Hypothesis testing for phylogenetic composition: a minimum-cost flow perspective.

Area of Science:

• Microbiome research
• Phylogenetic analysis
• Statistical methods

Background:

• Comparing microbial composition between populations is crucial in microbiome studies.
• Phylogenetic information is increasingly leveraged for these comparisons.
• Existing methods like PERMANOVA have limitations with sparse differences.

Purpose of the Study:

• To develop a novel statistical test for microbial compositional data using phylogenetic information.
• To address the limitations of existing methods when compositional differences are sparse.
• To propose a powerful test for detecting sparse phylogenetic differences in microbiome studies.

Main Methods:

• A minimum-cost flow perspective is adopted for two-sample testing.
• Multivariate analysis of variance with permutation (PERMANOVA) using phylogenetic distances is analyzed.
• A new maximum type test, the detector of active flow on a tree, is proposed and investigated.

Main Results:

• PERMANOVA is identified as a sum-of-squares test, effective for dense alternatives.
• The proposed detector of active flow on a tree demonstrates superior power against sparse phylogenetic compositional differences.
• The new method shows practical merit through simulations and application to a human intestinal biopsy dataset.

Conclusions:

• The detector of active flow on a tree is a powerful and potentially optimal method for microbiome compositional analysis with sparse differences.
• This new method enhances the ability to detect subtle yet significant phylogenetic shifts in microbial communities.
• The findings have implications for understanding diseases like ulcerative colitis through microbiome profiling.