Dear all,

I recently measured the pairwise distances between pocket centers that I detected across an alignment of homologous proteins. I plotted the data in a correlation matrix and clustered them based on distance using hierarchical clustering as provided by seaborn. However, the matrix is quite hard to interprete because of the number of clusters. I would like to extract clusters from this matrix and apply a "cut-off" on what I would consider a cluster, i.e. I would like to define a cluster by atomic positions not further away from each other then 10 A.

Does anyone have experience with extracting cluster information from seaborn-based clustermatrices?

Any help is appreciated!

Best

Jonathan

you can filter a correlation matrix as this answer mention it: https://stackoverflow.com/questions/44889508/if-correlation-is-greater-than-0-75-remove-the-column-from-dataframe-of-pandas