Thanks for the detailed instruction of this package!
I have a question about merging similar clusters, which seems to relate to the Step 10 and 14
First, I don't quite understand the purpose of making the cluster dendrogram in Step10, which seems to be different clustering method from KNN graph based clustering. Only the major cell groups in the cluster dendrogram are comparable to the graph based clustering and hard to merge similar small clusters based on this step. Is this just an extra step to look at the overall clustering?
Then I look into Step14 for the criteria of merging similar clusters. I found the FDR plot seems to be more helpful than Z score map. The reasons are as follows:
some of my clusters with very high FDR score indicating I should merge with other clusters but based on Z-score this cluster still stands out. So, should I look at FDR more closely for merging small similar clusters?
I am not sure whether my understanding is correct. Thanks a lot for help!
Hi SnapATAC team,
Thanks for the detailed instruction of this package!
I have a question about merging similar clusters, which seems to relate to the Step 10 and 14
First, I don't quite understand the purpose of making the cluster dendrogram in Step10, which seems to be different clustering method from KNN graph based clustering. Only the major cell groups in the cluster dendrogram are comparable to the graph based clustering and hard to merge similar small clusters based on this step. Is this just an extra step to look at the overall clustering?
Then I look into Step14 for the criteria of merging similar clusters. I found the FDR plot seems to be more helpful than Z score map. The reasons are as follows: some of my clusters with very high FDR score indicating I should merge with other clusters but based on Z-score this cluster still stands out. So, should I look at FDR more closely for merging small similar clusters?
I am not sure whether my understanding is correct. Thanks a lot for help!
Yang