Revealing the microbial assemblage structure in the human gut microbiome using latent Dirichlet allocation.

Area of Science:

• Microbiology
• Computational Biology
• Human Health

Background:

• The human gut microbiome significantly influences health and is extensively studied.
• Enterotypes, clusters of individuals by microbiome composition, are well-characterized.
• Assemblages, clusters of co-occurring bacteria, are less understood, especially their relation to enterotypes.

Purpose of the Study:

• To investigate gut microbiome assemblages using a statistical approach.
• To explore the relationship between enterotypes and bacterial assemblages.
• To identify novel patterns in gut microbial community structure.

Main Methods:

• Latent Dirichlet Allocation (LDA) was employed to detect gut microbiome assemblages.
• LDA was applied to a large-scale human gut metagenome dataset.
• A 4-assemblage model was developed to analyze enterotype-assemblage relationships.

Main Results:

• A 4-assemblage LDA model demonstrated high interpretability for enterotype-assemblage relationships.
• Three assemblages aligned with known enterotypes, while a fourth was enterotype-independent.
• The independent assemblage, rich in butyrate producers like Faecalibacterium prausnitzii, correlated with butyrate production functions.

Conclusions:

• Assemblage analysis using LDA revealed an enterotype-independent gut microbiome structure.
• This finding highlights a previously unrecognized dimension of gut microbial organization.
• The study provides a new framework for understanding gut microbiome complexity and its health implications.