[gmx-users] Pulling Mechanics

[Justin Lemkul]

On 1/13/17 6:37 PM, Alexander Yang wrote:
> Hi everyone,
>
> I am pulling a water molecule into a bilayer using an absolute reference
> (mdp file below). I have tried to adapt methodology in Justin Lemkul's
> umbrella sampling tutorial, but I have encountered a couple issues:
>

If you're trying to get water to move into a bilayer, why are you using an 
absolute restraint?  Intuitively, the groups should be the bilayer and the water.

> 1) The bilayer (already well-equilibrated) is either being pulled or
> drifting. Without the pull code, the bilayer does not move. I've defined
> center of mass motion removal for the non-waters, which I would expect to
> keep the bilayer still. I've copied the mdp file below (Tracer2117
> corresponds to 2117SOL, but I've defined the Tracer2117 group in the index
> file as the 3 atoms that correspond to this water)
>

Applying COM motion removal to one group and not another is highly artificial. 
Couple that with your use of an absolute restraint (also somewhat artificial) 
and you're layering bad things upon bad things.

-Justin

> 2) I'm getting an error that the distance between the pull groups is larger
> than 0.49 times the box size (the box is about 6.12 nm x 5.17 nm x 7.42
> nm). The origin of my absolute reference is only 0.1 nm away from the water
> molecule. The way I've specified my pull-coord-rate, at the timestep when
> the pull distance is larger than half the box size, the distance between
> the origin of the reference at time 0 and the water molecule at time of
> crash exceeds half the box size. Is the reference position actually moving,
> or does the pull-coord-rate specify a change to a quantity besides the
> reference position such that the water molecule can be pulled while the
> reference position stays where it originally was specified?
> @ time 0 ns: Water (3.794, 1.447, 1.150), reference origin (3.794, 1.447,
> 1.250)
> @ crash (time 47.19ns): Water(3.997, 0.032, 4.751). My pull-coord-rate is
> 7.7e-5 nm/ps, so would the reference be at (3.794, 1.447, 4.883)? If this
> is the case, I'm not sure how this distance violates the box size rule.
> However, the distance between reference @ t=0 and water @ t = 47.19ns,
> violates the 0.49 box size rule.
>
>
>
> ******* Mdp File ***********
>
>  Run MDP parameters
>
> integrator                = md
>
> dt                        = 0.002
>
> nsteps                    = 25000000
>
> comm-mode                 = Linear
>
> nstcomm                   = 1
>
> comm-grps                 = non-water
>
>
> ; Output parameters
>
> nstxout                   = 0
>
> nstvout                   = 0
>
> nstxtcout                 = 5000
>
> nstenergy                 = 5000
>
> nstlog                    = 5000
>
> nstfout                   = 0
>
>
> ; Bond parameters
>
> continuation              = yes
>
> constraint-algorithm      = lincs
>
> constraints               = all-bonds
>
> lincs-iter                = 1
>
> lincs-order               = 4
>
>
> ; Neighbor searching
>
> cutoff-scheme             = Verlet
>
> nstlist                   = 10
>
> rcoulomb                  = 1.4
>
> rvdw                      = 1.4
>
>
> ; Electrostatics
>
> coulombtype               = PME
>
> fourierspacing            = 0.16
>
> pme_order                 = 4
>
>
> ; Temperature coupling
>
> tcoupl                    = nose-hoover
>
> tc_grps                   = non-water   water
>
> tau_t                     = 0.4         0.4
>
> ref_t                     = 305         305
>
>
> ; Pressure coupling
>
> pcoupl                    = Parrinello-Rahman
>
> pcoupltype                = isotropic
>
> tau_p                     = 2.0
>
> ref_p                     = 1.0
>
> compressibility           = 4.5e-05
>
> refcoord_scaling          = com
>
>
> ; Misc stuff
>
> gen_vel                   = no
>
> pbc                       = xyz
>
> DispCorr                  = EnerPres
>
>
> ; Pull parameters
>
> pull                      = yes
>
> pull-nstxout              = 5000
>
> pull-nstfout              = 5000
>
> pull-ngroups              = 1
>
> pull-ncoords              = 1
>
> pull-group1-name          = Tracer2117
>
> pull-coord1-groups        = 0 1
>
> pull-coord1-type          = umbrella
>
> pull-coord1-geometry      = direction
>
> pull-coord1-origin        = 3.794    1.447    1.250
>
> pull-coord1-dim           = N N Y
>
> pull-coord1-rate          = 7.7e-05
>
> pull-coord1-vec = 0 0 1
>
> pull-coord1-k             = 500.0
>
> pull-coord1-start         = no
>
> *************
>
> I've attached URLS to snapshots below (time0 and time47). Red is water,
> grey is the lipid bilayer, yellow is the water molecule (enlarged for
> visibility), and I've placed axes in the center. Before the crash, the
> water molecule was definitely moving in the correct direction (positive z).
> The bilayer drifted in the direction the water molecule was moving.
>
> (time0) http://imgur.com/a/jWu33
>
> (time47) http://imgur.com/a/nekmL
>
> Thanks for the help,
>
> Alex
>

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

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

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