Functional Integrative Bayesian Analysis of High-dimensional Multiplatform Clinicogenomic Data

Area of Science:

• Genomics
• Proteomics
• Computational Biology

Background:

• Multi-platform molecular and genomics data offer opportunities for disease understanding and treatment.
• Current multi-omic integration methods need enhanced approaches for detailed cellular function evaluation.
• Identifying precise cellular functions is crucial for understanding complex disease mechanisms.

Purpose of the Study:

• To propose a novel framework, fiBAG, for the simultaneous identification of upstream functional evidence of proteogenomic biomarkers.
• To incorporate functional knowledge into Bayesian variable selection models for improved signal detection.
• To enhance the understanding of cellular functions governing complex disease mechanisms.

Main Methods:

• Functional Integrative Bayesian Analysis of High-dimensional Multiplatform Genomic Data (fiBAG) framework.
• Utilizing Gaussian process models to quantify functional evidence via Bayes factors.
• Mapping Bayes factors to a calibrated spike-and-slab prior for guided variable selection.

Main Results:

• Simulations show integrative methods with functional calibration possess higher power for detecting disease-related markers.
• fiBAG demonstrates profitability in a pan-cancer analysis across 14 cancer types.
• Identified and assessed cellular mechanisms of proteogenomic markers associated with cancer stemness and patient survival.

Conclusions:

• fiBAG provides a data-driven approach to evaluate cellular functions and improve biomarker discovery.
• Integrating functional evidence enhances the detection of biologically relevant markers for patient outcomes.
• The framework offers a powerful tool for proteogenomic biomarker assessment in cancer research.