Open ChaptCarole opened 3 months ago
Hi,
This is a bug and have been reported at https://github.com/satijalab/seurat/issues/8561. Our team is trying to fix it.
In the meantime, a workaround would be running FindMarkers() using the "RNA" assay's data
slot instead of using SCT assay. In most cases you should get similar results especially if you are searching for cluster markers.
Hello,
For my analysis, I have 12 samples that I merged before performing the normalization with SCTransform. When pursuing the analysis, I saw that I could use only 3 samples to perform a good annotation of the clusters, because for the other samples, there was an abnormal expression of a lot of genes and it was complicated to conclude. So I splitted the merged seurat object and merge again only the 3 seurat object that were interesting for the annotation. Then, I wanted to check the markers of these clusters using FindAllMarkers(). Since the data was normalized with SCTransform, I had first to perform a PrepSCTFindMarkers(). However, each time I tried to run this, I got the error:
Found 3 SCT models. Recorrecting SCT counts using minimum median counts: 5574 | | 0 % ~calculating Error in .subscript.2ary(x, i, j, drop = TRUE) : subscript out of bounds
Here is my code:
Here is my sessionInfo()
I suspect that splitting then merging some samples without the others is the problem. So, is there any way to do this correctly ?
Thanks a lot for your help,