A Bayesian approach to differential edges with probabilistic interactions: applications in association and...

Area of Science:

• Systems biology
• Bioinformatics
• Computational biology

Background:

• Differential network (D-Net) analysis identifies genetic variations under different conditions.
• Existing methods offer deterministic inference, overlooking probabilistic insights into differential edges.

Purpose of the Study:

• To propose a Bayesian approach for inferring differential networks with uncertainty.
• To develop a method for probabilistic evaluation and prioritization of differential edges.

Main Methods:

• Developed PRobabilistic Interaction for Differential Edges (PRIDE), a Bayesian regression model for D-Net analysis.
• Translated probabilistic estimation to inferential D-Net analysis for genetic association and classification.
• Evaluated PRIDE's performance against state-of-the-art methods using simulations and real-world cancer data.

Main Results:

• PRIDE performs comparably to or better than existing tools on deterministic criteria.
• PRIDE provides probabilistic evaluation and prioritization of differential edges.
• PRIDE highlights hub node importance and identifies potential sub-networks within D-Nets.

Conclusions:

• PRIDE offers a novel probabilistic framework for differential network analysis.
• The method enhances the understanding of genetic variations by quantifying edge uncertainty.
• PRIDE provides valuable insights for genetic association and classification studies.