Fast kernel-based association testing of non-linear genetic effects for biobank-scale data.

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genetics

Background:

• Understanding non-linear genetic effects on complex traits is limited by statistical power.
• Existing kernel-based tests are not scalable to biobank-level datasets (hundreds of thousands of individuals).

Purpose of the Study:

• To introduce FastKAST, a scalable kernel-based method for detecting non-linear genetic effects.
• To enable analysis of non-linear genetic influences on quantitative traits in large-scale biobanks.

Main Methods:

• Developed FastKAST, a computationally efficient kernel-based approach.
• Implemented calibrated hypothesis testing for robust statistical inference.
• Applied the method to UK Biobank data comprising approximately 300,000 individuals.

Main Results:

• FastKAST successfully analyzes biobank-scale datasets with hundreds of thousands of unrelated individuals.
• The method provides calibrated hypothesis tests for non-linear genetic effects.
• Applied to 53 quantitative traits in the UK Biobank, identifying sets of variants with genome-wide significant non-linear effects.

Conclusions:

• FastKAST significantly advances the ability to detect non-linear genetic effects in large populations.
• This method opens new avenues for exploring the genetic architecture of complex traits.
• Enables genome-wide association studies for non-linear genetic influences in biobank-scale cohorts.