Computational approaches for detecting protein complexes from protein interaction networks: a survey

Area of Science:

• Proteomics
• Bioinformatics
• Systems Biology

Background:

• Proteins function in macromolecular complexes, but experimental determination of these complexes, especially large ones, is limited.
• High-throughput techniques like yeast-two-hybrid generate extensive pairwise protein-protein interaction data.
• Computational approaches are crucial for inferring protein complexes from interaction data.

Purpose of the Study:

• To review and evaluate state-of-the-art computational techniques for protein complex detection.
• To discuss future research directions in the field of computational protein complex identification.

Main Methods:

• Construction of protein interaction networks using proteins as nodes and interactions as links.
• Application of graph clustering techniques to detect protein complexes from binary interaction graphs.

Main Results:

• Computational methods accurately identify protein complexes, validated by yeast protein interaction data.
• Performance of computational approaches for protein complex detection has improved over time.

Conclusions:

• Computational methods are valuable complements to experimental techniques for mapping protein interactions and understanding cellular functions.
• Future advancements depend on improving protein interaction data quality and integrating diverse biological evidence.