Executable pathway analysis using ensemble discrete-state modeling for large-scale data.

Area of Science:

• Computational biology
• Systems biology
• Bioinformatics

Background:

• Pathway analysis is crucial for interpreting high-throughput omics data.
• Existing methods often neglect pathway topology and signal integration, leading to biased enrichment results.
• Incorporating signal flow and topology can better rank pathways based on key regulatory gene modulation.

Purpose of the Study:

• To introduce a novel algorithm, Boolean Omics Network Invariant-Time Analysis (BONITA), for pathway analysis.
• To model cellular heterogeneity and measure pathway activities using discrete state dynamics.
• To enhance the sensitivity and accuracy of pathway and gene prioritization from omics data.

Main Methods:

• Developed BONITA, a discrete state network modeling approach for signal propagation and integration.
• Modeled intercellular heterogeneity in transcriptomic data and propagated binary signals across networks.
• Inferred logic rules for signal integration using a genetic algorithm and local search for pathway scoring.

Main Results:

• BONITA demonstrated higher sensitivity than state-of-the-art methods at lower levels of source node modulation.
• Applied to RNA-sequencing studies, BONITA identified additional relevant pathways in cell line and infant studies.
• BONITA successfully detected disease-specific pathway modulation and identified known drug targets among high-impact genes.

Conclusions:

• BONITA offers a powerful approach for pathway analysis, improving mechanistic insights from omics data.
• The algorithm effectively prioritizes pathways and identifies key genes with mechanistic roles.
• BONITA enhances the discovery of relevant biological pathways and potential therapeutic targets.