[gmx-users] Umbrella sampling

[Justin A. Lemkul]

Nilesh Dhumal wrote:
> Hello
> I am trying to run umbrella sampling for my system. I have system with
> four ion pairs (4 cation and 4 anion) of ionic liquids and 1024 water
> molecules. I am trying to PMF calculation with cation-anion distance as a
> reaction coordinate.  I choose cation as group1 and anion group2.
> I am getting the free energy around ~400 kcal.  Can you tell where I am
> going wrong?
> 

Given this little detail, no one can do anything but speculate.  You haven't 
even said how long your simulations are, or how many windows you've used, or how 
you generated configurations for the sampling.  I could go on.  If you want 
diagnostics, you've got to provide substantial detail about what you're doing 
and how you're doing it.

So let's clarify: you're trying to restrain them all at the same time?  That 
sounds like a recipe for failure.  Presumably these molecules will diffuse 
through the water in all directions, but then with the umbrella restraint you're 
trying to fix the COM distance of each group.

Why not try to restrain one anion and one cation in a box of water, and 
calculate PMF as a function of straight line distance?  What is your objective? 
  If nothing else, a simpler system will tell you if your methods are correct, 
if your simulations are long enough, etc.

> I have one more question, In my equilibrated strcutrue cations and anions
> are far from each other (around 7-8 A). I am more interested in structure
> where cation and anion close (around 2-4A)
> How can I do the PMF simulation to get this structure.

You have to sample in windows that correspond to this spacing, so you have to 
have an input configuration with the desired COM distances.  Umbrella sampling 
will restrain the configurations at a given distance, as specified in the .mdp 
file.  It will not allow significant deviation from this position, unless the 
spring constant is very weak, in which case you're probably undermining the 
purpose of the sampling and need very long simulations.

-Justin

> I have pasted pull code used in mdp file.
> 
> Nilesh
> 
> 
> ; Pull code
> pull            = umbrella
> pull_geometry   = distance
> pull_dim        = Y Y Y
> pull_start      = YES             ; define initial COM distance > 0
> pull_ngroups    = 1
> pull_group0     = EMI
> pull_group1     = ETS
> pull_init1      = 0.0
> pull_rate1      = 0.001          ; 0.01 nm per ps = 10 nm per ns
> pull_k1         = 1000          ; kJ mol^-1 nm^-2
> 
> 

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

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

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