METRO

METRO integrates multi-ancestry gene expression and genome-wide association study (GWAS) data to increase statistical power and calibrate inference in transcriptome-wide association studies (TWAS).

Key Features:

• Ancestry-Inclusive Expression Models: Leverages gene expression prediction models constructed from diverse genetic ancestries to accommodate population-specific variation in gene expression.
• Likelihood-Based Inference Framework: Implements a likelihood-based inference framework that calibrates p-values.
• GWAS–Expression Integration: Integrates summary-level GWAS findings with multi-ancestry expression models for joint analysis.
• Enhanced Statistical Power: Produces improved power to detect gene-trait associations relative to traditional TWAS approaches.

Scientific Applications:

• Simulation and Real-Data Evaluation: Validated using both simulated scenarios and real-world analyses of complex traits and diseases.
• Multi-Ancestry GWAS Analysis: Applied to seven traits and diseases using four GWAS datasets (European ancestry n=188,577 and n=339,226; African ancestry n=42,752 and n=23,827).
• Small-Sample Ancestry Discovery: Demonstrated effectiveness in smaller-sample contexts, notably African ancestry cohorts.
• Gene-Trait Discovery: Enabled identification of associations including PLTP and PPARG (lipid metabolism) and MAPT (type II diabetes, implicated in impaired insulin secretion).

Methodology:

Utilizes gene expression data from the Genetic Epidemiology Network of Arteriopathy (GENOA) study (1,032 African Americans and 801 European Americans), integrates these expression models with GWAS findings, and applies a likelihood-based inference framework to calibrate p-values.