BEARscc: robustness of single-cell clusters determined using simulated technical replicates

Technical variance is a major confounding factor in single-cell RNA sequencing, not least because measurements on the same cell are not replicable. We developed BEARscc, a tool that simulates experiment-specific technical replicates based on a probabilistic model of technical variance trained on RNA spike-in measurements. We demonstrate that the tool improves the unsupervised classification of cells and aids the interpretation of single-cell RNA-seq experiments. Overall design: scRNA-seq was performed on dilute whole brain tissue and water samples in order to assess BEARscc as cluster robustness tool.