Simultaneous test and estimation of total genetic effect in eQTL integrative analysis through mixed models

Area of Science:

• Genetics and Genomics
• Statistical Genetics
• Bioinformatics

Background:

• Integrating expression quantitative trait loci (eQTL) into genome-wide association studies (GWAS) is crucial for identifying functional single-nucleotide polymorphisms (SNPs) in complex traits.
• Efficiently incorporating eQTL mapping into GWAS for causal gene prioritization remains a significant challenge in the post-GWAS era.

Purpose of the Study:

• To propose a novel statistical method, Mixed transcriptome-wide association studies and mediated Variance estimation (MTV), for integrating eQTL and GWAS data.
• To enhance the identification and prioritization of causal genes underlying complex phenotypes by modeling SNP effects as a function of eQTL.

Main Methods:

• Developed MTV, a unified framework using mixed models to integrate TWAS and eQTL information, encompassing prior methods as special cases.
• Utilized mediation analysis and two-stage Mendelian randomization to statistically justify the MTV framework.
• Implemented a computationally efficient parameter-expansion expectation-maximization algorithm for MTV.

Main Results:

• MTV demonstrated superior performance over existing methods, accurately controlling Type I error and exhibiting increased power in discovering true genetic associations.
• The method effectively processes direct SNP effects and assesses joint effects of SNPs and genetically regulated gene expression (GReX) using a likelihood ratio test.
• Application of MTV to 41 complex traits identified novel associated genes missed by previous approaches and revealed GReX mediates a substantial fraction of phenotypic variation.

Conclusions:

• MTV provides a robust and realistic modeling foundation for integrative omics analysis, significantly improving the biological interpretation of GWAS results.
• The method offers enhanced power and accuracy in identifying causal genes for complex traits by leveraging eQTL information within a unified statistical framework.