Accurate multiple network alignment through context-sensitive random walk.

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Comparative network analysis reveals insights into biological network structure and organization.
• Global network alignment identifies similarities and differences across multiple biological networks.
• Existing network alignment methods primarily focus on pairwise comparisons.

Purpose of the Study:

• To develop a novel multiple network alignment algorithm.
• To improve the accuracy of detecting conserved pathways and modules across biological networks.
• To address limitations of pairwise alignment methods.

Main Methods:

• A context-sensitive random walk model is proposed.
• The random walker switches between individual and simultaneous network walks.
• Node correspondence is estimated by integrating node and topological similarity.

Main Results:

• The algorithm predicts maximum expected accuracy (MEA) alignment.
• Quantitative estimation of node correspondence between networks is achieved.
• The method effectively integrates node and topological similarity measures.

Conclusions:

• The proposed algorithm constructs more accurate multiple network alignments.
• Performance is validated on synthetic and real protein-protein interaction networks.
• The novel approach outperforms leading existing methods in accuracy.