[gmx-users] g_covar on multiple simulations

[Bert de Groot]

Anton Feenstra wrote:
> 
> Hi,
> 
> Does anyone have experience in g_covar (essential dynamics) on multiple,
> combined, trajectories? My experience for single (or a few comined)
> simulations, is that only a few (<10, sometimes <5) eigenvectors capture
> about 60% of MSF, but in the case I'm studying now (Cyt.P450), 12 are
> needed for 60%, and 30 for 80%...!
> 
> Is this unusual, 

kind of, mostly it's more concentrated

> or have other people seen similar things?
> 

no. I've done plenty combined analyses and always have had the majority
of the MSF concentrated in a low-dimensional subspace. I can see two reasons
for your observation:

1) a fitting/alignment inconsistency. Are you sure that the (sub)set of
atoms you extract from each trajectory match exactly to each other and
to the reference (-s in g_covar) structure? Taking a few snapshots and
superimposing them with g_confrms usually helped me to identify such problems.
Large jumps from one trajectory piece to the next in the projections 
of the combined trajectory onto the eigenvectors might also be a sign.
2) your protein is kind of floppy and different modes are sampled in
different trajectories. Is the MSF more concentrated in PCA's from single
trajectories than in the combined trajectory?

--

Bert

____________________________________________________________________________
Dr. Bert de Groot

Max Planck Institute for Biophysical Chemistry
Computational biomolecular dynamics group
Am Fassberg 11 
37077 Goettingen, Germany

tel: +49-551-2011306, fax: +49-551-2011089

email: bgroot at gwdg.de
http://www.mpibpc.gwdg.de/abteilungen/071/bgroot
____________________________________________________________________________