A Scalable Permutation Approach Reveals Replication and Preservation Patterns of Network Modules in Large Datasets

Area of Science:

• Computational biology
• Systems biology
• Bioinformatics

Background:

• Network modules, topologically distinct groups of nodes and edges, are vital for understanding biological systems.
• Identifying preserved modules across datasets aids in uncovering conserved biological features.
• Existing statistical methods for assessing module preservation often rely on heuristics and can be systematically biased, leading to skewed p-values.

Purpose of the Study:

• To address the limitations of current methods for assessing network module preservation.
• To introduce a novel, rapid, and computationally efficient method for unbiased module preservation testing.
• To apply the new method to identify conserved gene coexpression modules across multiple murine tissues and bacterial gut microbiota.

Main Methods:

• Development of NetRep, a permutation-based approach for scoring module preservation.
• NetRep does not assume normal distribution of data, providing unbiased p-values.
• Application of NetRep to gene coexpression networks in murine brain, liver, adipose, and muscle tissues.
• Quantification of bacterial network preservation in gut microbiota between sexes.

Main Results:

• Current methods for module preservation testing are systematically biased and produce skewed p-values.
• NetRep provides unbiased p-values and effectively distinguishes true from false positives in multiple hypothesis testing.
• Complex multi-tissue preservation patterns were identified, including a liver-specific housekeeping module associated with body weight in adipose and muscle tissues.
• Preservation of bacterial networks in gut microbiota between men and women was successfully quantified.

Conclusions:

• NetRep is a rapid, computationally efficient, and statistically robust method for assessing network module preservation.
• The method overcomes biases in existing approaches, enabling more reliable biological interpretations.
• NetRep facilitates the discovery of conserved biological features and cross-tissue or cross-population network similarities.
• This tool has broad applicability in systems biology, comparative genomics, and microbiome research.