Protein complex prediction for large protein protein interaction networks with the Core&Peel method

Area of Science:

• Computational Biology
• Network Science
• Bioinformatics

Background:

• Biological networks are crucial for understanding cellular organization.
• Clustering algorithms are primary tools for analyzing these networks.
• Predicting protein complexes (PC) in large protein-protein interaction networks (PPIN) is a key challenge.

Purpose of the Study:

• To develop and evaluate a novel, efficient algorithm for clustering large sparse networks.
• To specifically address the prediction of protein complexes in large-scale PPINs.
• To assess algorithm performance on real-world biological networks.

Main Methods:

• Introduced the Core&Peel algorithm for clustering large sparse networks.
• Evaluated Core&Peel on multiple large PPI networks from yeast and Homo sapiens.
• Compared Core&Peel against ten state-of-the-art clustering algorithms.

Main Results:

• Core&Peel demonstrated superior performance in identifying known protein complexes.
• The algorithm showed high functional coherence in its predictions.
• Core&Peel is robust against random interactions and offers competitive running times.

Conclusions:

• Core&Peel advances the state-of-the-art in PPIN clustering.
• It provides an efficient algorithmic solution for large, real-world networks.
• The algorithm achieves high output quality and speed, outperforming competitors.