Parallel Exchange of Randomized SubGraphs for Optimization of Network Alignment: PERSONA

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Network alignment in protein-protein interaction networks (PPINs) aims to identify conserved functional regions across different organisms.
• A key challenge is balancing multiple similarity objectives (e.g., topological, functional) within alignment strategies, which can compromise biological relevance.
• Existing methods often face trade-offs, potentially overlooking critical biological insights by prioritizing certain objectives over others.

Purpose of the Study:

• To address the multi-objective optimization challenge in network alignment.
• To propose a novel parallel approach, PERSONA, for enhanced network alignment.
• To improve the biological relevance and balance of network alignment solutions.

Main Methods:

• Developed PERSONA, a parallel network alignment approach utilizing a reactive actor environment.
• Employed a particle swarm optimization strategy where multiple aligners (actors) continuously share and refine partial solutions.
• Incorporated randomized parameter values for individual actor heuristics to explore the search space comprehensively.
• Actors exchange and evaluate subgraphs representing partial solutions, iteratively improving the overall alignment.

Main Results:

• Achieved remarkably optimized and balanced solutions for network alignment problems.
• Demonstrated significant improvements in both topological and node similarity objectives simultaneously.
• PERSONA effectively mitigates the trade-offs inherent in multi-objective network alignment.

Conclusions:

• PERSONA offers a robust framework for multi-objective network alignment by leveraging parallel processing and collaborative solution refinement.
• The approach enhances the discovery of functionally similar regions in protein-protein interaction networks.
• This method provides a more biologically relevant and balanced outcome compared to traditional network alignment strategies.