A gene-specific variance-control approach corrects polygenicity-driven inflation observed in transcriptome-wide...

Area of Science:

• Genetics
• Statistical Genomics
• Bioinformatics

Background:

• Transcriptome-wide association studies (TWASs) and related methods (xWASs) are crucial for linking genetic variation to disease via molecular traits.
• The impact of polygenicity on the accuracy of these mediator-trait association tests has been largely unaddressed.
• Complex traits are often highly polygenic, necessitating an evaluation of current association test validity.

Purpose of the Study:

• To assess the accuracy of mediator-trait association tests in the presence of polygenicity.
• To develop and validate a method for controlling false-positive rates in genetic association studies.
• To address inflation issues in methods analogous to TWASs that use genetic predictors.

Main Methods:

• Investigated the effect of polygenicity on linear regression-based association tests using simulated and real data.
• Developed a novel variance-control method, analogous to genomic control but gene-specific.
• Evaluated the performance of the proposed method against existing approaches and theoretical derivations.

Main Results:

• Standard linear regression tests show inflated false-positive rates for highly polygenic traits, increasing with sample size and heritability.
• The proposed variance-control method effectively calibrates false-positive rates, outperforming current methods.
• TWAS-analogous methods also exhibit inflation, highlighting the need for correction in genetic predictor development.

Conclusions:

• Polygenicity poses a significant challenge to the validity of TWASs and related methods.
• The developed variance-control method provides a robust solution for accurate genetic association testing.
• Developers of genetic predictors, including polygenic risk scores (PRSs), should implement inflation parameter correction for reliable results.