Re: [AMBER] Regarding SHAKE Parameter

From: David A Case <>
Date: Fri, 23 Jul 2021 09:17:25 -0400

On Fri, Jul 23, 2021, Sumon Naskar wrote:
> I am working on a DNA-protein system. According to the
>protocol, SHAKE should be applied to H containing bonds from heating
>onwards. The protocol did not mention anything about the minimization in
>particular. My doubt is if, in the 2 minimization stages, we do not
>apply the SHAKE (ntf=1), would it really change the system as a whole? I
>completed the production as well and the geometry was fine till the end.
>There is one minimization with restraints on the biomolecules and one
>with no restraints at all. Does not applying shake in minimization make
>the simulation go wrong? What is the general purpose of applying the
>SHAKE algorithm?

You should be fine in leaving SHAKE off during an initial minimization step.
Some notes:

1. SHAKE constaints the lengths of bonds, generally only to hydrogen. It is
controlled by the ntc paramter. You can leave ntf at 1 if you wish.

2. For dynamics, SHAKE is important, especially for water, since the common
water force fields all assume a rigid water molecule, which SHAKE provides,
since there is also a "bond" between the two hydrogen atoms. You will get
wrong behavior for water if you run without SHAKE.

3. SHAKE is, by construction, an MD algorithm. Handing similar constraints
in minimization requires a different approach, and this approach has never
been consistently implemented in sander or pmemd. Unless you have a bad
initial geometry, running short minimizations without SHAKE rarely causes

...hope this helps...dac

AMBER mailing list
Received on Fri Jul 23 2021 - 06:30:02 PDT
Custom Search