A Mixed Effect Similarity Matrix Regression Model (SMRmix) for Integrating Multiple Microbiome Datasets at Community...

Area of Science:

• Microbiome Research
• Statistical Bioinformatics
• Computational Biology

Background:

• Human microbiota plays a vital role in health and disease.
• Individual microbiome studies yield inconsistent results due to small sample sizes and heterogeneity.
• There is a critical need for integrative analysis of multiple microbiome datasets, accounting for study variations.

Purpose of the Study:

• To develop a statistical method for integrating multiple microbiome datasets.
• To address the challenge of study heterogeneity in microbiome research.
• To identify community-level microbiome shifts associated with health outcomes.

Main Methods:

• Developed a mixed-effect similarity matrix regression (SMRmix) approach.
• SMRmix builds upon the microbiome kernel association test but accommodates multiple studies.
• The method is designed to consolidate findings from diverse microbiome datasets.

Main Results:

• SMRmix demonstrated well-controlled type I error and higher statistical power in simulations.
• Analysis of HIV data from 17 studies confirmed associations between gut microbiome, HIV infection, and MSM status.
• Analysis of colorectal cancer data from 11 studies revealed significant microbiome dysbiosis in affected individuals.

Conclusions:

• SMRmix enables the integration of multiple microbiome studies, effectively managing heterogeneity.
• This provides a powerful tool for uncovering consistent disease-microbiome associations.
• The method enhances the reliability and power of microbiome-wide association studies.