A cluster robustness score for identifying cell subpopulations in single cell gene expression datasets from heterogeneous tissues and tumors
Kanter I, Dalerba P, Kalisky T.
Significance: The identification of stem cells populations is often made difficult by their low numbers in primary tissues, and by the lack of unique surface markers that can be used for their differential purification. Single-cell genomics technologies, such as singe-cell qPCR and single-cell RNA-seq, can be used to de-convolute the cell composition of primary tissues and the discovery of novel biomarkers with specific expression in stem cell populations, but are often limited in their resolution because of the high levels of noise intrinsic to such gene-expression measurements. This study describes a new mathematical approach that can aid in understanding whether novel cellular subtypes identified using single-cell genomics techniques can be considered robust (i.e. resistant to the confounding effects of noise).
Link:
Kanter et al., Bioinformatics, article in press (2018)