[gmx-users] Re: TI, sampling, sc_power, and sc_alpha

[David Mobley]

Matt,

On 5/30/06, mernst at tricity.wsu.edu <mernst at tricity.wsu.edu> wrote:
> David, I forgot to ask:
>
> You said that
>
> " the soft core settings that work well for the vdW portion make the charging portion a
> more "difficult" transformation than normal, and the settings that work well for
> charging make the vdW portion "difficult". "
>
> What settings would you initially suggest to make the changes less difficult? Using
> sc_power=1 and sc_alpha=0.5 is what you suggest for vdW changes, but what for charge
> changes? Should I be using sc_power=2 and sc_alpha=1.5 to emulate older behavior? Using
> sc_power=1 and sc_alpha=X.Y?

Oh. Electrostatics are usually a smooth, nearly linear function of
lambda without soft core. I would recommend *not* using soft core for
electrostatics -- just use the default linear scaling with lambda that
you get with alpha=0. Soft core actually puts *more* structure in the
curve, which is not what you want.

> I'm also a little puzzled by the decomposition of the two calculations if I'm going to
> be changing sc_power and/or sc_alpha. Should my second calculation merely add an extra
> modification to the B-states (i.e. after changing atom types in B-states, I extend the
> trajectory using a .top file with charge changes added to B-states) or should I use a
> new topology file with my initial change placed in the A-states and only the new
> (charge) changes placed in the B-states?

You want your series of calculations to create a thermodynamic cycle
to take you from where you start to where you want to end up. So if
you do this in multiple steps, you need to do something like (or in a
different order):

PL -> PL(new electrostatics) -> PL(new electrostatics and vdW) or some such.

So, in the first transformation, you would change the electrostatics
in your small molecule (normal A state; new electrostatics in B
state). Then in the second transformation the A state would be the B
state you end the first step with (new electrostatics) and the B state
would be the state where you have your new electrostatics and vdw
interactions.

You may want to refer to Michael Shirts' hydration free energy papers
(2003 and 2005) for some more details on the soft core potentials and
stuff. He used the approach I'm describing, at least in the 2005
paper, and explains why.

David


> Thank you again for your help.
>
> Matt Ernst
> Washington State University
>
> _______________________________________________
> 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
>