To Embed or Not: Network Embedding as a Paradigm in Computational Biology

Area of Science:

• Bioinformatics
• Computational Biology
• Network Science

Background:

• High-throughput biomedical data generation is accelerating.
• Network-based analyses are crucial for interpreting complex biological data.
• Graph embedding techniques simplify and visualize biological networks.

Purpose of the Study:

• To review and compare traditional and novel graph embedding methods.
• To evaluate their application in network biology problems.
• To contrast embedding methods with direct network analysis.

Main Methods:

• Survey of graph embedding approaches.
• Comparative analysis of embedding versus direct network analysis.
• Application in protein network alignment, community detection, and protein function prediction.

Main Results:

• Both graph embedding and direct network analysis have value in network biology.
• Performance depends on specific evaluation measures and project goals.
• Network embedding methods show superior performance in certain benchmarks.

Conclusions:

• Graph embedding is a valuable and essential tool in bioinformatics research.
• These techniques aid in deciphering complex biological networks.
• The choice of method depends on the specific biological question and metrics.