Genotype Imputation Performance of Three Reference Panels Using African Ancestry Individuals

Genotype imputation is used to estimate unobserved genotypes from genome-wide maker data, to increase genome coverage and power for genome-wide association studies. Imputation has been most successful for European ancestry populations in which very large reference panels are available. Smaller subsets of African descent populations are available in 1000 Genomes (1000G), the Consortium on Asthma among African-Ancestry Populations in the Americas (CAAPA) and the Haplotype Reference Consortium (HRC). We aimed to compare the performance of these reference panels when imputing variation in 3,747 African Americans (AA) from 2 cohorts (HCV and COPDGene) genotyped using the Illumina Omni family of microarrays. The haplotypes of 2,504 individuals (from 1000G), 883 (from CAAPA) and 32,611 (from HRC) were used as reference. We compared the performance of these panels based on number of variants, imputation quality, imputation accuracy and coverage. In both cohorts, 1000G imputed 1.5-1.6x more variants compared to CAAPA and 1.2x more variants than HRC. Similar findings were observed for variants with higher imputation quality (R2>0.5) and for rare, low frequency, and common variants. When merging the results of the three panels the total number of imputed variants was 62M-63M with 20M overlapping variants imputed by all three panels, and a range of 5 to 15M unique variants imputed exclusively with one of the three panels. For overlapping variants, imputation quality was highest for HRC, followed by 1000G, then CAAPA, and improved as the minor allele frequency increased. The 1000G, HRC and CAAPA participants of African ancestry provided high performance and accuracy for imputation of African American admixed individuals, increasing the total number of variants with high quality available for subsequent analyses. These three panels are complementary and would benefit from the development of an integrated African reference panel, including data from multiple sources and populations.