How to choose the number of cluster in single cell multiome (RNA & ATAC)

Question

[Chris]

Hi all,

I am following some vignettes such as this one: https://satijalab.org/seurat/articles/weighted_nearest_neighbor_analysis, but I don't know the parameter to group cell into 2 or 3 clusters only instead of around 10 clusters automatically. The samples I have supposed to have only two cell types. Would you please have a suggestion? I appreciate your input!

[Ming Tommy Tang]

lower the resolution to small values such as 0.1 may work

Answer 1

[bk11]

You can run FindClusters with different resolutions and choose one that makes sense to you-

FindClusters(bm, graph.name = "wsnn", algorithm = 3, resolution = c(01, 0.2, 0.4, 0.8, 1.2, 1.4, 1.6, 1.8, 2), verbose = FALSE)

[Chris]

Thank you so much! It worked. I don't know why RNA + ADT, the vignette mention resolution but in RNA + ATAC, it is not. Do we have any good practice for choosing resolution?