themetagenomics

themetagenomics applies topic modeling and multilevel, fully Bayesian regression to 16S rRNA amplicon sequencing data in R to identify co-occurring taxa (topics) and infer within-topic functional potential associated with metabolic pathways.

Key Features:

• Topic modeling: Identifies groups of co-occurring taxa ("topics") from 16S rRNA amplicon sequencing data while preserving microbial community configuration.
• Within-topic functional prediction: Infers functional content within identified topics by estimating interactions between topics and metabolic pathways using a multilevel, fully Bayesian regression model.
• PICRUSt and Tax4Fun integration: Implements PICRUSt in R and integrates Tax4Fun to generate predicted metagenomic functions from taxonomic profiles.

Scientific Applications:

• Capture co-occurring taxa: Identify and characterize taxa sets that co-occur and contribute to sample characteristics such as disease states or environmental conditions.
• Uncover functional potential: Link taxonomic topics to predicted gene functions and metabolic pathways to determine functional potential within communities.
• Link communities to host features: Associate topic composition and predicted functions with host or environmental features.
• Temporal analysis: Analyze dynamics of co-occurring taxa and their predicted functions in time-series studies.
• Disease-focused analysis: Apply topic-based taxonomic and functional analyses to datasets related to inflammatory bowel disease and oral cancer.

Methodology:

Topic modeling of 16S rRNA sequencing data; multilevel, fully Bayesian regression to estimate interactions between topics and metabolic pathways for within-topic functional inference; R implementation of PICRUSt and integration with Tax4Fun for predicted metagenomic functions.