[gmx-users] PMF calculation using g_wham using different simulation boxes

[chris.neale at utoronto.ca]

Sorry, I don't use gromacs g_wham so I can't help you on the specifics  
there. I don't know if the -zprof0 does anything special, but you know  
that your PMF can be arbitrarily translated as dG is unknown to an  
additive constant, right?

## assume rx coord in first column and dG in second column
desired_position=0.0;
constant=$(grep "^${desired_position} " file | awk '{print $2}')
awk '{print $1,$2+'${constant}'}' file > file.translated

Again, I am not talking specifically for g_wham, but no you don't need  
to use the same force constant or sampling time at each window while  
performing wham in general.

-- original message --

Hi

Thank you Justin and Chris, for your comments and suggestions.

All the restrained distances are less than half of the box length. The  
reason I want to use different box sizes is:  g_wham gives zero value  
for the first bin, so I need extend my simulation for long restrant  
distances for the use of -zprof0 option to make the PMF zero.  
Unfortunately, after finished simulations for short restraints, I  
realized that the box size may not be long enough for the use of  
-zprof0. Therefore I think of using larger box sizes for the rest of  
the windows with longer restraints. As you suggested, I need to do a  
test of finite-size effect before switching to larger box sizes.

Btw, for using g_wham, do I need to use the same parameters for  
different windows, e.g., the force constant, the simulation time and  
the width of the window?


Cheers,
Li Jianguo
Research fellow
Singapore-MIT Alliance
National University of Singapore


I have never heard of such a thing.  Is the restraint distance always  
less than
1/2 the box dimension in the pulling direction?  If not, then this approach
doesn't work.

I have always been under the impression that when running umbrella  
sampling, you
have to actually be doing sampling in different regions of  
configurational space
in the same system, in order to get a continuous energy function and  
appropriate
sampling.

-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

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




Message: 4
Date: Wed, 02 Dec 2009 08:42:36 -0500
From: chris.neale at utoronto.ca
Subject: [gmx-users] PMF calculation using g_wham using different
     simulation    boxes
To: gmx-users at gromacs.org
Message-ID: <20091202084236.bk7yqoke0w0wcwcs at webmail.utoronto.ca>
Content-Type: text/plain;    charset=ISO-8859-1;    DelSp="Yes";
     format="flowed"

Hi Li,

In my opiniun, what you propose is acceptable, but not rigorous, and
you will need to prove that there are no unexpected artifacts of
changing box size. For example, you should definitely run a couple of
your small box systems as larger boxes as well and show that you get
the same histograms for the distances between your polyatomic particles.

Justin is correct about the need for a box that is 2x larger than your
displacement, and pull_pbcatom is an important option to set as well.

Chris.