scMitoMut for calling mitochondrial lineage–related mutations in single cells

Abstract

AbstractTracing cell lineages has become a valuable tool for studying biological processes. Among the available tools for human data, mitochondria DNA (mtDNA) has a high potential due to its ability to be used in conjunction with single-cell chromatin accessibility data, giving access to the cell phenotype. Nonetheless, the existing mutation calling tools are ill-equipped to deal with the polyploid nature of the mtDNA and lack a robust statistical framework. Here we introduce scMitoMut, an innovative R package that leverages statistical methodologies to accurately identify mitochondrial lineage related mutations at the single-cell level. scMitoMut assigns a mutation quality q-value based on beta-binomial distribution to each mutation at each locus within individual cells, ensuring higher sensitivity and precision of lineage related mutation calling in comparison to current methodologies. We tested scMitoMut using single-cell DNA sequencing, scATAC sequencing and 10× Genomics single cell multiome datasets. Using a single-cell DNA sequencing dataset from a mixed population of cell lines, scMitoMut demonstrated superior sensitivity in identifying small proportion of cancer cell lines compared to existing methods. In a human colorectal cancer scATAC dataset, scMitoMut identified more mutations than state-of-the-art methods. Applied to 10× Genomics multiome datasets, scMitoMut effectively measured the lineage distance in cells from blood or brain tissues. Thus, the scMitoMut is a free available (https://www.bioconductor.org/packages/devel/bioc/html/scMitoMut.html.), well-engineered toolkit for mtDNA mutation calling with high memory and CPU efficiency. Consequently, it will significantly advance the application of single-cell sequencing, facilitating the precise delineation of mitochondrial mutations for lineage tracing purposes in development, tumor and stem cell biology.