[gmx-users] Info on umbrella sampling simulation

[Matteo Busato]

Dear all,


I'm performing an umbrella sampling simulation on an ion pulled from the bulk of a phase1 to the COM of a phase2 (taken as reference) along the z-axys in a biphasic system to calculate the PMF related to the transfer from phase1 to phase2. The box is 3.62 X 3.62 X 7.24 nm and the interphase is more or less at z = 3.62 nm. This is the pulling part of the .mdp file for the pulling simulation:


; Pull code
pull                    = umbrella
pull-ngroups            = 2
pull-group1_name        = Other
pull-group2_name        = zn
pull-geometry    = distance
pull-coord1-groups      = 1 2
pull-ncoords = 1
pull-dim = N N Y
pull-coord1-rate        = -0.02
pull-coord1-k           = 1000
pull-start       = yes


36 configurations have been chosen to cover a path from the bulk of phase1 (distance with COM of phase2 = 3.4 nm) to the COM of phase2. Each configuration was equilibrated for 5 ns and then simulated for 10 ns in NPT conditions (298.15 K, 1 atm). This is the pulling part of the .mdp file for the NPT simulation of each configuration:


; Pull code
pull                    = umbrella
pull-ngroups            = 2
pull-group1_name        = Other
pull-group2_name        = zn
pull-geometry    = distance
pull-coord1-groups      = 1 2
pull-ncoords = 1
pull-dim = N N Y
pull-coord1-rate        = 0.0
pull-coord1-k           = 5000
pull-start       = yes
refcoord_scaling = com


In addition to the umbrella potential, position restraints of 2000 KJ mol-1 nm-2 on the x and y-axes have been applied to the ion to prevent movement on the xy-plane.


This is the obtained PMF profile:

https://ibb.co/hj9je6


and these are the histograms:

https://ibb.co/cYssXR


I was wandering:


1) why the PMF curve doesn't reach a plateau when the ion is in the middle of the phase1, i.e. for d close to zero. Could it depend on the simulation time or should I build a bigger box so that the ion is sorrounded by a larger amount of phase1?


2) why I get those little jumps on the plateau from d = 2.5 nm to d = 3.4 nm (ion close to the middle of phase 1). The histograms seem to be quite well overlapped.


Thank you in advance for your answers


Kind regards,

Matteo Busato