AMBER Archive (2003)

Subject: Re: free energy perturbation (Gibbs/Amber 5)?

From: Lepsa (lepsik_at_marilyn.uochb.cas.cz)
Date: Fri Jan 10 2003 - 03:56:50 CST

Dear Heike and AMBER community,
I've gone thru and extended the Val=to=Ala tutorial. I'll present my results
and raise some questions.

> But I don't understand the perturbation charges.
> Can you tell me, why the pert.charge in the tutorial for CB is 0.105 and
> not the charge for alanine (CB -0.1825)? Why are the other charges not
> changed as well?

CB 0.105 and HC zeroes make the resulting alanine neutral but it's not the
reality in my mind (just pedagogical reasons,
I suspect?). Though, I cannot just put in the charges from alanine residue
because you'd end-up in non-neutral ALA!!
To be right, you have to adjust the charges or even better calculate them
from QM. I used the former case (I'm not hunting the very real results) and
used -0.165 for CT and 0.09 for HCs.

> My last question, what is the actual result (kcal/mol) for this simulation
> (valine to alanine)?

Here are some of my results (kcal/mol):
tutorial charges
FEP (9000steps,21win,idsx0=10) 3.8/-4, (9000steps,21win,idsx0 = -1) 4.1/-4.1
TI (9000steps,21win,idsx0=10) 4.0, (9000steps,21win,idsx0= -1) 3.9
dyn mod win TI (9000steps,21win,idsx0=1) 4.0

Summary: transition val-to-ala with tutorial charges has free energy gain
~ -4.0kcal/mol
Quest: Elstat and 1-4 elstat cancel out and the favorable terms are vdW!
What should that mean?

my charges:
TI (9000steps,21win) -2.2,
dyn mod win TI (9000steps,21win) -2.2

With the more proper charges, the change is UNFAVORABLE!! Has someone
something to say to that? It's unfavorable
1-4 electrostatics that outweighs favorable elstat! Why is that? What about
solvation?

Just a minor correction: isldyn is not a flag for thermodyn integrat as is
said in the tutorial.

Looking forward for your comments and discussion.

All the best,

    Martin Lepsik