Pairwise Versus Multiple Global Network Alignment.

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Biological Network Alignment (NA) seeks conserved regions across species' molecular networks.
• Global NA methods include pairwise (PNA) and multiple (MNA), with MNA gaining recent focus for potentially higher quality alignments across more networks.
• Current comparisons are limited to PNA vs. PNA and MNA vs. MNA, hindering evaluation of MNA's claimed superiority and justification for its higher computational cost.

Purpose of the Study:

• To introduce a novel framework enabling direct comparison between PNA and MNA methods.
• To rigorously evaluate the performance of prominent PNA and MNA methods.
• To determine if MNA consistently yields higher-quality alignments than PNA, justifying its computational complexity.

Main Methods:

• Development of a unified framework for comparing PNA and MNA.
• Evaluation of eight leading PNA and MNA algorithms.
• Utilized both synthetic and real-world biological networks.
• Employed topological and functional measures to assess alignment quality.
• Performed comparisons in both pairwise and multiple network settings.

Main Results:

• PNA methods performed better than MNA in the native pairwise evaluation framework, as expected.
• Contrary to expectations, PNA methods frequently outperformed MNA in the multiple network evaluation framework.
• The relative performance of PNA and MNA was dependent on the specific evaluation test used.

Conclusions:

• The study introduces a crucial framework for a standardized comparison of PNA and MNA.
• Findings challenge the assumption that MNA inherently provides superior alignments.
• PNA methods can be more effective than MNA, even in multiple network alignment scenarios, necessitating careful method selection based on evaluation criteria.