What is the difference between FlowSOM_cluster and FlowSOM_metacluster? #93
-
|
Hi there, Thank you again for this well-constructed and explained package. But, what is the difference between FlowSOM_cluster and FlowSOM_metacluster? And, when to use which one? 😅 |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 1 reply
-
|
@denvercal1234GitHub excellent question! Essentially FlowSOM creates the clusters initially, and then groups them together into metaclusters. There are a number of reasons for this, but fundamentally it is about providing levels of resolution -- the clusters are far more granular and are finding subtle changes in data structure, metaclusters will (hopefully) collect those into groups that more closely resemble 'populations'. I would focus mainly on the metaclusters, where the clusters can be used in some situations to dive deeper where required. |
Beta Was this translation helpful? Give feedback.
-
|
Just to increment Thomas's answer, you have to take care of the ratio between the number of clusters to the number of meta-clusters. 30 or 40 meta-clusters might be the upper limit for 100 clusters. If you want more meta-clusters, you have to increase the number of meta-clusters : for example a grid of 14x14 = 196 clusters for 60 meta-clusters. This still depends on the dataset and trials... |
Beta Was this translation helpful? Give feedback.
-
|
Thank you! 😊 |
Beta Was this translation helpful? Give feedback.
@denvercal1234GitHub excellent question! Essentially FlowSOM creates the clusters initially, and then groups them together into metaclusters. There are a number of reasons for this, but fundamentally it is about providing levels of resolution -- the clusters are far more granular and are finding subtle changes in data structure, metaclusters will (hopefully) collect those into groups that more closely resemble 'populations'.
I would focus mainly on the metaclusters, where the clusters can be used in some situations to dive deeper where required.