Imputation Accuracy Across Global Human Populations.

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• Genotype imputation is crucial for genome-wide association studies (GWAS).
• Existing imputation reference panels, like TOPMed, underrepresent non-European ancestries, leading to fairness issues.
• Imputation accuracy for populations outside North America remains a concern despite advances.

Approach:

• Curated genome-wide array data from 23 publications (2008-2021) for over 43,000 individuals across 123 global populations.
• Assessed imputation accuracy (r-squared) comparing non-European populations to European-ancestry cohorts.
• Evaluated meta-imputation using Taiwan Biobank whole-genome sequencing data to improve imputation for underrepresented groups.

Key Points:

• Significant disparities in imputation accuracy were observed, with lower r-squared values in populations like Saudi Arabians, Vietnamese, Thai, and Papua New Guineans compared to Europeans.
• Imputation accuracy generally decreased with increased genetic distance from European references outside of Africa and Latin America.
• Imputation software may overestimate accuracy for non-European populations, exacerbating disparities.

Conclusions:

• Current reference panel diversity and size are insufficient for equitable genotype imputation.
• Meta-imputation, in the tested design, did not enhance imputation accuracy for underrepresented non-European populations.
• Increasing diversity and sample size in reference panels is essential to promote equity in genetic research.