scSTAR reveals hidden heterogeneity with a real-virtual cell pair structure across single cell samples

Abstract We present scSTAR, single-cell States Transfer Across-sample of RNA-seq data, which estimates individual cell state dynamics by creating real-virtual cell pairs across samples/conditions, therefore each cell can act as its own control when comparing between conditions. Using simulated and published scRNA-seq datasets, we show that new cell subtypes/functions which are more linked to the biological problem under investigation are discovered, especially when current methods give blur patterns.