Is reclustering of a cluster obtained from hierarchical clustering "allowed"?

Question

[Jonathan Lefebre]

Hey all,

I obtained a clustermap from hierarchical clustering of pairwise distances of atoms in an protein alignment. Using a specific cut-off for cluster selection, i found several clusters that "biologically" speaking make sense to me. However, I also obtained one very big cluster that seems to be rather composed of several clusters, which do not appear as single clusters at the cut off i used. So here is my question:

Is it allowed to extract this cluster and cluster this again? By allowed I mean, does it make sense or am I undermining the principal of hierarchical clustering by reclustering?

Best,

Jonathan

[German.M.Demidov]

There are no fundamental principles behind clustering. Do whatever suits your ideas.

Seriously, it is a very non-perfect and poorly formulated tool since 2D matrices of distances are most of the time impossible to properly decompose into clusters. Think of it as about an instrument, not as a "theory" or smth like that