Re: AMBER: MD problems/shake

From: Thomas E. Cheatham, III <>
Date: Tue, 20 Dec 2005 12:20:26 -0700 (Mountain Standard Time)

> Yes, this is a ligand (ATP) problem. The atoms HO-OH in the first
> phosphoreus group. The parameters to ATP are from parm94.dat.
> I have tried to do dynamics on isolated ligand having shake on,
> and that is a problem. I got the SHAKE error again on the same
> atoms:
> "Coordinate resetting (SHAKE) cannot be accomplished,
> deviation is too large
> NITER, NIT, LL, I and J are : 0 0 1 1 2

As a followup to the list regarding what was happening and what to look
for... If you have an isolated ligand and SHAKE failures are evident in
MD simulation, something is likely funny with the parameters that allows
atoms to move within vdw spheres leading to large forces such that the
integrator or SHAKE fails. Working with Pawel, I obtained the
prmtop/inpcrd and did some experimenting.

(1) If you get SHAKE failures on your drug/ligand, try first extensive
minimization of the isolated compound in-vacuo. If atoms are penetrating
vdw spheres (leading to overlap), this will be readily apparent when
looking at the structure. In this case, the HO atom (with no vdw
parameters) overlapped with a nearby oxygen. This is occurring since the
HO atom has a relatively large (~ +0.4) charge and strongly interacts
favorably with the nearby oxygens (~ -0.6 charge). As there is no vdw
repulsion, this attraction overcomes the bond and angle forces holding
it away.

(2) Running MD simulation on the isolated system in-vacuo shows that
temperature cannot be maintained (it quickly rises) which is further
evidence of the instability.

I got around this problem (assuming the charges were reasonable) by
adding a vdw potential to this hydrogen. I changed the OH parameters
(for this compound) to have a small vdw like HS (0.6 0.0157) and then
minimization and MD were stable (although the hydrogen in question is
still very strongly attracted to the nearby oxygen, it does not penetrate
the vdw sphere).

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Tue Dec 20 2005 - 19:53:01 PST
Custom Search