Good morning,

I wonder if anyone can answer this question: If I subset my Seurat object into celltype A and celltype B and then I merge it back together I see some differences in the counts@p. Why is this the case?

Please find below some pictures attached for illustration:

Before subsetting:

After subsetting into celltypes A (Macrophages) and Celltype B (all other cells):

Celltype A (Macrophages) :

Celltype B (all other cells):

Newly merged object:

The UMAP cluster seem to look as before however in slightly different positions.

However, comparing the counts I can see a difference:

Before:

After:

I had to re-run FindVariableFeatures, ScaleData, RunPCA, FindNeighbors, FindClusters and RunUMAP in the newly merged object since all reductions, commands were gone. However, I repeated the commands exactly in the way as before.

Any idea what is causing this difference? Thank you for any hint!