An integrative association method for omics data based on a modified Fisher's method with application to childhood...

Area of Science:

• Genetics and Bioinformatics
• Computational Biology
• Systems Biology

Background:

• High-throughput biotechnologies generate diverse omics data (genomics, epigenomics, transcriptomics) for complex disease research.
• Individual analysis of omics data types is suboptimal for understanding biological mechanisms.
• Integrating multi-platform omics data is crucial for robust variant identification and biological process articulation.

Purpose of the Study:

• To introduce a novel statistical method, Omnibus-Fisher, for integrating correlated omics data in disease association studies.
• To extend Omnibus-Fisher to an optimal test using perturbations to account for various disease models.
• To evaluate the performance of Omnibus-Fisher compared to standard methods in simulations.

Main Methods:

• Developed Omnibus-Fisher, a modified Fisher's method, to combine p-values from kernel machine regression of genomics, epigenomics, and transcriptomics data.
• Extended Omnibus-Fisher using perturbations to create an optimal test for comprehensive disease model consideration.
• Utilized simulations to assess type I error rates and statistical power.

Main Results:

• Standard Fisher's method showed inflated type I error rates with correlated omics data.
• Omnibus-Fisher preserved expected type I error rates, demonstrating robustness with correlated omics data.
• Omnibus-Fisher showed increased power when all omics data types were involved, while the optimal version was more powerful when only one data type was causal.

Conclusions:

• Omnibus-Fisher provides a robust approach for integrating correlated omics data in genetic association studies.
• The proposed method enhances statistical power and maintains accurate error rates compared to traditional methods.
• Demonstrated the method's utility by analyzing multi-omics data in a childhood asthma study.