souporcell: Robust clustering of single cell RNAseq by genotype and ambient RNA inference without reference genotypes

Abstract

Methods to deconvolve single-cell RNA sequencing (scRNAseq) data are necessary for samples containing a natural mixture of genotypes and for scRNAseq experiments that multiplex cells from different donors¹. Multiplexing across donors is a popular experimental design with many benefits including avoiding batch effects², reducing costs, and improving doublet detection. Using variants detected in the RNAseq reads, it is possible to assign cells to the individuals from which they arose. These variants can also be used to identify and remove cross-genotype doublet cells that may have highly similar transcriptional profiles precluding detection by transcriptional profile. More subtle cross-genotype variant contamination can be used to estimate the amount of ambient RNA in the system. Ambient RNA is caused by cell lysis prior to droplet partitioning and is an important confounder of scRNAseq analysis³. Souporcell is a novel method to cluster cells using only the genetic variants detected within the scRNAseq reads. We show that it achieves high accuracy on genotype clustering, doublet detection, and ambient RNA estimation as demonstrated across a wide range of challenging scenarios.