GLIMPSE2

GLIMPSE2 performs whole-genome imputation and phasing of low-coverage sequencing data using large-scale reference panels to increase genotype accuracy and reduce computational cost.

Key Features:

• Cost Efficiency: Achieves genome-wide imputation for less than US$1 per sample.
• Scalability: Scales sublinearly with respect to number of samples and markers, enabling application to large datasets such as the UK Biobank (150,119 sequences).
• High Accuracy: Maintains high imputation accuracy across the allele frequency spectrum, including rare variants, for very low-coverage sequencing down to 1× coverage.
• Versatility Across Populations: Demonstrates robust performance across different human populations and across varying levels of genomic coverage.

Scientific Applications:

• Disease and Population Genetics: Enables cost-effective genotyping for disease-association and population genetics studies using low-coverage sequencing.
• Gene Expression Association Studies: Provides imputed genotypes that often outperform dense SNP arrays in gene expression association analyses.
• Rare Variant Burden Tests: Facilitates rare variant burden testing by accurately imputing rare alleles from low-coverage data.
• Large-scale and Inclusive Study Design: Reduces per-sample cost and computational requirements to support more extensive and inclusive genetic studies.

Methodology:

Leverages large reference panels to perform phasing and imputation of low-coverage sequencing datasets, scales sublinearly with sample and marker counts, and retains high accuracy for both ancient and modern genomes, demonstrated on datasets such as the UK Biobank.