MitoH3: Mitochondrial Haplogroup and Homoplasmic/Heteroplasmic Variant Calling Pipeline for Alzheimer's Disease Sequencing Project

Abstract

Background: Mitochondrial DNA (mtDNA) is a double-stranded circular DNA and has multiple copies in each cell. Excess heteroplasmy, the coexistence of distinct variants in copies of mtDNA within a cell, may lead to mitochondrial impairments. Accurate determination of heteroplasmy in whole-genome sequencing (WGS) data has posed a significant challenge because mitochondria carrying heteroplasmic variants cannot be distinguished during library preparation. Moreover, sequencing errors, contamination, and nuclear mtDNA segments can reduce the accuracy of heteroplasmic variant calling.

Objective: To efficiently and accurately call mtDNA homoplasmic and heteroplasmic variants from the large-scale WGS data generated from the Alzheimer's Disease Sequencing Project (ADSP), and test their association with Alzheimer's disease (AD).

Methods: In this study, we present MitoH3-a comprehensive computational pipeline for calling mtDNA homoplasmic and heteroplasmic variants and inferring haplogroups in the ADSP WGS data. We first applied MitoH3 to 45 technical replicates from 6 subjects to define a threshold for detecting heteroplasmic variants. Then using the threshold of 5% ≤variant allele fraction≤95%, we further applied MitoH3 to call heteroplasmic variants from a total of 16,113 DNA samples with 6,742 samples from cognitively normal controls and 6,183 from AD cases.

Results: This pipeline is available through the Singularity container engine. For 4,311 heteroplasmic variants identified from 16,113 samples, no significant variant count difference was observed between AD cases and controls.

Conclusions: Our streamlined pipeline, MitoH3, enables computationally efficient and accurate analysis of a large number of samples.

Keywords: Alzheimer’s disease; haplogroup; homoplasmic and heteroplasmic variant calling; mitochondrial DNA; whole genome sequencing.

Fig. 1

Sum of inconsistent ratios of (A) Homoplasmic calls and (B) Heteroplasmic calls among technical replicates of each subject (n = 6) according to 5 different VAF thresholds.

Fig. 2

MT haplogroup distribution in African American (AA), Caribbean Hispanic (CH), and European ancestry (EA) cohorts. Bolded numbers indicate that the haplogroup is more predominant in the particular ancestry group compared to other groups. Red-highlighted numbers signify that the haplogroup constitutes more than 5% within the ancestry, while green-highlighted numbers indicate that the haplogroup comprises less than 5%, yet the frequency is 5-fold greater than in other ancestral groups at least.

Fig. 3

Among individuals in the total sample, (A) variant allele fraction (VAF) distribution of mtDNA variant calls and (B) proportion of heteroplasmic calls and homoplasmic calls.

Fig. 4

MT variant distributions in the total sample. A) Proportion of variants by plasmy status. B) Number of variants detected according to allele frequency strata. C) Proportion of variants by plasmy status in each frequency strata.

Fig. 5

Boxplots showing the mean number of homoplasmic calls according to (A) ancestry group and (B) tissue type among European ancestry individuals. Bar plots show the proportion of heteroplasmic calls according to (C) ancestry group and (D) tissue type among European ancestry individuals.