PMGs

PMGs aggregates operational taxonomic units (OTUs) into Principal Microbial Groups using principal balances to model the compositional structure of microbiome relative abundance data for downstream biomarker discovery and disease prediction.

Key Features:

• Compositional Data Handling: Uses principal balances to manage the compositional structure of microbiome data, addressing high dimensionality and sparsity.
• OTU Grouping by Relative Abundance: Groups OTUs based solely on their relative abundances without relying on phylogenetic trees or predefined taxonomic classifications.
• Dimensionality Reduction: Aggregates OTUs into Principal Microbial Groups to reduce data dimensionality while preserving essential compositional information.
• Aggregation for Microbial Balances: Provides an aggregation procedure to construct microbial balances that can be applied in disease prediction models.

Scientific Applications:

• Biomarker Discovery: Facilitates identification of microbial biomarkers associated with specific health conditions by analyzing relative abundance–based groups.
• Disease Prediction: Enables construction of microbial balances for predictive models that correlate microbial community structure with disease states.
• Cirrhosis Microbiome Analysis: Can be applied to cirrhosis datasets to perform coherent analyses aimed at identifying candidate biomarkers within the human microbiota.

Methodology:

Groups OTUs based on their relative abundances using principal balances to create Principal Microbial Groups and construct microbial balances that reflect the compositional nature of the data.