Deep Pathway Analysis V2.0: A Pathway Analysis Framework Incorporating Multi-Dimensional Omics Data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Pathway analysis is crucial for interpreting complex, high-throughput omics data in cancer research.
• Network topology enhances the power of pathway information for omics data interpretation.

Purpose of the Study:

• To propose DPA V2.0, a topology-based pathway analysis method capable of integrating multiple heterogeneous omics data types.
• To demonstrate the effectiveness of DPA V2.0 in identifying perturbed regulatory signals within cancer pathways.

Main Methods:

• DPA V2.0 encodes each pathway route as a Bayesian network, initialized with conditional probabilities to represent regulatory relationships.
• The method was applied to TCGA breast cancer (BRCA) and ovarian cancer (OV) datasets, integrating mRNA-seq, mutation, copy number variation, and phosphorylation data.
• Survival and patient subtype analyses were performed to evaluate model outcomes.

Main Results:

• DPA V2.0 successfully identified perturbed pathways, with outcomes aligning with biological literature.
• The model demonstrated improved accuracy, identifying 10% more pathways correctly compared to methods using only mRNA-seq and mutation data for breast cancer.
• Integration of diverse omics data significantly enhances the accuracy of perturbed pathway detection.

Conclusions:

• DPA V2.0 offers a powerful approach for in-depth pathway analysis by integrating multiple omics data types.
• The method's effectiveness in identifying key regulatory signals and improving detection accuracy is validated.
• The study encourages the use of multi-omics data in pathway analysis for advancing cancer research.