Learning functional groups in complex microbiomes.

Area of Science:

• Microbiology
• Computational Biology
• Systems Biology

Background:

• Microbial communities are essential for ecosystem functions like carbon sequestration and immune regulation.
• Understanding the link between microbial community structure and function is crucial but complex due to the vast number of microbes and genes involved.

Purpose of the Study:

• To develop a data-driven approach for identifying key microbial groups or genes responsible for community functions.
• To simplify complex microbiome data and create interpretable structure-function maps.

Main Methods:

• Utilized a neural-network based clustering algorithm for dimensionality reduction.
• Applied the algorithm to gut microbial communities, ocean metagenomes, and soil bacterial species.
• Integrated interpretable machine learning (ML) with strain isolation and sequencing experiments.

Main Results:

• Successfully recovered known functional groups in gut communities.
• Distilled ~500 gene modules in ocean metagenomes into three sparse groups representing survival strategies.
• Reduced ~4400 soil bacterial species into two groups crucial for nitrate metabolism modeling.
• Connected microbial group specialization to community responses to perturbations.

Conclusions:

• The developed approach effectively simplifies complex microbiome data, revealing key microbial players and their functional roles.
• This method facilitates the discovery of molecular mechanisms underlying human and environmental health.
• Demonstrated a generalizable strategy for function-informed dimensionality reduction in biological systems.