MitoMut: An Efficient Approach to Detecting Mitochondrial DNA Deletions from Paired-end Next-generation Sequencing Data

C. Shane Elder, Catherine E. Welsh

Abstract: Next-generation sequencing technologies paired with bioinformatics analysis pipelines are necessary to elucidate the causes and functional consequences of mitochondrial DNA (mtDNA) deletions. Various methods exist that accurately detect and quantify deletions in nuclear DNA, but due to special characteristics of mtDNA, these methods are inaccurate in the detection of mtDNA deletions. Recently, a few tools have emerged specifically for detecting deletions in mtDNA. However, these tools often require large amounts of RAM and have significant run times. For a scientist using a personal computer to analyze their data without access to a high-performance cluster, these limitations can be barriers to use. Therefore, efficient tools for detecting mtDNA deletions are necessary. Here we present MitoMut, a tool capable of efficiently and effectively analyzing mitochondrial deletions while maintaining user freedom. MitoMut runs easily on personal computers and scales well for use in large bioinformatics pipelines. We thoroughly tested MitoMut on simulated and real data, showing that it can detect deletions at low heteroplasmy ($<$ 1%) with high precision and recall. MitoMut consistently outperforms alternate approaches in both time and space efficiency, as well as overall sensitivity. In addition, we report the results of MitoMut for 970 phase 3 1000 Genomes Project samples. MitoMut is implemented in Python and is freely available at https://github.com/shane-e945/MitoMut.