search

BAUDOTlab / scDataPipeline

Reusable standard single-cell RNA sequencing data analysis pipeline
MIT License
2 stars 0 forks source link

Leiden clustering fails for datasets with ~50K cells #113

Closed celinech closed 5 months ago

celinech commented 5 months ago

I have a dataset with 49235 cells from the integration of 2 datasets. When I try to run the process step on the combined data, the step of the clustering fails.

The command to run the clustering is the following:

SO <- FindClusters(
    SO,
    resolution = 1,
    algorithm = 4,
    random.seed = 17,
    verbose=FALSE
)

The error message is this one:

Error in unlist(object) : 
  long vectors not supported yet: ../../src/include/Rinlinedfuns.h:537
Calls: lapply ... FindClusters.default -> RunLeiden -> leiden -> leiden.matrix -> unlist

After a quick research, I found on GitHub another issue, similar to it. The solution they suggest https://github.com/satijalab/seurat/issues/2294#issuecomment-771333280 is to change the implementation method for big datasets