Re: AMBER: perturbed atoms are not bonded on using TIP5P water

From: David A. Case <>
Date: Sun, 14 Jan 2007 09:39:13 -0800

On Fri, Jan 12, 2007, sugino tatsuro wrote:
> I created my system such that 1293 TIP5P waters were placed around a water
> molecule by using xLaep. Then I chose one of the water molecules to be
> perturbed.
> Only electric charge of atoms of the water molecule was perturbed into 0
> from default value.
> Bond information of perturbed water molecule could be confirmed by the
> command "desc" before saving topology and coordinate files.

> But there was no bond information in the saved topology file.

How did you establish this? Were you looking inside the prmtop file to see
that the perturbed water had no bonds, but all the others did?

How did you distinguish the one atom to be perturbed from the others? (A
different unit name?)

> So,when I performed MD simulation with these saved topology and coordinate
> files, perturbed atoms moved separately (because these are no bond).

This does sounds like a possible bug. Someone will probably have to try to
reproduce it to go further. Answers to the above questions may help.

Concerning your related email:

> Before perturbation, I performed energy minimization by using the system
> composed of 1293 TIP5P water molecules created by xLeap. But their
> coordinates were not changed in the calculation, and little of the total
> energy of the system was decreased.

You only ran 30 steps of steepest descent minimization; this will indeed
not move things very much.

> Still, after that I carried out standard MD calculation, and it was
> performed without problems.

Sounds like you are OK here.

> > loadamberparams gaff.dat
> > source TIP5P.cmd

You don't really need either of the above commands. The TP5 unit is part of
solvents.lib, which is loaded with any of the standard leaprc files. Still,
it should not do any harm.

> > loadamberparams parm99EP.dat

For pure water, this is not needed; for protein/water simulations, it is
is not recommended any more. Protein simulations should be using one of the
more recent forcefields where the backbone torsion potentials have been
adjusted: possibilities are ff99SB, ff03, ff02r1 and ff03ua. (Clearly, this
last statement is somewhat subjective, but I believe it represents good


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Jan 17 2007 - 06:07:16 PST
Custom Search