Mitochondrial DNA Imputation Accuracy and Its Application in a Southern African Mitochondrial Genome-wide Association Study

Abstract

Southern African populations harbor exceptional mitochondrial genetic diversity and include the earliest diverging mitochondrial haplogroups. Yet, they remain significantly underrepresented in genetic studies and reference databases. This underrepresentation raises concerns about the performance and generalizability of mitochondrial imputation tools in populations with complex demographic histories. In this study, we conducted a comprehensive evaluation of the publicly available MitoImpute reference panel to assess its accuracy and suitability for use in southern African cohorts. We examined imputation performance for mitochondrial DNA (mtDNA), the impact of imputation on haplogroup assignment, and the extent to which population-specific variants were recovered by the panel. As a proof-of-principle application, we used the MitoImpute reference panel and imputation workflow to merge five southern African tuberculosis (TB) case-control mtDNA datasets and perform a mitochondrial genome-wide association study (miWAS) of TB susceptibility. Although MitoImpute performed well in predicting mitochondrial haplogroups and accurately imputing a subset of mtDNA variants, substantial population-specific variation was not captured. No significant associations with TB susceptibility were identified, highlighting key limitations of array-based mtDNA genotyping and imputation in highly diverse populations. Together, these findings underscore the need for improved reference panels enriched for African mitochondrial diversity and provide practical guidelines for mitochondrial data harmonization and analysis in multi-ancestry cohorts.