AMBER Archive (2009)

Subject: Re: [AMBER] Best rms fit for correlation and covariance analysis

From: Bill Ross (
Date: Tue Apr 28 2009 - 11:57:10 CDT

> I was wondering the best way to do an RMS fit on all MD trajectories to
> obtain correlation and covariance matrices to remove the rotational and
> translational degrees of freedom to the best extent.

If you do an rms fit, that removes the overall translation and rotation
from the rms result.

> My simulations
> include a two domain complex which is connected by a flexible linker
> region and non-bonded contacts. I want to analyze the effect of these
> non-bonded contacts through the analysis of inter-domain correlations
> and projection of quasi harmonic modes on a variety of stuff...

Fitting on one section and measure resulting rms on another may be useful.

> I was thinking about using these structures for reference...
> - Minimized pre-equilibrated structure
> - Water & ion equilibrated structure @ 300K (the last snapshot before MD
> runs)
> - Average structure calculated using the whole MD trajectory

Bear in mind that the average might not be a physically possible structure,
e.g. atoms may be superimposed.

> And these alignments
> - All CA's
> - CA's in structured core subdomains in either domain of the complex
> - Selected residues with the lowest B-values
> - Selected residues from double-difference plots
> I tried combining some of these and obtained different results and quite
> confused on which way to go... If you can provide me any insight I will
> be grateful. If you provide me a better way than I thought, I will be
> more grateful.

How did you combine what? How did the results differ?


AMBER mailing list