[gmx-developers] Bug in g_cluster/gmx_cluster

[Berk Hess]

>The program runs ok, but the cluster centers that are written to
>cluster.log/clusters.pdb are not the correct cluster centers.

Ah, I somehow did not recompile completely.
Now I get the correct results.
This was a quite serious bug as all cluster avg rmsd's and cluster
middle structures were reported incorrectly.
Thanks for reporting this.

>As far as I recall, in full linkage clustering a structure would be added
>to a cluster if its distance to *all* other cluster members is below the
>cutoff. What is advertised in the g_cluster -h help is correct where it
>says "add a structure to a cluster when its distance to *any* element of
>the cluster is less than cutoff" (to me as a non-native english speaker
>this  means that the distance to only one cluster member needs to be below
>the cutoff  for a structure to be added to a cluster. Correct me if I'm
>wrong. At least that's what is done by the function 'gather', and this I
>think is called single linkage clustering).

OK, I checked on the net and indeed it should be called single linkage.
I fixed it in the help.

When running g_cluster without options I still get::
Writing rms distance/clustering matrix Fatal error: Can not plot more than 
16 discrete colors
I hope somebody else can fix this.

Berk.

_________________________________________________________________
Protect your PC - get McAfee.com VirusScan Online 
http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963