Principal microbial groups: compositional alternative to phylogenetic grouping of microbiome data

Area of Science:

• Microbiology
• Bioinformatics
• Statistical analysis

Background:

• Microbiome data analysis faces challenges due to high dimensionality, sparsity, and compositional nature.
• Existing taxon grouping methods for microbiome data present difficulties in disease correlation.
• Operational Taxonomical Units (OTUs) require effective aggregation for robust analysis.

Purpose of the Study:

• To present a novel approach for grouping Operational Taxonomical Units (OTUs) based on relative abundances.
• To introduce Principal Microbial Groups (PMGs) as an alternative to traditional taxon grouping.
• To demonstrate the utility of PMGs for dimensionality reduction and disease prediction in microbiome studies.

Main Methods:

• A new procedure for grouping OTUs using principal balances on compositional data.
• Development of Principal Microbial Groups (PMGs) independent of phylogenetic information.
• Application of PMGs for dimensionality reduction and as an aggregation method for microbial balances.

Main Results:

• The proposed method effectively groups OTUs based on relative abundances, creating Principal Microbial Groups (PMGs).
• PMGs serve as a dimensionality reduction technique for compositional microbiome data.
• The approach facilitates disease prediction by enabling the construction of microbial balances.

Conclusions:

• Principal Microbial Groups (PMGs) offer a coherent data analysis framework for identifying microbial biomarkers.
• This method overcomes limitations of traditional taxon grouping and addresses statistical challenges in microbiome research.
• The PMG approach provides a flexible alternative for analyzing complex microbiome datasets, as illustrated with cirrhosis data.