NUMT Confounding Biases Mitochondrial Heteroplasmy Calls in Favor of the Reference Allele

Abstract

Homology between mitochondrial DNA (mtDNA) and nuclear DNA of mitochondrial origin (nuMTs) causes confounding when aligning short sequence reads to the reference human genome, as the true sequence origin cannot be determined. Using a systematic in silico approach, we here report the impact of all potential mitochondrial variants on alignment accuracy and variant calling. A total of 49,707 possible mutations were introduced across the 16,569 bp reference mitochondrial genome (16,569 × 3 alternative alleles), one variant at-at-time. The resulting in silico fragmentation and alignment to the entire reference genome (GRCh38) revealed preferential mapping of mutated mitochondrial fragments to nuclear loci, as variants increased loci similarity to nuMTs, for a total of 807, 362, and 41 variants at 333, 144, and 27 positions when using 100, 150, and 300 bp single-end fragments. We subsequently modeled these affected variants at 50% heteroplasmy and carried out variant calling, observing bias in the reported allele frequencies in favor of the reference allele. Four variants (chrM:6023A, chrM:4456T, chrM:5147A, and chrM:7521A) including a possible hypertension factor, chrM:4456T, caused 100% loss of coverage at the mutated position (with all 100 bp single-end fragments aligning to homologous, nuclear positions instead of chrM), rendering these variants undetectable when aligning to the entire reference genome. Furthermore, four mitochondrial variants reported to be pathogenic were found to cause significant loss of coverage and select haplogroup-defining SNPs were shown to exacerbate the loss of coverage caused by surrounding variants. Increased fragment length and use of paired-end reads both improved alignment accuracy.

Complete vertebrate mitogenomes reveal widespread gene duplications and repeats

“…The high level of similarity is supportive of the overall quality of the mitoVGP assembly, which is also confirmed by its Q44. 30 Overall, over 50% of mitoVGP assemblies (52/100) presented one or more repeats (N = 45) and/or duplications (N = 18) ( Fig. 3a, Supplementary Table 1, columns AJ-AQ).…”

“…Theoretically, whenever repeats longer than the reads are present, assemblies are limited within the boundaries of repetitive elements. Another important challenge for mitogenome assembly is posed by the nuclear DNA of mitochondrial origin (NUMT) 29,30 . NUMTs originate from the partial or complete transposition of the mtDNA into the nuclear genome, potentially leading to multiple copies of the mtDNA sequence scattered throughout it and independently evolving 31 .…”

An Innovative Data Analysis Strategy For Accurate NGS Detection of Tumor mtDNA Mutations

“…The rst is mapping to the human reference mtDNA sequence rCRS, and the second is mapping to rCRS and hg19 (human genome 19) that contributes to remove NUMTs during analysis. Therefore, the accuracy of mtDNA mutation calling can be largely affected by mapping strategies with different selection of reference genomes and mismatch number, owing to existence of sequence similarity between NUMTs and mtDNA (a known source of confounding in mtDNA NGS studies) (24). However, there are few studies focusing on the applicability of the two mapping strategies under different circumstances.…”

Haplocheck: Phylogeny-based Contamination Detection in Mitochondrial and Whole-Genome Sequencing Studies

“…Nuclear DNA of mitochondrial origin Nuclear DNA of mitochondrial origin (NUMTS) can either result in a coverage drop on mtDNA sites due to the alignment of mitochondrial reads to NUMTS or false positive heteroplasmy calls due to the alignment of NUMT reads to the mitochondrial genome (Maude et al 2019) .…”

“…Nevertheless, sufficient coverage for the haplogroup defining variants is still required when dealing with NUMTS. In a study conducted by (Maude et al 2019) , an in-silico model has been set up to analyze the homology between mitochondrial variants and NUMTS. They show that 29 variants representing haplogroups A, H, L2, M, and U did not cause loss of coverage, nevertheless substantial loss of coverage has been identified for specific sites (e.g.G1888A, A4769G).…”