The cell subpopulation that I am interested in only accounts for around 1.2% of the total cells. I have previous FACS experiments that sort out the subpopulation from the samples (using markers CD166+ CD49fhi CD104- Lin-, see images). May I ask if there are any ways to identify this cell population from single-cell RNA-seq data of the tumor epithelial samples using the marker information?
Run the usual steps including clustering and visualization via something like UMAP and then color the UMAP by these four markers. That assumes that the surface marker separation you see in flow holds true on a transcriptional level which is not necessarily the case. For example, hematopoietic surface-defined populations such as the Granulocyte-Macrophage Progenitor (GMP) are well-separated in flow but on transcriptional single-cell cell level span many clusters.
Alternatively, check the tool
UCell which can calculate a score per cell based on a combination of positive and negative markers. You can then color the UMAP by this to see whether a single cluster stands out or whether this celltype contributes to many clusters ir is not even present at all. Include some marker combinations that you know are oresent and some nonsense combination that must be absent to have positive and negative controls. That will give you a feel whether that score for your cells is reliable.
This all assumes that the cells are even in the dataset. Depending on cell size and frequency (underrepresentation) they might even have dropped out during library prep.