A Mixed-Effects Model for Powerful Association Tests in Integrative Functional Genomics

Area of Science:

• Genetics
• Statistical genetics
• Computational biology

Background:

• Genome-wide association studies (GWASs) identify genetic variants for complex diseases but explain limited heritability.
• External transcriptome data offers promise for novel variant discovery, with methods like PrediXcan using predicted gene expression.
• Existing methods like PrediXcan have limitations, including potential bias in predicted expression and overlooking variants with non-expression-related mechanisms.

Purpose of the Study:

• To develop a unified statistical framework that overcomes limitations of current gene-based association methods.
• To enhance the power and interpretability of genetic association studies by incorporating both predicted gene expression and individual variant effects.
• To identify novel disease-associated genes missed by existing approaches.

Main Methods:

• Developed a unified mixed effects model incorporating fixed effects for imputed gene expression and random effects for individual variant residual effects.
• Proposed MiST (mixed effects score test), a set-based score testing framework with two data-driven combination approaches.
• Established asymptotic distributions for rapid p-value calculation in genome-wide analyses and provided separate p-values for fixed and random effects.

Main Results:

• Extensive simulations demonstrated that the proposed MiST approaches are significantly more powerful than existing methods.
• Application to a large-scale GWAS of colorectal cancer identified two novel associated genes, POU5F1B and ATF1.
• These identified genes were missed by the PrediXcan approach, highlighting the advantage of the unified model.

Conclusions:

• The MiST framework provides a powerful and interpretable approach for genetic association studies, enhancing the discovery of disease-related genes.
• MiST effectively integrates gene expression and individual variant effects, addressing limitations of previous methods.
• This method has significant implications for understanding the genetic architecture of complex diseases and identifying potential therapeutic targets.