[gmx-users] Free energy for charged molecules

[David Mobley]

Ignacio,


> Is there some precaution or additional term I should take into account
> when calculating the free energy of solvation for a charged molecule?
>
> I plan to make the molecule "disappear" whith lambda moving from 0 to 1
> and then perform a thermodynamical integration, this seems to work
> reasonably well for neutral molecules. Is there any "gotcha" with
> charged molecules?

In my opinion, yes. This has been discussed some on the list before so
I encourage you to search the archives. The basic problem is that,
when using PME (as you are) the system must be net neutral. When you
turn off or on a charge in your system, you are necessarily making the
system charged. Most PME codes are actually designed to be able to
handle this (in the sense of not exploding) by introducing a uniform
neutralizing background charge to enforce neutrality.

Unfortunately, this is actually problematic for free energy
calculations: If you just turn off the charge on your molecule,
instead of getting the free energy of turning off that charge, you're
getting the free energy of turning off the charge plus the free energy
of turning on a uniform neutralizing background charge, which is
obviously not what you want.

There have been some comments on the list suggesting that one might be
able to subtract off this contribution by either (a) making a series
of separate calls to PME (which would require code changes), or (b)
doing a series of free energy calculations with the charges on various
components of the system set to zero. I have a rough idea of how to do
option (a), but h aven't been convinced yet that there is a rigorous
way to do it, and option (b) sounds like it might be reasonable, but I
again haven't been convinced it's rigorous, nor I have I seen any
publications using this method.

I would encourage you to dig in to this issue; in my opinion, it is a
sufficiently large problem that I would not let someone publish a free
energy calculation where they change the net charge of a system while
using PME unless they can demonstrate that they have correctly handled
this. (It will, of course, "work" to do it with the uniform
neutralizing background charge. The only problem is that your answer
will be wrong by some unknown amount which is likely to be fairly
large.)

Please let me know if you get anywhere on it: I'd love to be able to
do these calculations on charged systems but so far don't have a way
to do it  (with PME) that I'm convinced is rigorous.

Thanks,
David


Of course, I'm using PME, pbc=xyz and a large enough
> simulation box.
>
> Thanks
>
>
>
> ___________________________________________________________
> To help you stay safe and secure online, we've developed the all new Yahoo! Security Centre. http://uk.security.yahoo.com
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>