Imputation aware tag SNP selection to improve power for multi-ethnic association studies

Abstract

1The emergence of very large cohorts in genomic research has facilitated a focus on 2 genotype-imputation strategies to power rare variant association. Consequently, a new generation 3 of genotyping arrays are being developed designed with tag single nucleotide polymorphisms 4 (SNPs) to improve rare variant imputation. Selection of these tag SNPs poses several challenges 5 as rare variants tend to be continentally-or even population-specific and reflect fine-scale linkage 6 disequilibrium (LD) structure impacted by recent demographic events. To explore the landscape of 7 tag-able variation and guide design considerations for large-cohort and biobank arrays, we 8 developed a novel pipeline to select tag SNPs using the 26 population reference panel from Phase 9 3 of the 1000 Genomes Project. We evaluate our approach using leave-one-out internal validation 10 via standard imputation methods that allows the direct comparison of tag SNP performance by 11 estimating the correlation of the imputed and real genotypes for each iteration of potential array 12 sites. We show how this approach allows for an assessment of array design and performance that 13 can take advantage of the development of deeper and more diverse sequenced reference panels. 14 We quantify the impact of demography on tag SNP performance across populations and provide 15 population-specific guidelines for tag SNP selection. We also examine array design strategies that 16 target single populations versus multi-ethnic cohorts, and demonstrate a boost in performance for 17 the latter can be obtained by prioritizing tag SNPs that contribute information across multiple 18 populations simultaneously. Finally, we demonstrate the utility of improved array design to provide 19 meaningful improvements in power, particularly in trans-ethnic studies. The unified framework 20 presented will enable investigators to make informed decisions for the design of new arrays, and 21 help empower the next phase of rare variant association for global health. 22

Genotype imputation performance of three reference panels using African ancestry individuals

“…Imputation of rare SNVs is more challenging since rare alleles are often ethnicity or population-specific and reflect fine-scale linkage disequilibrium (LD) structure impacted by recent demographic events (Wojcik et al 2017). Options for imputing low-frequency and rare variants more accurately in any specific population include increasing the size of the imputation reference panel to capture more reference haplotypes, or increasing the sequencing depth in the reference samples to minimize error rates inherent in low-coverage sequencing (Browning and Browning 2009).…”

The critical needs and challenges for genetic architecture studies in Africa

“…Genotyping arrays have traditionally been biased towards alleles most frequent and imputable in European populations [9,10], compounding biases in which GWAS identify variant associations most common in the study population [11,12]. In contrast, array backbones prioritizing SNPs that maximally tag variants across all populations improve imputation performance, providing more even genomic coverage [13]. Perhaps more importantly, imputation panels are vastly Eurocentric, shortchanging representation of the greater haplotypic diversity present in Africans from deeper recombination history [12,14,15].…”

Genotype Imputation Performance of Three Reference Panels Using African Ancestry Individuals

“…Imputation of rare SNVs is more challenging since rare alleles are often ethnicity or population-specific and reflect fine-scale linkage disequilibrium (LD) structure impacted by recent demographic events [7]. Options for imputing low-frequency and rare variants more accurately in any specific population include increasing the size of the imputation reference panel to capture more reference haplotypes, or increasing the sequencing depth in the reference samples to minimize error rates inherent in low-coverage sequencing [8].…”