Genome-wide mediation analysis: an empirical study to connect phenotype with genotype via intermediate transcriptomic...

Area of Science:

• Genetics and Genomics
• Causal Inference
• Bioinformatics

Background:

• Traditional 2-variable association methods (e.g., GWAS, TWAS) offer limited resolution and causality in mapping genotype to phenotype.
• Genomic variations' molecular mechanisms and candidate genes are challenging to interpret and validate due to low mapping resolution.
• Gene expression is hypothesized as a partial mediator in the causal chain from genotype to phenotype.

Purpose of the Study:

• To test the hypothesis of a causal chain from genotype to phenotype, mediated by gene expression.
• To apply high-dimensional mediation analysis for integrating multi-omics data (genotype, gene expression, phenotype) in maize.
• To identify mediating genes and understand their role in connecting genotype to phenotype.

Main Methods:

• Utilized high-dimensional mediation analysis, a causal inference method, on a maize association panel (N=280 lines).
• Integrated publicly available agronomy traits (40), newly generated metabolite traits (66), and RNA-seq data from 7 tissues.
• Analyzed the causal chain: genotype (exposure) -> gene expression (mediator) -> phenotype (outcome).

Main Results:

• Identified 736 unique mediating genes connecting genotype to various agronomy and metabolite traits.
• Observed pleiotropic effects, with 11% (83/736) of mediating genes influencing multiple traits.
• Validated several identified mediating genes against known functions, supporting the method's efficacy.

Conclusions:

• High-dimensional mediation analysis is a powerful tool for integrating multi-omics data to establish genotype-phenotype relationships.
• The study provides explicit hypotheses for functional validation of identified mediating genes.
• This approach enhances the interpretation of molecular mechanisms underlying genomic variations and their phenotypic consequences.