[gmx-users] Non-overlapping region appears in histograms when using SMD+Umbrella Sampling for strong binding complex

[Justin Lemkul]

On 4/20/17 2:30 PM, Jason Zhu wrote:
> Dear All,
>
> I am modeling the binding free energy between a molecule of sodium cholate
> and a 2D nanosheet by using steered MD and umbrella sampling as guided by
> Gromacs Tutorial 3 (
> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/umbrella/index.html
> ).
>
> The small molecule has a strong binding with the surface of nanosheet. I
> first performed a Steered MD to pull the molecule away from the surface.
> The loading rate is low as 0.01 nm/ns, the force constant for pulling is
> 1000 kJ/nm2 and the output frequency of configuration is 1 frame very 100
> ps.
>
> However, due to the strong binding, the COM distance between the molecule
> and nanosheet has a big jump (~0.4 nm) at the detachment. I cannot extract
> successive sufficiently configurations at the moment of detachment for
> umbrella sampling (10 ns sampling). Even if I choose every configurations
> near the detachment for sampling, the histograms still have a wide gap and
> the curve in the energy profile is noncontinuous.
>
> I have tried to increase the force constant for pulling and even decreased
> the loading rate. But the problem still exists.
>
> I wonder if you have any experience to solve this problem. It would be
> appreciated if you could provide any suggestion.
>

Force reasonable starting structures by taking the closest snapshot you have to 
the desired distance, set the reference distance to the one you want, set the 
pull rate to zero, and just let the system equilibrate under the biasing 
potential.  After a bit of time, you should have a reasonable starting structure.

> Should I use the other method for free energy calculation as in the Gromacs
> Tutorial 6 (
> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/free_energy/index.htm),
> instead of using SMD and umbrella sampling? But there are also some
> questions when I use this method. First, the molecule is negatively
> charged. When I uncouple the electrostatic interactions, should I uncouple
> the sodium ions? If so, I am calculating the free energy of the molecule
> and its neutralizing ion, not only the molecule. Is it correct? If I am
> only interested in the molecule, what should I do? In this method, unlike
> the umbrella sampling, we only can get the free energy change, not the
> energy curve that includes the information of energy barrier. Is it correct?
>

You're unlikely to get good convergence with this approach.  SMD/PMF is the way 
to go.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================