Choosing appropriate models for protein-protein interaction networks: a comparison study.

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• Protein-protein interaction (PPI) networks are fundamental to understanding cellular mechanisms.
• Numerous models exist for PPI network simulation, but a lack of comparative studies hinders model selection.
• Accurate PPI network models are essential for exploring network evolution and dynamics.

Purpose of the Study:

• To conduct a comprehensive performance comparison of 12 existing PPI network models.
• To provide a guideline for researchers selecting appropriate PPI network models.
• To identify models that best capture the structural and statistical properties of real PPI networks.

Main Methods:

• Evaluated 12 PPI network models using datasets from yeast, mouse, fruit fly, and nematode.
• Compared global and local statistical properties of original PPI networks against model-reproduced networks.
• Utilized the mean reciprocal rank to consolidate model performance across various statistical metrics.

Main Results:

• The PS_Seed, STICKY, and DD_Seed models demonstrated the best fit with empirical PPI datasets.
• Models incorporating node duplication, link dynamics, and degree-weighted behaviors showed superior performance.
• The study identified specific models that accurately reproduce key features of biological PPI networks.

Conclusions:

• The PS_Seed, STICKY, and DD_Seed models are recommended for modeling protein-protein interaction networks.
• Evolutionary mechanisms involving node duplication, link dynamics, and degree-weighting are critical for accurate PPI network representation.
• This comparative analysis offers valuable insights for theoretical and computational biologists in model selection.