Adaptively Integrative Association between Multivariate Phenotypes and Transcriptomic Data for Complex Diseases.

Area of Science:

• Genomics
• Systems Biology
• Translational Research

Background:

• Phenotype-gene association studies are crucial for understanding complex diseases.
• Current methods often focus on single nucleotide polymorphism (SNP)-based associations, limiting holistic views.
• Integrating multiple phenotypes enhances statistical power and provides comprehensive insights.

Purpose of the Study:

• To extend and evaluate adaptive Fisher's methods (AFp and AFz) for phenotype-mRNA association analysis.
• To aggregate heterogeneous phenotype-gene effects and accommodate diverse phenotype data types.
• To identify gene modules associated with specific phenotypes through effect selection and clustering.

Main Methods:

• Utilized adaptive Fisher's methods (AFp and AFz) for p-value combination in phenotype-mRNA association.
• Employed bootstrap analysis to calculate variability indices for phenotype-gene effect selection.
• Generated a co-membership matrix to cluster gene modules based on phenotype-gene effects.

Main Results:

• Extensive simulations demonstrated AFp's superior performance over existing methods in type I error control and statistical power.
• The proposed methods effectively aggregated heterogeneous phenotype-gene effects and selected associated phenotypes.
• Applied to lung disease, breast cancer, and brain aging datasets, yielding significant biological findings.

Conclusions:

• The developed adaptive Fisher's methods offer a powerful approach for phenotype-mRNA association analysis in complex diseases.
• These methods enhance statistical power, enable robust phenotype-gene effect aggregation, and facilitate biological interpretation.
• The findings provide a foundation for uncovering novel disease mechanisms and identifying therapeutic targets.