I have several scRNA-seq datasets of tissue immune cells obtained from FACS sorting. When analyzing the data, I find when a dataset is rich in myeloid cells (e.g. macrophage), there tends to be a cell population that exhibits low-quality features. In addition, they are also featured with doublet gene expression (low expression of marker for several cell types), though I have performed the doublet removing.
I chose to remove this population when first having this observation. But after seeing these "low-quality" cells in several tissues, I started to worry about whether it was right to remove them. Does anyone come across this phenomenon when analyzing FACS-sorted immune cell scRNA-seq data?