[gmx-users] g_wham problem with negative COM differences

[Justin A. Lemkul]

Anni Kauko wrote:
> 
>     Date: Wed, 11 Apr 2012 08:38:05 -0400
>     From: "Justin A. Lemkul" <jalemkul at vt.edu <mailto:jalemkul at vt.edu>>
>     Subject: Re: [gmx-users] g_wham problem with negative COM differences
>     To: Discussion list for GROMACS users <gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>>
>     Message-ID: <4F857B2D.3050100 at vt.edu <mailto:4F857B2D.3050100 at vt.edu>>
>     Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> 
> 
>     Anni Kauko wrote:
>      > Hi!
>      >
>      > I try to perform pmf calculations for case where a peptide shifts
>      > through the membrane. My COM differences should vary from 2.3 to
>     -2.5.
>      >
>      > My problem is that g_wham plots negative COM difference as they
>     would be
>      > positive.  In pullx-files the COM differences are treated correctly
>      > (look below). My peptide is not symmetric, so profile curves are not
>      > symmetric, so loosing the sign for COM difference screws my profile
>      > curve completely.
>      >
>      > I did not manage to find any pre-existing answers to this problem
>     from
>      > internet.
>      >
>      > First datalines from pullx files:
>      > (sorry for strange file names...)
>      >
>      > pull_umbr_0.xvg:
>      > 0.0000  6.26031 2.27369
>      >
>      > pullz_umbr_23.xvg:
>      > 0.0000  6.09702 0.0233141
>      >
>      > pullz_umbr_50.xvg:
>      > 0.0000  6.02097 -2.50088
>      >
>      > g_wham command:
>      > g_wham -b 5000 -it tpr_files.dat  -ix pullz_files.dat -o
>      > profile_test.xvg -hist histo_test.xvg  -unit kCal
>      >
>      > My pull code:
>      >
>      > pull            = umbrella
>      > pull_geometry   = distance
>      > pull_dim        = N N Y
>      > pull_start      = yes
>      > pull_ngroups    = 1
>      > pull_group0     = POPC_POPS         ; reference group is bilayer
>      > pull_group1     = C-alpha_&_r_92-94 ; group that is actually pulled
>      > pull_init1      = 0
>      > pull_rate1      = 0.0
>      > pull_k1         = 1000     ; kJ mol^-1 nm^-2
>      >
> 
>     Your problem stems from the use of "distance" geometry.  This method
>     assumes the
>     sign along the reaction coordinate does not change, i.e. always
>     positive or
>     always negative.  If the sign changes, this simple method fails.
>      You should be
>     using something like "position" to allow for a vector to be
>     specified.  Perhaps
>     you can reconstruct the PMF by separately analyzing the positive
>     restraint
>     distances and negative restraint distances (note here that
>     "distance" really
>     refers to a vector quantity, and thus it can have a sign), or
>     otherwise create
>     new .tpr files using "position" geometry, though I don't know if
>     g_wham will
>     accept them or not.
> 
>     -Justin
> 
>     --
>     ========================================
> 
>     Justin A. Lemkul
>     Ph.D. Candidate
>     ICTAS Doctoral Scholar
>     MILES-IGERT Trainee
>     Department of Biochemistry
>     Virginia Tech
>     Blacksburg, VA
>     jalemkul[at]vt.edu <http://vt.edu> | (540) 231-9080
>     <tel:%28540%29%20231-9080>
>     http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin
> 
>     ========================================
> 
> 
> Thank's!
> 
> I managed to solve my g_wham problem by doing two things:
> 
> 1. New tpr-files with proper pull code for g_wham.
> 2. I also needed to modify signs of pullf values: If value for pullx 
> distance was negative, I reversed the sign of corresponding pullf value. 
> I did that by my own script.
> 
> The new pull code:
> 
> ; Pull code
> 
> pull            = umbrella
> pull_geometry   = direction
> pull_vec1       = 0 0 1
> pull_start      = yes
> pull_ngroups    = 1
> pull_group0     = POPC_POPS         ; reference group is bilayer
> pull_group1     = C-alpha_&_r_92-94 ; group that is actually pulled
> pull_init1      = 0
> pull_rate1      = 0.0
> pull_k1         = 1000     ; kJ mol^-1 nm^-2
>    
> I am little bit confuced, why I needed to tweak signes of pullf values. 
> But like that I got the curve that resembles two half curve made for 
> positive and negative pullx distances separately. That curve also makes 
> sense from biochemical point of view.
> 

Such changes do not seem appropriate to me.  If you change the sign of the 
pulling force, you change the implication of what that value means.  What 
happens if you run your simulations with the new (more appropriate) .mdp file? 
Do the forces have the same magnitude, but opposite sign?

> -Anni
> 
> PS. Thank's for your excellent tutorials. They have been indispensable 
> help for me to get started with gromacs!

Glad to hear it :)

-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

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