AMBER Archive (2009)
Subject: Re: [AMBER] RE:xleap problem
From: M. L. Dodson (activesitedynamics_at_comcast.net)
Date: Thu Jan 29 2009 - 10:22:22 CST
oguz gurbulak wrote:
> Dear D Case and M. L. Dodson,
> *Thank you* very much for your advices. It *works well* after using the set
> command in LEaP to add the box information. And I run a min and then run a
> constant pressure PME dynamics (1 atm ). I plotted the density over the
> trajectory. The density must be 0.7636 g/ml. The min.in and md.in files are
> in below. I attached the density graph to the mail. I'm planning to run a
> constant volume to equilibrate the temperature as you said to me. I want to
> learn if the operations I did are true ? What do you think about that ? What
> do you suggest to me ? And I also want to make analysis without using ptraj.
> So I want to learn the structure of the output files ( restrt, mdcrd, mdvel,
> mden ) and meaning of the colums and rows in the output files. Can I easily
> learn the informations of the atom's xyz coordinates, velocities and
> accelerations in any time step and pressure tensors ? Or Should I write a
> script in order to get these informations ? And I want to ask a question.
> For example
> NSTEP = 2000 Density = 0.7636
> NSTEP = 2500 Density = 0.7689
> I want to get the output informations for NSTEP = 2000 in target Density =
> 0.7636 for input and start a md with little time steps. Is it possible to do
> this in Amber? If so, what should I do to do this operation ?
> Thanks in advance.
I am curious why you think the density MUST BE 0.7636 g/ml (aside from
the four decimal precision which is very much more precise than would
be expected over a molecular dynamics run). Is this an experimental
The graph clearly shows that the density is increasing beyond 0.76
g/ml and approaching 0.82 g/ml. Whether it is equilibrated at 10000
steps or not would require a longer dynamics run.
What is the temperature (since you have not equilibrated temperature
at this stage)? Is there a physical reason why the density should be
very temperature sensitive?
I should have been a little more expansive in my recipe for
equilibration. I always run a short constant volume simulation (to
ramp up the temperature to the desired value) before I equilibrate the
density (in the constant pressure run). One can sometimes get into
difficulty with just dumping a lot of kinetic energy into an
unequilibrated constant pressure simulation and/or trying to ramp the
temperature during a constant pressure simulation (depending on the
details of how you do it). This has been discussed on the email list
at least 2 or 3 times in the last year or so. If you had no problem,
then I guess it is OK to use the last frame of your simulation as a
restart. If the temperature is approximately what you set it to be,
then restart the constant pressure run and look to see to what the
If, when your temperature is correct, and the constant pressure run is
equilibrated, the density is still too large, I would say this
strongly hints that there is a problem with your force field for this
molecule. A good force field should be able to reproduce (on average)
the experimental density fairly closely, I think. Someone can correct
me if my logic is wrong on that, though, since I only have first hand
knowledge using (mostly) standard force fields and aqueous solutions.
> initial minimization
> ntc=2, ntf=2,
> MD equilibration
> imin=0, irest=0, nstlim=5000000, dt=0.002, ntx=1,
> ntc=2, ntf=2, ntpr=2000, ntwx=2000, ntwr=2000,
> cut=9.825, ntb=2, ntp=1, ntr=0, ntt=1,tautp=2.0,
> tempi=0.0, temp0=273.0, pres0=1.0, taup=2.0,
> 2009/1/27 David A. Case <case_at_biomaps.rutgers.edu>
>> On Mon, Jan 26, 2009, oguz gurbulak wrote:
>>> I'm now on Windows so I can only use rar or zip format for
>>> compressing files.
>> This is hardly true....just Google for "gzip windows" for example.
>>> the mail. I used packmol in order to get my pdb file that includes 100
>>> n-dodecane molecules. I added TER cards between each seperate n-dodecane
>>> molecule with using a script. You can examine my packmol pdb file
>>> to the mail.
>> You need to use the "set" command in LEaP to add the box information
>> after you load the packmol pdb file.
>> Also note: a box size of 9 Ang or so is too small for Amber. The box
>> size has to be at least twice the cutoff, so as a practical matter, the
>> smallest box that really works is about 18 Ang.
M. L. Dodson
Business Email: activesitedynamics-at-comcast-dot-net
Personal Email: mldodson-at-comcast-dot-net
AMBER mailing list