METAINTER

METAINTER performs meta-analysis of summary statistics from multiple studies to integrate and synthesize complex regression models, including linear and logistic regressions, in genome-wide association studies (GWAS).

Key Features:

• Versatility in Model Handling: Manages models with an arbitrary number of parameters to support single-SNP tests, global haplotype tests, and gene-gene or gene-environment interaction analyses.
• Incorporation of Study-Specific Information: Integrates study-specific parameter estimates, effect directions, standard errors, and covariance structures for robust analysis of high-dimensional models.
• Adaptation from Educational Sciences: Adapts a slope-synthesis method originally proposed for linear regressions to accommodate multiple logistic regressions.
• Simulation Validation: Demonstrated appropriate type I error rates and power in simulations of two-SNP models, approximating joint analysis across pooled samples.
• Real-World Application: Applied to six GWAS datasets from dbGaP on type 2 diabetes, combining genome-wide pairwise SNP interaction tests performed with logistic regression via meta-analysis.

Scientific Applications:

• GWAS meta-analysis: Synthesis of summary statistics across studies to detect genetic associations in genome-wide scans.
• Complex regression synthesis: Integration of results from linear and logistic regression models including single-SNP, haplotype, gene-gene, and gene-environment tests.
• Genome-wide interaction screening: Meta-analysis of pairwise SNP interaction tests to investigate genetic contributions to diseases such as type 2 diabetes.

Methodology:

Modifies a slope-synthesis meta-analysis method to extend from linear to multiple logistic regressions, integrates study-specific parameter estimates (effect directions, standard errors, covariance structures), validates performance via two-SNP simulations, and combines results from genome-wide pairwise SNP logistic regression tests by meta-analysis.