[gmx-users] How to do cluster analysis in the essential space after PCA?

[Anton Feenstra]

Zhenting Gao wrote:

> Hi  gmx-users ,
> 
> After PCA of the trajectory, a 2D(3D) graph with the first 2(3) eigenvectors with the largest eigenvalue is obtained.
> Then how should I cluster on this essential space?
> What exactly software will be chosen?

You could do part(s) of it yourself, I see two ways to do something like
what you describe:
1) write a filtered trajectory (using g_anaeig, I believe) and cluster
that using g_cluster. (But, I'm not sure if it makes much sense this way.)
2) calculate pairwise frame-distances in the 2D (or 3D) projections,
write those out in a matrix, and read that in g_cluster to cluster. (You
may need to tune carefully the settings for g_cluster, as the defaults
may not be optimal for this type of clustering.)

-- 
Groetjes,

Anton

* NOTE: New Phone & Fax numbers (below) *

 _____________ _______________________________________________________
|             |                                                       |
|  _   _  ___,| K. Anton Feenstra                                     |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|(   |   )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands   |
| \_/ \_/ | | | Tel: +31 20 59 87608 - Fax: +31 20 59 87610           |
|             | Feenstra at few.vu.nl - www.few.vu.nl/~feenstra/         |
|             | "If You See Me Getting High, Knock Me Down" (RHCP)    |
|_____________|_______________________________________________________|