Deconvolution of Sample Identity in Single-Cell RNA Sequencing via Genome Imputation
Ghosh, R.; Hugh-White, R.; Nassiri, F.; Zadeh, G.; Boutros, P. C.
Show abstract
BackgroundDroplet based single-cell RNA sequencing (scRNA-seq) is a powerful tool for measuring RNA abundance profiles at cell-specific resolution. Droplet-based barcoding technology allows sample multiplexing, thereby facilitating high-scale of single cell sequencing. The resulting processing complexity, sample contamination and the underlying chemistries can all contribute to cell mis-labelling and consequent spurious cell-to-sample assignment. Approaches for barcode-free de-multiplexing which leverage natural genetic variation have been developed, but generally require an external source of genotype information. ResultsWe propose a novel method to exploit genome imputation and clustering to assign cells to inferred donor groups in the absence of a priori genetic information. Using tumor-derived single-cell RNA-sequencing (scRNA-seq) data, our workflow successfully assigned individual cells to donor-of-origin with high concordance. ConclusionsThis imputation-clustering approach represents a quality-assessment and quality-control strategy for barcode-free single cell donor-origin deconvolution with the capacity to resolve cases of sample cross-contamination.
Matching journals
The top 9 journals account for 50% of the predicted probability mass.