Bioinformatic removal of NUMT‐associated variants in mitotiling next‐generation sequencing data from whole blood samples

Abstract

Nuclear mitochondrial DNA segments (NUMTs) have arisen because of the transposition of segments of the mitochondrial DNA genome (mitogenome) into the nuclear genome. When using a "mitotiling" strategy, NUMTs may be more readily amplified when targeting the entire mitogenome compared to the control region, as hundreds of primers are required for complete sequencing coverage. In samples with a high percentage of nuclear DNA copies per cell, such as whole blood, NUMT coenrichment may be exacerbated. The present study examined bioinformatic approaches for removing NUMTs and NUMT-associated variants (NAVs) from next-generation sequence data generated using two mitotiling kits (Precision ID and QIAseq). Across 16 samples with low mtDNA copy number, NUMT coenrichment produced 890 NAVs with >5% variant frequency. The use of the consensus sequence to eliminate NUMT reads proved to be effective for QIAseq data, and resulted in >85% NAV removal in Precision ID data. This method was bolstered by NAV filtering in Precision ID analysis. Alternative high stringency mapping to the revised Cambridge Reference Sequence (rCRS) and the human genome reference GRCh38 for the QIAseq data caused a reduction in mitogenome coverage without complete NUMT removal. These bioinformatic solutions facilitate mitotiling sequence data analysis for low-level variant detection.

Validation of NGS for mitochondrial DNA casework at the FBI Laboratory

“…However, these analyses have been largely manual, which is not suitable to a production environment. As bioinformatic approaches (for example, [58]) and particularly software packages with filtering capabilities (e.g. Converge™ Software -compatible with Thermo Fisher assays only, Thermo Fisher Scientific) continue to become available, and as the efficacy of such filters is clearly established via rigorous testing and validation, it is possible that analysis parameters can be refined and assay sensitivity improved even further.…”

Mitigating the effects of reference sequence bias in single-multiplex massively parallel sequencing of the mitochondrial DNA control region

“…In non-forensic applications numt sequences could interfere with correct reporting of mitochondrial DNA variants [46]. However, in the analysis of casework samples where low-level minor components are of particular interest [39], numt sequences could become visible, requiring removal [47], or may be falsely interpreted as heteroplasmies or mixtures.…”

Impact of the sequencing method on the detection and interpretation of mitochondrial DNA length heteroplasmy

“…Supplemental haplotype analyses of the SA Ion BAM files were performed using a custom IGV mtDNA tool (variant caller v1.01b) that was developed specifically for Ion sequencing of the Precision ID mtDNA Panels [42,61]. Median read depth was determined for each amplicon, and a 20X minimum read depth threshold was used for variant detection.…”

“…These low-level variants can be attributed to incomplete removal of SA PCR primers, Ion-associated indel errors [34], and/or sequences from nuclear mtDNA segments (NUMTs). NUMTs are more likely to be co-amplified using SA PCR compared to LR PCR due to the large number of primers and small amplicons that may target homologous pseudogenes in the nuclear genome [44,61]. The IGV mtDNA tool analyses eased interpretation of the SA Ion data by automatically trimming the SA PCR primers and filtering NUMT-associated variants.…”