Differential Markov random field analysis with an application to detecting differential microbial community networks

Area of Science:

• Microbiology
• Computational Biology
• Bioinformatics

Background:

• Microbial communities form complex ecological networks influenced by diet and environment.
• Differential analysis seeks to understand systematic changes in these networks during adaptation.
• Existing methods may lack flexibility or power for detecting subtle network differences.

Purpose of the Study:

• To propose a flexible Markov random field model for microbial network analysis.
• To introduce a hypothesis testing framework for differential network analysis.
• To identify structural changes in microbial networks and their drivers, such as age.

Main Methods:

• Developed a flexible Markov random field model for microbial network structures.
• Introduced a hypothesis testing framework for detecting differences between networks (differential network analysis).
• Implemented a multiple testing procedure with false discovery rate control for identifying differential network structures.

Main Results:

• The proposed global test for differential networks demonstrates power, especially against sparse alternatives.
• The method successfully identified age-related structural changes in a UK twin gut microbiome study.
• The developed multiple testing procedure effectively controls the false discovery rate.

Conclusions:

• The flexible Markov random field model provides a robust framework for microbial network analysis.
• Differential network analysis is a powerful tool for understanding microbial community dynamics and adaptation.
• Age is a significant factor influencing the structure of the gut microbial community network.