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

Abstract

Background: The human gut microbiome has been suggested to affect human health and thus has received considerable attention. To clarify the structure of the human gut microbiome, clustering methods are frequently applied to human gut taxonomic profiles. Enterotypes, i.e., clusters of individuals with similar microbiome composition, are well-studied and characterized. However, only a few detailed studies on assemblages, i.e., clusters of co-occurring bacterial taxa, have been conducted. Particularly, the relationship between the enterotype and assemblage is not well-understood. Results: In this study, we detected gut microbiome assemblages using a latent Dirichlet allocation (LDA) method. We applied LDA to a large-scale human gut metagenome dataset and found that a 4-assemblage LDA model could represent relationships between enterotypes and assemblages with high interpretability. This model indicated that each individual tends to have several assemblages, three of which corresponded to the three classically recognized enterotypes. Conversely, the fourth assemblage corresponded to no enterotypes and emerged in all enterotypes. Interestingly, the dominant genera of this assemblage (Clostridium, Eubacterium, Faecalibacterium, Roseburia, Coprococcus, and Butyrivibrio) included butyrate-producing species such as Faecalibacterium prausnitzii. Indeed, the fourth assemblage significantly positively correlated with three butyrate-producing functions. Conclusions: We conducted an assemblage analysis on a large-scale human gut metagenome dataset using LDA. The present study revealed that there is an enterotype-independent assemblage.

Stratification of the Gut Microbiota Composition Landscape across the Alzheimer's Disease Continuum in a Turkish Cohort

“…Both parameters continued to partition the community without reaching a clear optimum; we therefore presented the first 10 subgroups, also to allow for comparison with the results of topological analysis. The perplexity parameter not reaching a clear optimum is unexpectedly consistent with recent publications using LDA in microbial ecology ( 47 – 49 ). Bacterial probability distributions (ranked by a probability of ≥1% in descending order) across the subgroups are displayed in Fig.…”

“…As a third method, we tested the LDA potential to stratify gut microbiota of the cohort participants. This unsupervised machine learning technique is increasingly finding acceptance in the field of microbiome studies ( 47 – 49 ) for its unique ability to reveal latent or hidden groups within the data cloud. Figure S4 in the supplemental material shows the LDA model’s perplexity parameter and log-likelihood values to find the optimal number of clusters.…”

Parallelized Latent Dirichlet Allocation Provides a Novel Interpretability of Mutation Signatures in Cancer Genomes

“…Owing to its versatility, there exist many LDA-based models that are widely used in the field of bioinformatics [ 24 , 25 ]. Among them, supervised LDA [ 20 ] should be noted as an extended model closely related to this research.…”

Revealing General Patterns of Microbiomes That Transcend Systems: Potential and Challenges of Deep Transfer Learning

“…In particular, we advocate using deep learning models that have already proven adept at transfer learning in natural language processing (NLP) to better understand baseline interactions present in microbiome data ( 19 – 21 ). There exist easily drawn parallels between natural language data and microbiome data, namely, that documents are equivalent to biological samples, words to taxa, and topics to microbial neighborhoods ( 22 – 24 ). While other language-inspired algorithms such as topic modeling methods have been employed to identify latent variables in microbiomes, deep learning approaches offer the unique advantage that they scale with the amount of data available, allowing our understanding and our predictive models to scale alongside the genomic revolution ( 25 ).…”