Clustering of molecular dynamics structures : new program available for testing.

[Nicholas M. Glykos]

Dear All,

We have developed a new program for clustering structures from a trajectory
using the data produced from dihedral PCA or Cartesian PCA.

The new program, called cwords, is multidimensional and uses by default the
top _5_ principal components (but can go up to 10 dimensions if you want to).

The program is fast : it can cluster 6.3 million frames (using 5 PCs) in
less than 85 seconds.

The program is a drop-in replacement for 'cluster5D' which is distributed
together with the carma/grcarma. If you want to give it a try, and if you
have installed carma/grcarma/cluster5D in /usr/local/bin, then the
following commands from the unix shell will do the trick :

cd /usr/local/
sudo git clone https://github.com/glykos/cwords.git
cd bin/
sudo mv cluster5D cluster5D_ORIG
sudo ln -s /usr/local/cwords/cwords ./cluster5D
sudo apt install tcsh

The last command is only needed if you do not have tcsh already installed.

To use the new clustering procedure, go to grcarma ⇒ Dihedral PCA and do
everything as normal with the addition that you click the "Perform five
dimensional clustering" and increase the maximum number of clusters to,
say, 100. Leave the "Percentage of frames ..." set to 'Auto'.

Further information is available from the program's GitHub page.¹

Report problems, bugs & flames using this thread.

Kind regards,
Nicholas


¹ https://github.com/glykos/cwords

--


Nicholas M. Glykos, Department of Molecular Biology
and Genetics, Democritus University of Thrace, University Campus,
Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620,
Ext.77620, Tel (lab) +302551030615, https://utopia.duth.gr/glykos/