A diverse ancestrally-matched reference panel increases genotype imputation accuracy in a underrepresented population

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• Variant imputation is essential for genome-wide association studies (GWAS) to infer unobserved genotypes using reference panels.
• Existing public reference panels vary in size, sequencing depth, and population representation.
• Performance evaluation of reference panels for underrepresented populations is limited.

Purpose of the Study:

• To compare the imputation performance of major public reference panels (1000 Genomes, HRC, GenomeAsia 100K, TOPMed) for the Thai population.
• To assess genotype yield and imputation accuracy against high-depth whole genome sequencing data.
• To determine the optimal reference panel for imputing genetic data in underrepresented populations.

Main Methods:

• Comparative analysis of four public reference panels: 1000 Genomes Project, Haplotype Reference Consortium (HRC), GenomeAsia 100K, and Trans-Omics for Precision Medicine (TOPMed).
• Imputation of Thai population samples using each reference panel.
• Assessment of genotype concordance rates by comparing imputed genotypes with high-depth whole genome sequencing data.
• Evaluation of imputation accuracy for common and rare variants.

Main Results:

• The Trans-Omics for Precision Medicine (TOPMed) panel provided the largest number of imputed variants (approximately 271 million).
• The GenomeAsia 100K panel achieved the highest imputation accuracy, with a median genotype concordance rate of 0.97.
• GenomeAsia 100K also showed the best accuracy for rare variants, though overall rare variant imputation accuracy was lower than for common variants.
• The superior performance of GenomeAsia 100K is attributed to its diverse population representation, including groups genetically similar to the Thai cohort.

Conclusions:

• Reference panel choice significantly impacts variant imputation accuracy, especially for underrepresented populations.
• The GenomeAsia 100K panel is recommended for accurate imputation in Thai and genetically similar populations.
• Increased sequencing and inclusion of diverse populations in reference panels are vital for advancing genomic research in underrepresented groups.