[Chimera-users] How to know the cut-off used in ensemble cluster?
conrad at cgl.ucsf.edu
Mon Jun 1 14:40:17 PDT 2020
As you mentioned, the cutoff is specific to each data set. As described
in the NMRCLUST paper, the cutoff is based on a "penalty value" as a
function of the number of clusters. The penalty value is the sum of the
average "spread" (distance between two samples within a cluster) and the
number of clusters. We simply choose the number of clusters as the one
with the lowest penalty value. I did not bother reporting the actual
penalty value because it does not really correspond to a physical
quantity, and the chosen penalty value is not useful without the context
of all other penalty values.
If I were doing it over, I would probably include a penalty-vs-#cluster
plot and allow users to change the cutoff, but we have moved on to
ChimeraX development and, sadly, there is not enough time to do
everything we want.
On 6/1/2020 10:16 AM, Ibrahim Mohamed wrote:
> i am using MD movie to cluster my Molecular Dynamics results. how can i
> know the cut-off used by chimera to do this? i know that it depends on
> the data itself, but how can i know this cut-off for specific data?
> Chimera-users mailing list: Chimera-users at cgl.ucsf.edu
> Manage subscription: https://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users
More information about the Chimera-users