Genome-wide discovery for biomarkers using quantile regression at biobank scale

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) traditionally use linear regression for quantitative traits.
• Linear regression models only the conditional mean of a phenotype, potentially missing complex genotype-phenotype relationships.

Purpose of the Study:

• To introduce and evaluate quantile regression as an alternative to linear regression for GWAS.
• To demonstrate the advantages of quantile regression in identifying variants with heterogeneous effects across phenotype distributions.

Main Methods:

• Application of quantile regression to analyze the entire conditional distribution of phenotypes.
• Simulations to assess the power of quantile regression under various genetic models.
• Analysis of 39 quantitative traits from the UK Biobank using quantile regression.

Main Results:

• Quantile regression effectively identifies genetic variants with heterogeneous effects across phenotype quantiles.
• This method accommodates diverse phenotype distributions and is invariant to trait transformations.
• UK Biobank data revealed variants with substantial effects in high-risk subgroups, missed by conventional GWAS.

Conclusions:

• Quantile regression is a valuable complement to linear regression in GWAS.
• It provides a more comprehensive understanding of genotype-phenotype associations, especially for complex traits.
• This approach can uncover genetic insights relevant to specific, high-risk subpopulations.