Statistical Methods for Microbiome Compositional Data Network Inference: A Survey

Area of Science:

• Microbial ecology
• Bioinformatics
• Network science

Background:

• Microbial communities are ubiquitous and interact with their environments.
• Studying these interactions is crucial for understanding ecosystem dynamics.
• Microbiome interaction network inference is a key approach, but faces data challenges.

Purpose of the Study:

• To provide a comprehensive review of emerging microbiome interaction network inference methods.
• To categorize and discuss different network types, their assumptions, and limitations.
• To identify current challenges and future prospects in microbial network analysis.

Main Methods:

• Review of existing literature on microbiome network inference.
• Categorization of networks into correlation, conditional correlation, mixture, and differential networks.
• Analysis of assumptions, advantages, and limitations of each network type.

Main Results:

• Microbiome network inference is challenged by data properties (compositional, high-dimensional, sparse, heterogeneous) and external factors.
• Four main categories of networks (correlation, conditional correlation, mixture, differential) are presented with their characteristics.
• No single method currently captures all aspects of complex and dynamic microbial interactions.

Conclusions:

• Current methods have limitations in fully representing microbial interactions.
• Future research requires integrated statistical and experimental approaches.
• Advancements in microbial network inference are essential for ecological and biological understanding.