I have a scRNA dataset that I'm analyzing, and have integrated cells from four different conditions following the Seurat vignette (Immune Integration). My levels of the combined data looks like:

levels(all.combined) [1] "CiliatedCell/Goblet_TAZE" "Neuronal_TAZE"
[3] "Neuroendocrine_TAZE" "CilliatedCell/CyclingBasal_TAZE"
[5] "MSC_TAZE" "Goblet/Tuft_TAZE"
[7] "Goblet/TransitAmplifyingCell_TAZE" "Neuroendocrine_TAZEH"
[9] "CiliatedCell/Goblet_TAZEH" "Goblet/Tuft_TAZEH"
[11] "CilliatedCell/CyclingBasal_TAZEH" "Neuronal_TAZEH"
[13] "MSC_TAZEH" "Goblet/TransitAmplifyingCell_TAZEH"
[15] "Neuronal_Hyp" "Neuroendocrine_Hyp"
[17] "CilliatedCell/CyclingBasal_Hyp" "CiliatedCell/Goblet_Hyp"
[19] "MSC_Hyp" "Goblet/TransitAmplifyingCell_Hyp"
[21] "Goblet/Tuft_Hyp" "Neuroendocrine_Control"
[23] "Neuronal_Control" "CiliatedCell/Goblet_Control"
[25] "Goblet/TransitAmplifyingCell_Control" "CilliatedCell/CyclingBasal_Control"
[27] "MSC_Control" "Goblet/Tuft_Control"

Each of these labels have the cluster name first, and then treatment condition after an underscore (e.g. Goblet/Tuft_Hyp are Goblet/Tuft cells in Hyp condition).

I have identified a mesenchymal stem cell population that interests me (e.g. labeled 'MSC')--I would like to be able to subset the MSC subset based on a specific gene expression (e.g. Egfr+ or Egfr-) and perform FindMarkers that separate these two groups in different settings (e.g. _Control, _Hyp, _TAZE, etc). However, if I create two subsets, e.g.

egfr_pos = subset(combined.data, subset = Egfr > 0.9)

egfr_neg = subset(combined.data, subset = Egfr < 0.1)

Then I can't perform FindMarkers on this (as it requires identies within an existing group). How do I go about this?