10:06:09 -0400

On Mon, Oct 04, 2021, Midhun K Madhu wrote:
>In my simulations of a lipid-protein system using
>Charmm36/Charmm36m forcefield, The ligand is becoming unstable and slowly
>coming out of the pocket. Originally, this should be a milliseconds
>timescale phenomenon and shouldn't happen within a few hundreds of
>nanoseconds, like in my case.
>Earlier when I tried accelerated MD (GaMD to be precise), the ligand
>completely came out, almost immediately, and that is why I tested the
>system using conventional MD. Can someone help me to understand what is
>going wrong in the simulations? The system was prepared using CHARMM GUI
>and the parameters for the ligand were assigned using the Paramchem server.

My experience has been that protein conformational changes sometimes
(often?) occur on simulation time scales that are faster than those observed
experimentally. I don't have any experience with lipid-protein systems, so
maybe others on the list can comment.

You might make sure that you have done extensive equilibration of the bound
state in the presence of restraints, so that the initial state is
well-adjusted to the force field you are using. Carrying out 100 ns of
restrained simulations (perhaps with slowly decreasing the weight of the
restraint) would seem appropriate. Also consider how well you know what the
bound configuration really is. If you were missing (say) a single
well-formed hydrogen bond, that could have a big effect on the simulated
residence times.

Also note that seeing a single ligand escape at hundreds of nanoseconds
doesn't directly tell you too much about the actual rate; although if your
time scale is "off" by four orders of magnitude there may indeed be something


AMBER mailing list
Received on Mon Oct 04 2021 - 07:30:02 PDT
Custom Search