Re: [AMBER] Free energy calculation SHAKE's failure when heating.

From: Hannes Loeffler <>
Date: Sat, 25 Feb 2017 09:35:04 +0000

On Fri, 24 Feb 2017 22:06:00 +0300
Artem Shekhovtsov <> wrote:

> Thanks for help, Hannes!
> Yes, I turned off (ifsc=0) softcore treatment for "decharge state"
> and put chargemask on second ligand softcore atoms. Do i need to turn
> this option on (ifsc = 1)?
> Also try minimize and heat at "lambda = 0.5, decharge" with the same
> effect (blowing up).
> About my system:
> I have 3 carbon 6 hydrogen atoms in LTA region (MCS) and 5 softcore
> atoms. All LTA atoms share same coordinates and have same order in
> pdb file.

You have 3 disappearing hydrogens and 3 appearing hydrogens. In the
discharge step you would have to first remove the charge of the
disappearing hydrogens (at least). For some reason you have five atoms
in your crgmask. If you are trying to keep the system as charge
neutral as possible you would need to remove H1 from the mask.

Your real problem however is that this step is a linear transformation
from the initial state to a partially discharged state which,
structurally, is exactly the same as the initial state. But your group
file suggests to you have been using the topology for the final state,
AAB_water.parm7. I can only recommend here to use FESetup to not run
into such issues.

As for SHAKE you may want to avoid the headache by simply using
noshakemask for the whole perturbed region and set the time step
dt=0.001. You will get different free energies when you have SHAKEn
bonds that change bond lengths (and/or force constant). I think sander
doesn't warn you about this (pmemd does).


AMBER mailing list
Received on Sat Feb 25 2017 - 02:00:03 PST
Custom Search