Improving power of association tests using multiple sets of imputed genotypes from distributed reference panels.

Area of Science:

• Genomics
• Population Genetics
• Statistical Genetics

Background:

• Genotype imputation accuracy relies on reference panel size and genetic similarity.
• Combining results from multiple, non-consented reference panels for imputation is challenging.
• Optimizing imputation strategies is crucial for increasing the power of genetic association studies.

Purpose of the Study:

• To compare the accuracy of genotype imputation using different reference panels.
• To evaluate strategies for combining imputation results to enhance genetic association study power.

Main Methods:

• Compared imputation accuracy of 9,265 Norwegian genomes using three reference panels: 1000 Genomes phase 3 (1000G), Haplotype Reference Consortium (HRC), and a Norwegian-specific panel (HUNT).
• Assessed imputation accuracy for variants with varying minor allele frequencies (MAF).
• Evaluated two strategies for utilizing multiple imputed genotype sets in association testing.

Main Results:

• A population-matched reference panel (HUNT) enabled imputation of more population-specific, low-frequency variants (0.05%–0.5% MAF).
• The HUNT panel achieved substantially higher imputation accuracy than 1000G and was comparable to the much larger HRC panel.
• Testing all imputed variants from any panel increased association power, particularly for low-frequency variants (MAF < 1%), compared to selecting single best-quality variants.

Conclusions:

• Population-specific reference panels are valuable for accurate genotype imputation.
• Combining imputation results from multiple panels by testing all variants boosts association study power, especially for rare variants, even with multiple testing adjustments.