[gmx-users] umbrella sampling

[Justin A. Lemkul]

Vijayaraj wrote:
> Hello,
> 
> I am doing umbrella sampling to calculate PMF curve for the detachment 
> of a terminal cyclic peptide with 8 aa's (CP) from the self-assembled 
> cyclic peptide nanotube. I extracted the reaction coordinates staring 
> from 5.5 ang COM distance between the terminal CP and the 2nd CP to 17.5 
> ang, I did umbrella sampling on 25 configurations (for 5ns) and the 
> window size is 0.5 ang. I used the following pulling code for umbrella 
> sampling,
> 
> pull            = umbrella
> pull_geometry   = distance
> pull_dim        = N N Y
> pull_start      = yes
> pull_ngroups    = 1
> pull_group0     = Chain_B
> pull_group1     = Chain_A
> pull_init1      = 0
> pull_rate1      = 0.0
> pull_k1         = 1000    
> pull_nstxout    = 1000    
> pull_nstfout    = 1000    
> 
> 
> I restrained the 2nd CP unit and the pull_rate1 is 0, so the COM 
> distance between the chain_A (terminal) and chain_B (2nd CP) should be 
> restrained. after 5ns of umbrella sampling, I calculated the COM 
> distance between chain A and B, but it was not restrained, for the 5.5 
> ang COM distance restrain, the COM distance varies from 4.5 to 5.5 ang. 
> and also the pulling cyclic peptide undergoes large conformational 
> sampling. from the WHAM analysis I understood the sampling window is 
> poorly represented. In addition to COM distance restrain, can I restrain 
> the pulling CP's backbone atoms? so that the pulling groups large 
> conformational sampling will be reduced.

You could implement dihedral restraints to fix the backbone secondary structure, 
but I can't comment on the stability of trying to use these restraints in 
addition to the pull code.  Seems like a lot going on at once, to me.

Also consider the fact that 5 ns is an extremely short timeframe to gather 
meaningful data.  Perhaps you just need more time in each window to equilibrate. 
  At the shortest COM distance, your two molecules are still likely experiencing 
some interactions, and it may require a great deal of sampling in this window to 
converge the simulations.  You haven't shown the rest of your .mdp file (always 
a good idea!), so we can only guess at whether or not your other settings should 
lead to a sensible result.

-Justin

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

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

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