Evaluation of mitochondrial DNA copy number estimation techniques

Abstract

Mitochondrial DNA copy number (mtDNA-CN), a measure of the number of mitochondrial genomes per cell, is a minimally invasive proxy measure for mitochondrial function and has been associated with several aging-related diseases. Although quantitative real-time PCR (qPCR) is the current gold standard method for measuring mtDNA-CN, mtDNA-CN can also be measured from genotyping microarray probe intensities and DNA sequencing read counts. To conduct a comprehensive examination on the performance of these methods, we use known mtDNA-CN correlates (age, sex, white blood cell count, Duffy locus genotype, incident cardiovascular disease) to evaluate mtDNA-CN calculated from qPCR, two microarray platforms, as well as whole genome (WGS) and whole exome sequence (WES) data across 1,085 participants from the Atherosclerosis Risk in Communities (ARIC) study and 3,489 participants from the Multi-Ethnic Study of Atherosclerosis (MESA). We observe mtDNA-CN derived from WGS data is significantly more associated with known correlates compared to all other methods (p < 0.001). Additionally, mtDNA-CN measured from WGS is on average more significantly associated with traits by 5.6 orders of magnitude and has effect size estimates 5.8 times more extreme than the current gold standard of qPCR. We further investigated the role of DNA extraction method on mtDNA-CN estimate reproducibility and found mtDNA-CN estimated from cell lysate is significantly less variable than traditional phenol-chloroform-isoamyl alcohol (p = 5.44x10 -4 ) and silica-based column selection (p = 2.82x10 -7 ). In conclusion, we recommend the field moves towards more accurate methods for mtDNA-CN, as well as re-analyze trait associations as more WGS data becomes available from larger initiatives such as TOPMed.Δ Ct values >3 standard deviations from the mean of the plate. Outlier replicates were identified and excluded for samples with a Δ Ct standard deviation >0.5. The sample was excluded if the Δ Ct standard deviation remained >0.5 after replicate removal. We

Blood-derived mitochondrial DNA copy number is associated with gene expression across multiple tissues and is predictive for incident neurodegenerative disease

“…19) and sex (p=0.17) were also in the expected direction. Effect sizes estimates for age and neutrophils were also consistent with prior literature [22] (Supp. Table 1).…”

“…mtDNA-CN was estimated as the number of mitochondrial reads divided by the difference between the number of total reads and the number of unaligned reads to obtain a ratio of mtDNA to nuclear DNA. Whole genome is a highly accurate method for estimation of mtDNA-CN [22,56].…”

Mitochondrial DNA copy number and incident heart failure: The Atherosclerosis Risk in Communities (ARIC) study

“…In addition, we used state-of-the art tools to measure mtDNA-CN. 23 The figure includes hazard ratios for comparing the 90 th to the 10 th percentile (reference) of mtDNA copy number. Pre-specified subgroups were age (<60 or ≥60 years), sex, race (White or Black), smoking status (never, former, or current), alcohol intake (never, former, current), BMI (underweight/normal, overweight, or obese), and prevalent CHD.…”

“…The methods for measuring mtDNA-CN have been described previously. 17,23 Briefly, DNA was extracted using the Gentra Puregene Blood Kit (Qiagen N.V., Venlo, The Netherlands) from buffy coat of whole blood samples collected in visits 1-4. mtDNA-CN was calculated from probe intensities of mitochondrial single nucleotide polymorphisms (SNP) on the Affymetrix Genome-Wide Human SNP Array 6.0 using the Genvisis software package (www.genvisis.org), which uses the median mitochondrial probe intensity of 25 high-quality mitochondrial probes as initial raw measure of mtDNA-CN. Batch effects, DNA quality, and starting DNA quantity were corrected for by using surrogate variable analysis applied to probe intensities of 43,316 autosomal SNPs.…”

SMART: Statistical Mitogenome Assembly with Repeats

“…The variability is not entirely a species effect -indeed, differences of more than an order of magnitude can be observed in the percentage of mtDNA reads in the human datasets. Most likely, other important contributing factors to this variability include DNA extraction and library preparation protocols [35], the tissue of origin [1], and the developmental stage of the sample [36]. Figure 3 compares the accuracy of the read filter employed by Norgal with that of the coverage-and alignment-based filters of SMART on the human datasets described in Table 1.…”