[gmx-users] Umbrella sampling - regd

[Justin Lemkul]

On 11/26/12 1:03 AM, ramesh cheerla wrote:
> Dear Gromacs users,
>
>
>       I am calculating  PMF for the ion permeation through a tunnel, using
> umbrella sampling. In my system I have some binding sites with ions, I have
> removed some ions from binding sites before pulling and In pulling
> simulations have pullaed pulled the adjacent ions to the vacant binding
> site (tunnel is aligned along  Y- axis).
> for this I have used the following options in .mdp file.
>     ; Pull code
> pull            = umbrella
> pull_geometry   = distance
> pull_dim        = N Y N
> pull_start      = yes
> pull_ngroups    = 1
> pull_group0     = REFA
> pull_rate1      = 0.01
> pull_k1         = 3000      ; kJ mol^-1 nm^-2
>
> Here  pull_group0 = REFA is the ion at another binding site and
> pull_group1     = ATBP  is the atom  that I am pulling to the vacant site.
> In next step I have performed umbrella sampling simulations to the selected
> configurations using the following options :
> ; Pull code
> pull            = umbrella
> pull_geometry   = distance  ; simple distance increase
> pull_dim        = N Y N
> pull_start      = yes       ; define initial COM distance > 0
> pull_ngroups    = 1
> pull_group0     = REFA
> pull_group1     = ATBP
> pull_init1      = 0
> pull_rate1      = 0.0      ; 0.01 nm per ps = 10 nm per ns
> pull_k1         = 3000      ; kJ mol^-1 nm^-2
> pull_nstxout    = 100
> pull_nstfout    = 100
> After that I have constructed PMF profile using Gromacs tool "g_wham",
> while using g_wham I am getting the following warnings,
> "WARNING, no data point in bin 34 (z=0.889382) !
> You may not get a reasonable profile. Check your histograms!"
> I have checked my histograms instead of one histogram for each
> configuration with perfect overlapping, I am getting only one histogram.

You're probably just plotting the file wrong.

xmgrace -nxy histo.xvg

> Here I am sending the link that containing PMF profile and histogram that
> have obtained.
> http://researchweb.iiit.ac.in/~bipin.singh/pmf.jpg
> http://researchweb.iiit.ac.in/~bipin.singh/hist.jpg
>

The PMF suggests massive undersampling (or complete lack of sampling) in several 
areas.

> the weired thing that I have observed in my pulling simulations is that I
> have expected motion of ion in positive Y-direction for my pull parameters
> but it is moving in negative Y- direction.
>   Can anybody please suggest me a solution for this,  Am I following correct
> protocol to get PMF of my system, is there any better method that suits for
> my system.
>

If you are trying to see a complete translocation of an ion through a channel, 
the "distance" geometry is inappropriate, as it does not deal correctly with the 
change of sign for the vector between the reference and pulled group, IIRC.  Try 
"position" geometry and associated settings.

-Justin

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

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

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