[gmx-users] Number of windows in umbrella sampling

Justin A. Lemkul jalemkul at vt.edu
Mon Feb 13 14:45:32 CET 2012

shahid nayeem wrote:
> Dear Justin
> I am doing umbrella pulling simulation of a protein complex wt and 
> mutant. I expect mutant to give lower deltG value. I am attaching a tif 
> file of energy vs time curve of wt and mutant protein on pulling 
> simulation. These energies are obtained by g_energy and selecting 11 
> option which is COM pulling energy. In this curve the first peak 
> decreases and again rises at longer time. How many windows should be 
> selected. As expected the peak of COM pulling energy is lower in 
> mutants. Please explain why the energy again rises at higher time. 
> Should I use the windows upto 160ps only because thereafter in both 
> curve there is rise in energy value. the pull code used is as follows.

What you have obtained is a path-dependent energy that may or may not signify 
anything useful - it almost certainly does not.  I cannot offer an explanation 
of the sharp increase towards the end of the simulation other than to speculate 
that your box is of insufficient size and you're encountering PBC issues.

As for how many windows are necessary, it's also impossible to tell.  You need 
enough windows to adequately sample the reaction coordinate.  Thus, it is 
decided based on how far you need to separate the two species, how strong the 
force constant is during the US simulations, the nature of the interactions in 
the system and how fast they converge, etc.


> pull_geometry   = distance 
> pull_dim        = Y     N  Y
> pull_vec1       = 0.75 0  1
> pull_start      = yes      
> pull_ngroups    = 1
> pull_group0     = Chain_B 
> pull_group1     = Chain_A 
> pull_rate1      = 0.01      
> pull_k1         = 1000     


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