Bayesian structural equation modeling in multiple omics data with application to circadian genes.

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Cancer Research

Background:

• Integrating multi-omics data reveals novel genomic expression functionalities.
• Multi-platform data analyses offer enhanced statistical power compared to single-source approaches.
• Circadian gene profiles, including copy number and RNA-sequence data, are crucial for understanding cancer survival.

Purpose of the Study:

• To develop a robust statistical model for integrating diverse omics data (copy number and RNA-sequence) for cancer survival prediction.
• To leverage Bayesian structural equation modeling for enhanced multi-platform data integration.
• To predict subject survival by analyzing circadian gene omics profiles.

Main Methods:

• Bayesian structural equation modeling (SEM) coupled with linear and log-normal accelerated failure-time regression.
• Application of conjugate priors on regression parameters.
• Derivation of a Gibbs sampler using conditional distributions for parameter estimation.

Main Results:

• The developed integrative model demonstrated a superior fit to the data compared to competing methods in simulation studies.
• Analyses of glioblastoma and breast cancer datasets from The Cancer Genome Atlas (TCGA) validated the model's effectiveness.
• The model successfully integrated copy number and RNA-sequence data to predict survival outcomes.

Conclusions:

• The proposed Bayesian SEM approach provides a powerful framework for integrating multi-omics data to improve cancer survival prediction.
• The method offers a significant advancement in analyzing complex genomic and transcriptomic datasets.
• The R package 'semmcmc' is available for broader application of this integrative methodology.