Loss of clusters during pbDS

[bkheira]

Hi everyone,

I am analyzing an immune dataset with 11 clusters.

When I run pbDS on this dataset, the results are available for only 9 clusters, even if I set the min_cells parameter at 1.

I am missing the B cells, which I am interested in, and the fibroblasts.

Do you know if there is something I am missing ?

Many thanks
Kheira

[HelenaLC]

Could you post the output of table(sce$kid, sce$sid) and table(sce$sid, sce$gid) or something like that? Where s/k/g = sample,cluster,group; thanks! Also, are you running a paired design (e.g., including patient identifiers) or just any samples in group A vs. any samples in group B?

[bkheira]

Hi Helena, thank you for your quick answer.
Here are the infos :

> table(sce$cluster_id, sce$sample_id)
                   
                     2-2  2-5  3-2  3-5 
  NK & ILCs          317  235  379  428
  Monocytes         3518 3713 2758 3026
  DC                 111  128   94   97
  Macrophages        302  267  172  158
  Endothelial cells   24   27   16   20
  Neutrophils          9   22    7    7
  Fibroblasts          3    0    0    1
  NKT                  8   14   28   32
  T cells             32   81   52   47
  B cells             20   73   47   54
  Other                4    3    2    3

> table(sce$sample_id, sce$group_id)
         
          Control Endometriosis
  2-2        4348             0
  2-5        4563             0
  3-2           0          3555
  3-5           0          3873

I am comparing just any samples in group control vs. any samples in group endometriosis. Honnestly I was not expecting any results in the group Fibroblasts, but maybe I can get something with the few B cells ?

Many thanks

[bkheira]

I tried adding the parameter "filter" on genes and now I have all my clusters in the results of pbDS. I don't know why but some low gene expression in the B cell cluster was probably not allowing it to be in the results.

Many thanks Helena for your help and for this package.