A Novel Sparse Compositional Technique Reveals Microbial Perturbations.

Area of Science:

• Microbiome research
• Computational biology
• Statistical ecology

Background:

• Microbiome studies aim to link microbial communities to outcomes but face challenges like high-dimensionality and data compositionality.
• Beta diversity metrics are crucial for analyzing microbial sample distances but vary in effectiveness.
• Existing methods struggle with the unique characteristics of microbiome data.

Purpose of the Study:

• To introduce a novel compositional beta diversity metric, robust Aitchison PCA, for microbiome analysis.
• To demonstrate the advantages of compositional transformations in beta diversity calculations.
• To improve effect size, classification accuracy, and robustness to sequencing depth in microbiome data.

Main Methods:

• Utilized centered log-ratio transformation and matrix completion to develop robust Aitchison PCA.
• Applied simulations to validate the benefits of compositional transformations.
• Tested the metric on real microbiome datasets with varying sample sizes.

Main Results:

• Robust Aitchison PCA showed improved effect size and classification accuracy compared to existing methods.
• The metric demonstrated enhanced robustness to varying sequencing depths.
• Successfully retained feature loadings, revealing important intercommunity niche features.

Conclusions:

• Robust Aitchison PCA effectively handles the sparse, compositional nature of microbiome data.
• This metric offers high discriminatory power and salient feature ranking for microbial niches.
• Open-source software and a QIIME 2 plugin are available for this analysis.