A Genomics England haplotype reference panel and imputation of UK Biobank

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• High-density genotype imputation is crucial for genome-wide association studies (GWAS).
• Existing reference panels may lack sufficient diversity or density for accurate imputation of rare variants.
• The Genomics England (GEL) dataset offers a valuable resource for building improved reference panels.

Purpose of the Study:

• To construct a large-scale autosomal variant reference panel using the GEL dataset.
• To evaluate the phasing accuracy and imputation quality of the developed reference panel.
• To assess the utility of the GEL-imputed data in large-scale genome-wide association analyses.

Main Methods:

• Generation of a reference panel comprising 342 million autosomal variants from 78,195 individuals in the GEL dataset.
• Phasing accuracy assessed by switch error rate (0.18% for European samples).
• Imputation quality evaluated using r² (0.75 for variants with minor allele frequencies as low as 2 × 10⁻⁴ in white British samples).

Main Results:

• Achieved a low phasing switch error rate of 0.18% for European samples.
• Demonstrated high imputation quality (r² = 0.75) for rare variants (MAF ≥ 2 × 10⁻⁴) in white British samples.
• GEL-imputed UK Biobank GWAS identified 70% of associations found by direct exome sequencing, extending rare variant testing across the genome.

Conclusions:

• The GEL reference panel significantly enhances imputation accuracy for rare variants.
• Imputed genotype data from this panel effectively supports large-scale genome-wide association studies.
• Rare coding variants play a dominant role in genome-wide association results, suggesting limited impact from rare non-coding variants.