[gmx-users] Umberella sampling

Justin A. Lemkul jalemkul at vt.edu
Wed Apr 6 17:39:41 CEST 2011

mohsen ramezanpour wrote:
> Dear Dr.Justin
> I had asked this question before but unfortunately I didn't answered good!
> Instead of Pulling and separating some structures in definite distances,
> I located my drug in definite distances along z axis (as my initial 
> structures for doing umbrella sampling).
> Am I right?

I have no idea.  Presumably, if you've defined the reaction coordinate 
appropriately and distributed configurations along this vector at adequate 
intervals to sample the whole coordinate, then yes.

> The main problem underlying this question is :
> Do we need to know the velocity of drug in every structure we choose for 
> doing umbrella sampling?(do they need to include the velocities too?)

PMF is calculated from either the resulting positions (in pullx.xvg) or forces 
(pullf.xvg) from the harmonically restrained simulations.

> if no,which criteria assures with this K our molecule can cover the 
> length of it's windows?

You can't necessarily know a priori if your force constants are adequate. 
Unfortunately, it's a bit empirical.  Run some simulations and analyze the 
result with g_wham.  A smooth PMF with good overlap between the histograms 
indicates your sampling is (at least) good, if not optimal on the first try.

> Because I have chosen windows with different lingthes,And I want now to 
> choose K so that it can sweep the lengthes.

Weaker force constants allow for more oscillation, but again, there is no 
prescribed way to know what values to choose.  If you have some very large 
window, you assign a weaker force constant, but then you may not cover the 
entire sampling window without a longer simulation in this window such that your 
molecule wanders through all of the applicable configuration space.


> Thanks in advance for your reply


Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list