Missing mitochondrial diversity hinders TB genetics in Southern Africa

Southern African populations carry extraordinary mitochondrial diversity and include some of the earliest diverging mitochondrial haplogroups known to science. Despite that richness, these populations are underrepresented in genetic studies and in public reference databases. This gap matters because many popular analytic approaches rely on reference data to fill in missing genetic information, and those tools may not perform well when a population’s history and variation are not represented. To test this directly, Desiree C. Petersen and colleagues carried out a focused study of a publicly available resource called the MitoImpute reference panel. They set out to see how well MitoImpute could predict mitochondrial DNA (mtDNA) variation in southern African groups, whether imputation would affect assignment to mitochondrial haplogroups, and whether population-specific mtDNA variants were recovered. As a practical demonstration, the team also applied the same imputation workflow to merge five southern African tuberculosis (TB) case–control mtDNA datasets and then ran a mitochondrial genome-wide association study (miWAS) to look for links between mtDNA and TB susceptibility.

The researchers performed a comprehensive evaluation of the MitoImpute reference panel and imputation workflow, focusing on three outcomes: overall imputation performance for mitochondrial DNA, impact on mitochondrial haplogroup assignment, and recovery of population-specific variants. For their real-data test, they used the MitoImpute reference panel to harmonize and impute data from five southern African TB case–control mtDNA datasets and then conducted a miWAS of TB susceptibility. Results showed a mixed picture. MitoImpute performed well at predicting mitochondrial haplogroups and accurately imputed a subset of mtDNA variants, meaning it can be useful for some analytic goals. At the same time, substantial population-specific variation present in southern African groups was not captured by the panel. In the combined TB miWAS the team did not identify any significant associations with TB susceptibility, a finding the authors link to the limitations of array-based mtDNA genotyping and imputation when applied to highly diverse populations.

These findings carry practical and scientific implications. First, they underscore the need for improved reference panels that are enriched for African mitochondrial diversity so that tools like MitoImpute can recover the full range of local variation. Second, the work highlights that array-based mtDNA genotyping followed by imputation may miss variants that are important in populations with complex demographic histories, reducing the chance of finding disease associations in studies like the TB miWAS described here. The study also provides useful guidance for researchers who must harmonize mitochondrial data across multiple cohorts: imputation can help with haplogroup assignment and some variants, but researchers should be cautious about relying on it as a substitute for direct sequencing in underrepresented populations. Overall, the work by Desiree C. Petersen and colleagues argues for targeted investments in reference data and careful interpretation of imputed mitochondrial results in multi-ancestry research.