Total-mRNA-Aware Analysis for Droplet-Based Single Cell Sequencing

Abstract Controlling total mRNA content differences between cell populations is critical in comparative transcriptomic measurements. Due to poor compatibility with ERCC, a good control for droplet-based scRNA-seq is yet to be discovered. Normalizing cells to a common count distribution has been adopted as a silent compromise. Such practice profoundly confounds downstream analysis and mislead discoveries. We present TOMAS, a computational framework that derives total mRNA content ratios between cell populations via deconvoluting their heterotypic doublets. Experiments showed that cell types can have total mRNA differences by many folds and TOMAS can accurately infer the ratios between them. We demonstrate that TOMAS corrects bias in downstream analysis and rectifies a plethora of previously counter-intuitive or inconclusive analytical results. We argue against the opinion that doublets are undesired scale-limiting factors and revealed the unique value of doublets as controls in scRNA-seq. We advocate for their essential role in future large-scale scRNA-seq experiments.