A framework for considering prior information in network-based approaches to omics data analysis.

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• Molecular biology research has generated extensive pathway information, leading to databases and computable network formats.
• Modern omics technologies enable systematic profiling of cellular processes, driving the development of algorithms using prior knowledge networks (PKNs).
• The choice of PKNs significantly influences the outcomes of omics data analysis.

Purpose of the Study:

• To introduce a five-level framework for categorizing biological network models.
• To assess network models based on scope, detail, and causal prediction capabilities.
• To contextualize the framework by reviewing omics analysis methods at each level.

Main Methods:

• Development of a five-level classification framework for network models.
• Review of omics analysis methods utilizing prior knowledge networks.
• Categorization of methods according to the proposed framework and their computational tasks.

Main Results:

• The proposed framework provides a structured way to understand and select biological network models.
• Different levels of the framework correspond to varying degrees of network detail and causal inference potential.
• The review highlights the importance of matching PKN characteristics to omics data types and analysis goals.

Conclusions:

• A standardized framework is essential for effective use of prior knowledge networks in omics research.
• The framework aids researchers in choosing appropriate network models for specific computational tasks.
• This work facilitates more robust and reproducible omics data analysis through informed network selection.