Re: [AMBER] if the ligand doesn't bind when it should, what does it say about quality of my simulations and what can I do?

From: Wesley Michael Botello-Smith <>
Date: Tue, 29 Oct 2019 14:37:07 -0700

I don't know if you checked already or not, but did you make sure that all
side chains are in their appropriate protonation state for the given
conformation? I.e. did you run something like prop-pka or a constant pH
simulation to test this?
If a nearby residue is in the wrong protonation state, that could make a
big difference. For instance if you have histidines near the binding site
it is possible that they may not be in the standard neutral protonation
state when CAMP is present.

I think your third idea (using NMR restraints) is probably the next thing
you should check before fiddling with the temperature or langevin damping
Using NMR or Collective Variable restraints to keep a ligand in place is
fairly common practice. At the very least, its a good idea during heating,
density, and equilibration stages. I.e. Keep it restrained while you
release the backbone restraints that were used during heating and density
and run a short (nanosecond or two) equilibration stage with the restraint
in place. Then slowly release the restraint. This way you can make sure
that its not getting jarred loose while the protein is acclimating to the
simulation environment.

On Tue, Oct 29, 2019 at 7:13 AM Homeo Morphism <>

> I'm modelling CAMP binding to the CAMP-binding domain in one of the
> appropriate receptors. The relative coordinates of superficially bound CAMP
> are available from both the crystallographic structures and the results of
> docking in Autodock Vina -- they can be said to agree with each other.
> But when I start simulating the complex in Amber, CAMP molecule
> disassociates from the domain unnaturally quick and often. Even when it
> doesn't, I don't see in CAMP-binding domain even a glimpse of the
> conformational transformations that CAMP is described to cause --
> reorientation of side-chains, etc.
> To help with the disassociation problem, I've tried to disable SHAKE and
> for a while go to 1fs-timestep, hoping that once the pose is refined I
> could revert to 2 fs. 1 fs seems to help, but once I revert to 2 fs, it
> comes back.
> I've also tried increasing cutoff from the often-advised value of 9A to 13A
> -- doesn't seem to help.
> The rest of parameters are pretty standard. The periodic box is large
> enough. There's ions added for neutralization purposes. Etc. etc. etc.
> I realize that diagnosing this remotely is impossible, but perhaps there's
> a typical set of mistakes that people do in experiments like this...
> I've been also thinking of:
> 1) lowering the temperature from 310K to 300K;
> 2) increasing the viscosity/damping coefficient in the Langevin;
> 3) using NMR-restraints to force the ligand to hold on to the receptor.
> Does any of these sound plausible? And what else can be done?
> Thanks.
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Tue Oct 29 2019 - 15:00:02 PDT
Custom Search