Quantifying Metagenomic Strain Associations from Microbiomes with Anpan

Area of Science:

• Microbiome research
• Genomic epidemiology
• Computational biology

Background:

• Microbial strain variation significantly impacts phenotypes and health.
• Existing inferential methods struggle with complex strain-level metagenomic data.
• Quantifying genetic and genomic differences at the strain level is crucial for understanding microbial epidemiology.

Purpose of the Study:

• To develop and validate quantitative methods for microbiome strain epidemiology.
• To address challenges in analyzing high-dimensional, variable, and phylogenetically related strain data.
• To identify strain-specific genetic elements, lineages, and pathways associated with host phenotypes.

Main Methods:

• Anpan utilizes adaptive filtering with linear models for gene carriage analysis.
• Phylogenetic generalized linear mixed models assess sub-species lineage associations.
• Random effects models identify phenotype-associated pathway retention or loss.

Main Results:

• Simulations show Anpan offers improved effect size estimation and reduced false positive rates.
• Application to colorectal cancer (CRC) data identified adaptive genes and phylogenetic effects.
• Findings complement and extend known species-level microbiome CRC biomarkers.

Conclusions:

• Anpan provides robust quantitative methods for strain-level microbiome epidemiology.
• The R library facilitates the study of microbial genetic epidemiology across diverse contexts.
• These methods enhance our understanding of microbial contributions to human health and disease.