Re: [AMBER] Total system charge and TI simulations

From: David Case <>
Date: Mon, 1 Oct 2018 23:49:24 +0000

On Mon, Oct 01, 2018, Veenis, Andrew Jay wrote:
> I would like to use TI to obtain pKa estimates of functional groups in
> RNA molecules. Water and ions will be modeled explicitly and the systems
> will typically consist of ~30,000 atoms. Because the functional
> group will be neutral at λ = 0 and charged at λ = 1, the overall
> system will also be neutral and charged, respectively. From what I
> have read, this should be okay as long as the system is sufficiently
> large, however, this is based on literature/discussion from 10+ years
> ago. Several suggestions were provided in an older post, found at
> Is there now a preferred way
> of handling these types of simulations? Is it best to just let the charge
> of the total system change or should I implement something to keep the
> system neutral?

Almost all practical pKa calculations let the total charge of the system
change from one end-point to the next, and rely on the fact that you are
always comparing two similar simulations (e.g. one transformation in RNA,
and a second in some model compound): one is relying on cancellation of
non-physical effects that arise when a "neutralizing background" appears
for charged systems in PME.

One aspect that is not covered in the post you mention above: if the size of
the simulation box for the model compound is much different from the box size
of the macromolecule (which is likely) then there are (still somewhat small)
size-dependent effects that won't cancel in this sort of calculation. It's
not feasible to be constantly re-doing the model compound calculations with
different box sizes, but it may be worth spending some effort on making sure
that the model compound simulations used as a reference are not "too small".

As usual, I encourage others on the list to chime in with more focussed
suggestions and computational results.


AMBER mailing list
Received on Mon Oct 01 2018 - 17:00:02 PDT
Custom Search