AMBER Archive (2004)

Subject: Re: AMBER: FEP error/early termination - why?

From: Annette Höglund (
Date: Tue Jun 01 2004 - 09:34:53 CDT

Dear Alfonso and everybody else,

Thank you for the detailed description you provided. I realize that this
parameter is going to be useful in this FEP case. I still have a few
questions to what you described:
For the first step (IELPER = 1) I should input the equilibrated crd, and
then which one: or
For the second step (IELPER = -1) which crd file (the same as went into
OR the one that came out of the first step), and which or

What should a "good" F_energy plot look like? Saturation type of thing?
Not linear until the end I guess?
Can one say from which step (IELPER = 1, or -1) the major contribution
to the free energy should come from?
It seems that I get linear F_energy plots from the first (seem perfectly
mirrored) fir IELPER = 1. However for IELPER = 1, I run into trouble and
the F_energy plots do not mirror eachother on the plus and minus sides.
What can I do to get a "less messy" second step?

Ok, thanks a lot in advance,


Alfonso Garcia Sosa wrote:

>Dear Annette
> Have you tried using the option for electrostatic decoupling (IELPER)?
>Basically, I think what might be happening is that as the atoms disappear,
>they become very small and cannot repel the close contacts due to
>charge-charge attractions. This makes the simulation crash because atoms
>get too close too each other. What electrostatic decoupling offers is to
>do the run in two legs, first turning off the electrostatic interactions
>so that the end state are the atoms with no charges. The second run would
>involve decreasing the nonbonded interactions such as vdW radii, and so
>the atoms can shrink until they disappear without other atoms crashing on
>top of them, because they have no charge. This option is explained in the
>manual of amber 7, option name IELPER (page 171). If I remember well, you
>don't need to do anything else than use this flag option in the ***.in
>file, that is one run with IELPER=1 and another run with IELPER=-1. You
>don't need to change the other files and after the runs you would then sum
>up the results from each run. Good luck and regards
>On Mon, 31 May 2004, Annette Höglund wrote:
>>Dear All,
>>I have tried to run an initial free energy perturbation for a ASP->ALA
>>sidechain in a protein-DNA complex (after minimization and
>>equilibration) using a similar gibbs input as in the val-to-ala tutorial.
>>For this particular sidechain the FEP terminates after about 75% with
>>When I plot the F_energy it seems to be a "concave" rather than "convex"
>>curve (as in the tutorial).
>>Where should I start looking for the problem?
>>What should an "optimal" F_energy plot look like for an FEP?
>>Regards and thanks in advance,
>>Annette Höglund

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to