Re: [AMBER] PCA analysis for side chain atoms and backbone atoms

From: Jorgen Simonsen <>
Date: Wed, 26 Jan 2011 12:12:10 +0100


thanks - so I tried to rms fit using the following syntax - the
trajectory has been centered

center :1-100 mass origin

rms first mass

and then do a entropy calculation

matrix mwcovar name cc.matrix out cc.matrix
analyze matrix cc.matrix out new.matrix thermo

but now I start to get NaN and some very strange numbers

 warning: setting vibrational entropy to zero for mode 320
with vtemp = -118786300.372960
 warning: setting vibrational entropy to zero for mode 321
with vtemp = -4988665.26942776

Is the syntax wrong? So I do have diffusion and different
conformations in my sample which will probably have a very bad rms

On Wed, Jan 26, 2011 at 11:27 AM, Hannes Loeffler
<> wrote:
> On Wed, 26 Jan 2011 10:52:42 +0100
> Jorgen Simonsen <> wrote:
>> Hi
>> Just to be quite sure - I think that I remove as much of the rotations
>> and translations by first aligning the atoms
>> center :1-100 mass origin
>> or for backbone .N,CA,C etc? I hope this is correctly understood.
> Centering alone will not remove the rotations.  What you should do is
> to RMS fit every frame with the same selection that you will use for the
> PCA analysis.
>> For the side chains, sc, versus backbone.
>> For sc, I would like to see how much is captured by the first and
>> second principal component and how big a "space" is expanded by this
>> side chains. Later, I would like to do quasiharmonic analysis and see
>> entropy contribution between bb and sc.
> You may have to run very long simulations if you want to quantify this
> difference.  Keep in mind that the entropy is dominated by
> low-frequency modes, i.e. the (possibly much) slower backbone motions.
> Those motions will also show up in the side chain analysis.  I am not
> sure how well both would cancel.
> _______________________________________________
> AMBER mailing list

AMBER mailing list
Received on Wed Jan 26 2011 - 03:30:02 PST
Custom Search