Re: [AMBER] Different trajectories with different RMSD with changing the parameters of gamma_ln, taup, timesteps and extend of density equilibration

From: Thomas Cheatham III <>
Date: Fri, 8 Jul 2011 17:11:59 -0600 (Mountain Daylight Time)

> Also one more doubt, how can I make sure that the system reached
> equilibration? My systems contains 437 res and 14500 water molecules.
> When I checked the density of the system its close to 0.99 or 0.98
> always.. So I thought it equilibrated very well..If the system doesnt
> reach the equlibration stage, how can I know that....?

This is the big question and it depends what you are talking about, i.e.
equilibration of water density, ion distributions, temperature,
conformational distribution, ... Each of these processes may relax on a
different time scale. Equilibration of water density is fast; ion
distributions may take 50-100+ ns. Conformational distributions? Depends
on the system.

Normally "equilibration protocols" focus on getting a stable density,
pressure, temperature. If these reach a plateau, we can say that the
density, pressure and temperature appear to have equilibrated. Then,
simulations move into the production phase...

We cannot state, generically, that the system is equilibrated at this
point unless we are specifically referring to a particular indicator (like
density/temperature or pressure). If you ask, is the system equilibrated?
I would likely respond "no" since we do not know what indicator you are
referring to.

Clearly at 2 ns, the conformational distributions (protein, DNA, ions)
will not be equilibrated as Dan Roe mentioned. To investigate the
convergence we choose and monitor various progress variables (RMSd,
2D-RMS, distances, angles, ion distributions, ...). Here we worry about
statistical significance of the results, so it is typically best if we can
run multiple simulations for comparison (with different initial
conditions such as random seed, ion positions, ...). In single
trajectories block averaging can be performed (comparing the distributions
or progress variables over different regions of the trajectory).

> can use SO Can I use FF10 forcefield? I have seen from Amber 11 manual
> that FF10 includes both of these...

Yes, ff10 is fine.


AMBER mailing list
Received on Fri Jul 08 2011 - 16:30:03 PDT
Custom Search