Re: [AMBER] Drifting RMSD

From: Thomas Cheatham III <>
Date: Fri, 7 Jan 2011 12:00:48 -0700 (Mountain Standard Time)

> Following equilibration, and during the production run the backbone rmsd
> of the protein is drifting. I'm attaching two snapshots for easy
> reference. The equilibration run was over 6ns and the production run
> over 14ns. The files are identical.
> Any ideas regarding this behaviour?

Force field are not perfect; also, biomolecules are dynamic and have
motions on many time scales with different ranges of motion. For example,
you almost never expect a "zero" RMSd with respect to a given snapshot
since there are inherent fluctuations around the average structure(s).
For a relatively rigid structure that primarily adopts a well defined
structure (for example a high resolution crystal structure) you can
estimate this motion from the B factors. Many biomolecular systems,
however, have dynamic or disordered regions, for example in loops, at
the termini, or at domain interfaces. Note that 3A backbone RMSd is not
particularly large and likely (if a protein) the structure is largely

Like Jason just mentioned, the single number is sometimes misleading.
Perhaps you are simply seeing expected dynamics. This is where you need
to dig deeper to look at the structures in these regions to "see" what is
going on. Based on the RMSd, I would state that these simulations are not
converged. I would look at the structures over stable regions (0-500,
750-1100, ~1250, and at the end) and make a 2D-RMSd plot. Convergence
could be ascertained by multiple independent simulations and/or by
demonstrating that conformations are re-visited during the trajectory.


AMBER mailing list
Received on Fri Jan 07 2011 - 11:30:03 PST
Custom Search