On the current failure-but bright future-of topology-driven biological network alignment

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Protein function is determined by interaction partners, suggesting conserved interaction patterns across species.
• Protein-protein interaction (PPI) network alignment aims to leverage these patterns for functional inference.
• Existing PPI network alignment algorithms struggle to link network topology with functional similarity or recover orthologs.

Purpose of the Study:

• To investigate the limitations of current protein-protein interaction network alignment algorithms.
• To develop and demonstrate a novel network alignment method capable of effective topological alignment.
• To show that topology-based network alignment can reliably identify functionally related proteins and orthologs across species.

Main Methods:

• Analysis of edge densities and topological similarity measures in existing PPI networks.
• Development of the Simulated Annealing Network Aligner (SANA) for optimizing network alignment objective functions.
• Demonstration of SANA's performance on both experimental and synthetic PPI networks.

Main Results:

• Identified low edge density in experimental PPI networks as a primary barrier to successful topological alignment.
• Showcased SANA's superior ability to optimize topological objective functions, achieving near-optimal solutions.
• Achieved statistically significant global network alignments based on topology alone, identifying functionally similar proteins with extremely low p-values.

Conclusions:

• Topological network alignment is a viable strategy for functional prediction, especially with increasing PPI network edge densities.
• SANA significantly advances the field by effectively aligning networks based on topology, enabling robust ortholog recovery.
• This work paves the way for high-throughput functional prediction driven by topology-based network alignment.