RE: AMBER: Average structure

From: Ross Walker <>
Date: Mon, 21 Apr 2008 19:27:08 -0700

Hi James,

I would also like to point out that if by average structure you just mean
the average of the cartesian coordinates explored then you should think very
carefully about what this actually means and whether it is at all useful.

E.g. consider a freely rotating methyl group:

Then think about what the average structure for this would be. It is


I.e. all 3 protons sitting on top of each other along the vector that forms
the center of rotation - so not really that interesting. Much more
interesting is to do some kind of cluster type analysis or something that
plots say a color plot of RMSD from a given representative structure.

See the following tutorial: which goes over
average structures and then more advanced analysis techniques.

Good luck,

|\oss Walker

| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- |
| | PGP Key available on request |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.

> -----Original Message-----
> From:
> [] On Behalf Of David Cerutti
> Sent: Monday, April 21, 2008 09:13
> To:
> Subject: Re: AMBER: Average structure
> Hi James,
> The average structure from a given length of simulation
> is only that.
> 10s is not a long trajectory by current standards, but it's
> not terribly
> short either. If you're interested in an average structure,
> it should be
> the average structure for whatever length of trajectory you
> have, less
> some portion at the start of the run which you'll discard for
> "equilibration" (I use the quotes because there can be many
> definitions of
> qhat it means to be equilibrated, some of them more difficult
> to satisfy
> than others).
> It is conceivable to me that you might observe all of the
> different
> motions that you're going to see from this protein in the
> space of 10ns,
> just not in all possible combinations (maybe you see loop A
> flips out and
> back, loop B could curl up, and loop C twists or turns into a
> helix, but
> observing all of these different motions in every possible
> combination
> would take a long time). It may be helpful to determine if
> there is a
> significant "core" to the protein that remains very stable
> and can thus be
> used to align the trajectory, then construct some sort of
> library of all
> the loop motions that you see, and use the point at which
> your library is
> no longer growing as a metric for "convergent" behavior. I'm
> not sure how
> to do this conveniently in AMBER, though; I'm thinking of
> writing my own
> program for this type of analysis.
> The average structure can also serve as your basis for
> aligning the
> trajectory, but I like to first look for any parts of the
> protein that
> move very little so you reference point corresponds to
> something that is
> easily recognizable, which the average structure may not be.
> To remove the overall translation and rotation, you just
> align to some
> structure (it can be one frame of the trajectory, but in this
> case the
> average structure is often preferred). Again, if you can
> find some very
> stable residues in the central regions of your protein,
> aligning to those
> (or, the average conformation of those particular residues)
> might be a
> more intuitive way to look at the protein motions.
> Dave
> On Mon, 21 Apr 2008, James Thomas wrote:
> > Dear Amber Community
> >
> > I would like to inquire if average structure of MD run,
> after 10 ns for a
> > protein of ~100aa is reliable as a starting point for
> further analysis?
> > I heard from a post-doc who is experienced in Amber, that
> average structure
> > is not reliable for short duration simulation.
> > I look forward to your comments / advices.
> >
> > How to remove the overall translational & rotational motion
> in Amber9
> > trajectories?
> >
> > Many thanks.
> >
> --------------------------------------------------------------
> ---------
> The AMBER Mail Reflector
> To post, send mail to
> To unsubscribe, send "unsubscribe amber" to

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Apr 23 2008 - 06:07:41 PDT
Custom Search