Hi all,
I have carried out some clustering through dist and hclust. This works well and I am happy with the results produced but i was wondering if there is a way of plotting the clustering PLUS annotating the plot with the samples that cause the clustering...
Context
This is hierarchical clustering of neuron activations in a layer obtained from a neural network. I would like to cluster the neuron activations together to see which are activated similarly. I would like to visualize which of my samples cause those neuron activations (and how highly)...
I have attached a plot of the results of the hierarchical clustering. the plot doesn't have
to be circular, it could be a normal dendogram...
But for each group (on the outside of the circle) i would like to be able to annotate the samples that actually cause that clustering and also maybe with some form of heatmap show the neuron activation score for those samples.
Has anybody done anything similar to this/ know a way of annotating a plot like this with the samples that cause the clustering and the scores that per sample that are causing this clustering? any suggestions would be highly appreciated!
Additionally, the dendogram is produced from a similarity matrix of a 100 x 100 marix.
the labelling exists in a separate data.frame of 140 by 100 (140 is the sample number) with their respective scores for how much they activate a node)... so the labelling and neuron score comes from a different dataframe for labelling...
many thanks!!
circle plot.pdf (15.8 KB)