MQuad enables clonal substructure discovery using single cell mitochondrial variants

Abstract

Mitochondrial mutations are increasingly recognised as informative endogenous genetic markers that can be used to reconstruct cellular clonal structure using single cell RNA or DNA sequencing data. Novel sequencing protocols have been developed to enrich mitochondrial content for a deep and uniform coverage. However, there is a lack of effective computational methods to identify informative mtDNA variants in noisy and sparse single-cell sequencing data. Presented here is an open-source computational tool MQuad that accurately calls clonally informative mtDNA variants in a population of single cells, and an analysis suite for complete clonality inference, based on single cell RNA or DNA sequencing data. Through a variety of simulated and experimental single cell sequencing data, MQuad is shown to be able to identify clonal mitochondrial variants with both high sensitivity and specificity, outperforming existing methods by a large extent. Furthermore, its wide applicability is demonstrated in different single cell sequencing protocols, particularly in complementing single-nucleotide and copy-number variations to extract finer clonal resolution. Lastly, it is shown here that MQuad can be used on not only cancer datasets, but also in normal cell types such as fibroblasts to infer clonal structures with mitochondrial variants. MQuad is a Python package available via https://github.com/single-cell-genetics/MQuad.