Identifying interpretable gene-biomarker associations with functionally informed kernel-based tests in 190,000 exomes

Area of Science:

• Genetics and Genomics
• Biomarker Discovery
• Computational Biology

Background:

• Rare genetic variants significantly influence human traits and disease susceptibility.
• Understanding the genetic architecture of blood biomarkers is crucial for personalized medicine.
• Previous studies often aggregate variants, potentially masking functional effects.

Purpose of the Study:

• To conduct an exome-wide rare genetic variant association study for 30 blood biomarkers.
• To develop and compare novel statistical tests for rare variant association analysis.
• To identify specific genes and functional mechanisms contributing to biomarker variability.

Main Methods:

• Exome-wide association study in 191,971 UK Biobank participants.
• Comparison of gene-based collapsing tests with variant-category-specific kernel-based tests.
• Integration of functional variant effect predictions (missense, splicing, RNA-binding protein interactions).
• Development of a computationally efficient likelihood-ratio and score test combination.
• Introduction of local collapsing by amino acid position for missense variants.

Main Results:

• Identified 193 significant gene-biomarker associations.
• Genes linked to biomarkers were 4.5-fold enriched for Mendelian disorders.
• Kernel-based tests identified 13% more associations than collapsing methods.
• The novel test combination increased associations by 36% compared to score tests alone.
• Identified potential novel gain-of-function variants in PIEZO1.

Conclusions:

• Rare genetic variants substantially contribute to blood biomarker variability.
• Variant-category-specific tests enhance the power and interpretability of association studies.
• Investigating diverse functional mechanisms is essential for comprehensive rare-variant analysis.
• The developed methods offer improved detection of genetic associations for complex traits.