How to interpret Clustree results?

Question

[leranwangcs]

Hi,

I'm trying to use Clustree to determine which resolusion is the best for my data but I have trouble to understand the results. Below is what I got from clustree:

I can see that from resolution 0.4 to 0.5 the clustering is pretty stable but from 0.5 to 0.6, there is one a cross happened in cluster 3 and cluster 6 which both of them point to cluster 5 of resolution of 0.6. Does this mean the resolution of 0.5 is enough?

And a further explanation of that would be very appreciated!!

Thanks! Leran

[LChart]

The clustree is not everything -- and probably not enough on its own to determine a "good" resolution. I see several points at which the cluster is stable and differentiated only by a single cluster splitting into 2 (1.2 -> 1.3 ; 0.7 -> 0.8 ; 0.4 -> 0.5).

I would visualize these three cuts to see if the cells looked over-clustered or bad -- in particular whether the small clusters are "sufficiently distinct".