[gmx-users] Re: Re:g_cluster output

[Mark Abraham]

sarbani chattopadhyay wrote:
> Hi,
> Actually I am not confused regarding the clustering method that is being 
> used in default. I
> know that the method that is used is "Single linkage " method.
> 
> I got to know the options that I had used in my "g_cluster" command 
> after going through
> g_cluster -h.
> 
> Regarding the command "g_cluster -cl -wcl" . it is written
> "-cl writes average (with option -av) or central structure of each 
> cluster or
> writes numbered files with cluster members for a selected set of 
> clusters (with option -wcl,
> depends on -nst and -rmsmin). "

OK that's great. The usual mode is for people not to bother reading 
documentation before they post on here about a command line they copied 
from some tutorial, so if you haven't clearly indicated that you have 
done some reading (and what you learned!), you rate to not get the best 
possible response. :-)

> My question is how the average structure is obtained. I had written that 
> it does not match
> with the members of the cluster, thus it signifies that it isn't writing 
> that cluster member
> which is the cluster centre( determined through some criteria).
> It is not necessary for the average structure to be physically present 
> in the cluster, but it is
> necessary to know how is it obtained. Moreover while going through the 
> mailing lists, it was
> found that the average structures written by g_cluster were doubtful, 
> though the Bug was
> reported to be removed.

30 seconds of reading the code suggests it's a straight arithmetic 
average of the coordinates of the members of the cluster once they've 
been fitted to the first member of the structure - roughly as you would 
expect. You could test this with some artfully chosen pseudo-clusters, 
of course.

Mark