mi-Mic: a novel multi-layer statistical test for microbiota-disease associations

Area of Science:

• Microbiology
• Bioinformatics
• Statistical Analysis

Background:

• Microbiome research involves complex statistical challenges.
• Differential abundance analysis is crucial for understanding microbial community changes.
• Existing methods struggle with high dimensionality, sparsity, and taxonomic structures.

Purpose of the Study:

• To introduce mi-Mic, a novel statistical approach for microbiome differential abundance analysis.
• To address limitations of current methods, including large numbers of tests, sparsity, and taxonomic relationships.
• To improve the accuracy and reliability of identifying microbial differences.

Main Methods:

• mi-Mic converts microbial counts into a cladogram of means.
• It utilizes a priori tests on cladogram levels to detect overall relationships.
• Mann-Whitney tests are applied to significant paths or leaves for detailed analysis.

Main Results:

• mi-Mic demonstrates significantly higher true to false positive ratios compared to existing tests.
• Performance was validated using a novel real-to-shuffle positive score.
• The method effectively handles sparsity and taxonomic structures in microbiome data.

Conclusions:

• mi-Mic provides a robust and accurate method for microbiome differential abundance analysis.
• The approach effectively integrates taxonomic information, improving statistical power.
• mi-Mic represents a significant advancement for microbiome data interpretation.