Re: AMBER: how can I deal with the puzzle?

From: Thomas E. Cheatham, III <>
Date: Tue, 8 Mar 2005 20:53:48 -0700 (Mountain Standard Time)

> I have two conformations of a protein which have a RMSD of 440
> angstrom. I want to transfor from one conformation to another. In the

Are you sure that the difference is 440 angstroms or 44 nm? If so this is
of such an extremely large size that a few picoseconds (10000 * 0.0005 or
5 ps) will not be sufficient to move the system at all. It is very
unlikely that the system can adapt to such large force/structure changes
(on the order of 400 angstroms) in such a short period of time.

> imin=0, nstlim=10000, dt=0.0005,

Why do you need to have the timestep so short (using SHAKE); are their
instabilities in the system to begin with?

> ntpr=10, scee=1.2, cut=25,
> igb=0, irest=1, nmropt=1,

It appears that you are not using generalized Born (implicit solvent) yet
you have a large cutoff and PME was not turned off (in the &ewald namelist
setting use_pme=0 ?); this is a bit confusing. Do you want gas phase?

> itgtmd=1, tgtrmsd=0.0, tgtmdfrc=0.5,
> &end
> &wt
> TYPE='TGTRMSD', istep1 =1, istep2 = 1000,
> value1 = 441.2, value2 = 440.,
> &end

400 angstroms is a really large number in the atomistic world of AMBER

> 2.And In the output file, I found the temperature increased from 300K to
> 320K during 10000 steps, the whole energy was increased from negative
> value to positive. is it right? 3.Should I use the Born model(igb=1) for
> targeted MD?

Do you have solvent? If not, figure out carefully how to run a vacuum
simulation. If you can afford it, run GB (or implicit solvent) and even
better explicit solvent (which can be debated); however the size system
you are investigating (if the 400 angstroms RMSd is correct) is really at
the cutting edge of what is currently possible.

To better your understanding of targeted MD and these simulation
protocols, it may be advantageous to run a small system first to test
things out; watch movies, plot the energies and values and see what is

Feel free to ask the list for further advice as many, if they understand
your problem well, will provide excellent insight beyond my comments.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Mar 09 2005 - 04:53:01 PST
Custom Search