I ran SoupX using .h5 files for a sample using "SCTransform" workflow in seurat for generating initial clusters. The result seemed a bit odd as few genes in a cluster were called as 0 counts, but in reality the raw counts are in a range of 15-30. Then I ran "NormalizeData" workflow in seurat for clustering and the SoupX results are close to my expectation of how my data should look like. I need to understand what went wrong while using "SCTransform"? I have seen tutorials where they have used "SCTransform" without facing issues. As my downstream analysis relies on "SCTransform" in seurat I was wondering using "NormalizeData" for SoupX and then changing the workflow is a permissible approach?
I ran SoupX using .h5 files for a sample using "SCTransform" workflow in seurat for generating initial clusters. The result seemed a bit odd as few genes in a cluster were called as 0 counts, but in reality the raw counts are in a range of 15-30. Then I ran "NormalizeData" workflow in seurat for clustering and the SoupX results are close to my expectation of how my data should look like. I need to understand what went wrong while using "SCTransform"? I have seen tutorials where they have used "SCTransform" without facing issues. As my downstream analysis relies on "SCTransform" in seurat I was wondering using "NormalizeData" for SoupX and then changing the workflow is a permissible approach?