[gmx-users] Jarzynski and PMF

Sai Kumar Ramadugu sramadugu at gmail.com
Tue Aug 2 22:43:20 CEST 2011

Dear All,

I have tried to calculate the free energy of binding using Jarzynski
equality. I employed the following procedure.

--I did force pulling simulations along Z-direction as exemplified in
Justin's umbrella sampling simulations. I did 50, 250 and 500 pulling
simulations to test the convergence and also two different pulling rates
(2nm/ns and 4nm/ns), two different force
  constants (2000 kJ mol^-1 nm^-2 and 4000 kJ mol^-1 nm^-2).

--From the pulling simulation, I get the pullf.xvg and pullx.xvg.

--I use the force from pullf.xvg and calculate the work at each time step by
multiplying with v*dt and calculate average work for each simulation.

--Once I have the work for each simulation, I calculate the exp(-beta*W) for
each simulation and calculate the average of exp(-beta*W).

--Once I have the average of exp(-beta*W), I calculate the free energy delF
= -kB*T ln(exp(-beta*W)).

The question I have is whether my approach for calculating the work is
correct/feasible or not.
Does the pullf.xvg written by mdrun during a pulling simulation contain the
pull force only?
Do I need to add any corrections?
I'm doing NPT simulations, so can I add the pdV correction as suggested in
one of the paper (Macromolecules, 2008, 41 (6), pp 2283–2289) by Berk Hess?
Or did I understand the paper incorrectly?

For a publication level of work, I'll use NVT ensemble. But for now my
biggest concern is calculation of work.

Any suggestions will be greatly helpful.

Thanks for you time,

Sai Ramadugu
Dept of Chemistry,
University of Iowa.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110802/ce03d9b2/attachment.html>

More information about the gromacs.org_gmx-users mailing list