A two-platform design for next generation genome-wide association studies.

Area of Science:

• Genetics
• Genomics
• Statistical Genetics

Background:

• Genome-wide association studies (GWAS) have identified common genetic variants linked to complex traits and diseases.
• Advancements in array technology and reference sets enable the next generation of GWAS to focus on uncommon single nucleotide polymorphisms (SNPs) (1% ≤ Minor Allele Frequency [MAF] ≤ 10%).
• Genotyping all participants for uncommon SNPs is often prohibitively expensive for large GWAS.

Purpose of the Study:

• To evaluate the efficiency of a two-platform genotyping and imputation strategy for identifying uncommon SNPs in GWAS.
• To assess the trade-offs between cost, power, and data density in next-generation GWAS designs.

Main Methods:

• A two-platform approach was proposed: standard genotyping array for all participants and a denser array for a subset.
• The subset genotyped on the denser array serves as part of the imputation reference set.
• The strategy was evaluated using a dataset of 756 individuals genotyped on Illumina Human OmniExpress and Omni2.5 Quad arrays.

Main Results:

• Genotyping only 100 individuals on the denser array, combined with imputation, resulted in only a modest loss of statistical power for association detection.
• The efficiency of the two-platform method was demonstrated on a real-world dataset.
• Imputation can lead to more substantial power loss if rare variants have significantly larger relative risks than common variants.

Conclusions:

• A two-platform genotyping and imputation approach offers a cost-effective strategy for next-generation GWAS focusing on uncommon SNPs.
• This method provides a practical alternative to genotyping all participants on dense arrays, balancing cost and power.
• Careful consideration of potential imputation power loss is necessary, particularly when rare variants exhibit large effect sizes.