search

broadinstitute / infercnv

Inferring CNV from Single-Cell RNA-Seq
Other
566 stars 166 forks source link

infercnv.observations_dendrogram.txt contains only part of the cell information to be predicted #336

Closed hah2468 closed 2 years ago

hah2468 commented 3 years ago

Thanks for developing this great software. But when I use it, I divide all the cells to be inferred into two categories, and select the parameter cluster_by_groups=F. I want to get the clustering results of all the cells to be inferred（all observations）, but there is only one type of cell information in infercnv.observations_dendrogram.txt. What should I do, or can I get the clustering results from other files?

GeorgescuC commented 3 years ago

Hi @hah2468 ,

What clustering information are you looking for, the dendrogram/hclust or the results of the subclustering mode?

If you are looking for the dendrogram/hclust, the file you are looking at will contain the tree in the commonly used newick format. If you are looking for subclustering, done through leiden clustering or random trees for example, you need to run infercnv with the HMM enabled HMM=TRUE and the analysis mode set to subclusters analysis_mode="subclusters". You can then find the cluster to cell table in 17_HMM_predHMMi6.*.hmm_mode-subclusters.cell_groupings with the asterisk being the subclustering method (default is leiden).

You can also access the results directly in R by looking at the infercnv_obj@tumor_subclusters slot in the final object.

Regards, Christophe.