False discovery rate control incorporating phylogenetic tree increases detection power in microbiome-wide multiple...

Area of Science:

• Microbiome research
• Bioinformatics
• Statistical genetics

Background:

• Next-generation sequencing generates vast microbiome data.
• Microbiome data possesses inherent phylogenetic relationships among bacterial species.
• Phylogenetic information can enhance microbiome-wide association studies (MWAS).

Purpose of the Study:

• To propose a novel false discovery rate (FDR) control procedure for MWAS that leverages phylogenetic tree structure.
• To improve the statistical power of detecting microbe-phenotype associations.

Main Methods:

• Developed a hierarchical model with a structure-based prior distribution utilizing the phylogenetic tree.
• Implemented a new FDR control procedure incorporating phylogenetic information.
• Utilized extensive simulations and real microbiome datasets for validation.

Main Results:

• The proposed method significantly improves statistical power for detecting associated bacterial species while controlling FDR.
• The method demonstrates robustness even with mis-specified or non-informative phylogenetic trees.
• Identified substantially more alcohol-drinking associated bacterial species compared to traditional methods.

Conclusions:

• Incorporating phylogenetic structure into FDR control procedures is effective for microbiome data analysis.
• The developed method enhances the discovery of microbe-phenotype associations in MWAS.
• The StructFDR R package is available for public use.