[AMBER] clustering with ptraj

From: Francesco Gentile <fgentile.ualberta.ca>
Date: Thu, 19 May 2016 18:08:37 -0600

Hello everyone,
I am trying to perform clustering of MD trajectories using ptraj. I am
measuring the DBI and the SSR/SST by varying the number of clusters from 1
to 200. I tried two things that seems (at least to me) the same but I got
completely different results:
1) loading the trajectories by taking a snapshot every five, removing water
and ions, rms fitting to all the backbone atoms and then clustering in the
same ptraj input file. This leads to an optimal value (by combining a DBI
local minimum and the elbow criterion to the ratio) of 116 cluster.
2) loading the trajectories with a snapshot every 5, rms fitting and
writing out a fitted trajectory in one file. Loading the fitted one and
clustering in another one. This save of course time but gives a different
result, with DBI values exceptionally low (starting at 0.15 and going down
to 0.06 in the minimum).

Does anyone have a clue about that? I though that both methods were the
same but it seems like it is not like this. Best regards

Francesco Gentile
PhD Student, Biophysics
CCIS 3-215, Department of Physics
7-112, Li Ka Shing Centre for Health Research Innovation
University of Alberta, Edmonton, AB T6G 2E1
AMBER mailing list
Received on Thu May 19 2016 - 17:30:03 PDT
Custom Search