[gmx-users] Re: free energy calculation & TI

David Mobley dmobley at gmail.com
Thu Dec 14 19:47:22 CET 2006

Anirban and all,

I am putting this on list since I don't want to make a full-time job
out of answering free energy questions (although stay tuned for a more
extensive tutorial) and hopefully I can avoid answering the same
question more than once.

> I have seen your posts on gromacs mailing list regarding free
> energy calculation through TI and the need to partition the calculation
> into two steps :
> 1. change in VDW only
> 2. change in electrostatics only

This isn't essential, but I think it's a good idea, at least in terms
of efficiency.

> I was trying to run a free energy calculation run at lambda= and lambda=1
> and I found that at lambda=1 my job blew up i.e. it showed LINCS warning
> and
> dumped stepxx.pdbs...could you tell me what might  be the reason? This
> happened during stage 1 as mentioned above. In stage 2 the runs at
> lambda=0 and lambda=1 stops after sometime...seems there is some
> destabilizing forces.

By stage 2, you mean changing electrostatics? It would be helpful if
you could be more specific about the transformations you're doing, and
how. Any number of things could be going on:
(a) If, for example, you are doing things in a funny order (like first
turning off van der Waals interactions, and then modifying the
electrostatics) you may run into problems. (i.e. in that example, you
have the possibility of bare charges with opposite signs overlapping
with one another).
(b) If you are changing Lennard-Jones interactions without using soft
core, you can run into problems
(c) If you don't do adequate minimization or equilibration, you can
run into problems

Overall, you are basically getting forces that are too large for your
timestep for some particular reason. This could be related to the fact
that you're doing free energy calculations, or it could have nothing
to do with it. The basic solution is to figure out why you're getting
such large forces, and fix it (if it's a problem); otherwise, reduce
your timestep.

> I have also seen some posts where three steps are mentioned for free
> energy calculation..could you clarify on that ?

It will be helpful if you quote them or point me to them; I don't have
time to go through the archives to figure out what you're talking
about or how it applies here. Sorry. I'm happy to help if you can be
more specific.

Sorry this stuff is so scattered through the archives. One thing on my
to-do list is a "best practices" paper for free energy calculations
(which will initially be a wiki page, before I publish it). When I
pull that together, I will send an e-mail to the list so people can
refer to that rather than having to dig through the archives.

Best wishes,

> thanks
> anirban
> Dr. Anirban Mudi
> Postdoctoral Fellow
> Biocomputing Group
> Department of Biological Sciences
> University of Calgary
> 2500 University Drive NW
> Calgary AB
> Canada T2N 1N4
> (403) 220 6873
> http://moose.bio.ucalgary.ca/
> BE THE CHANGE YOU WISH TO SEE-> www.aidindia.org

More information about the gromacs.org_gmx-users mailing list