Human Microbiome Mixture Analysis Using Weighted Quantile Sum Regression.

Area of Science:

• Microbiome research
• Statistical modeling in biology
• Human health and disease

Background:

• Current microbiome studies often analyze diversity metrics or individual taxa, neglecting correlated relationships between microbes.
• Understanding the complex interplay of microbial communities is crucial for deciphering their role in health and disease.

Purpose of the Study:

• To apply and evaluate the random subset weighted quantile sum regression with repeated holdouts (WQS_RSRH) method for analyzing microbiome data.
• To assess the association between the gut microbiome and body mass index (BMI) using real-world data.
• To compare the performance of WQS_RSRH against standard microbiome analysis methods.

Main Methods:

• Utilized WQS_RSRH, a mixture method designed for 'omic data, on Human Microbiome Project amplicon sequencing data.
• Simulated a binary variable associated with 20 operational taxonomic units (OTUs) to test method performance.
• Compared WQS_RSRH with other standard methods and applied it to the Growth and Obesity Cohort in Chile dataset.

Main Results:

• WQS_RSRH demonstrated high sensitivity (75%) and specificity (70%) in identifying associated OTUs in simulated data, outperforming comparison methods.
• The analysis revealed an inverse association between BMI and the gut microbiome mixture in the Chilean cohort.
• Key genera identified as important in this negative association included Bacteroides, Clostridium, Prevotella, and Ruminococcus.

Conclusions:

• WQS_RSRH provides a robust approach to analyze the mixture effects of microbial taxa while accounting for correlations and covariates.
• The method successfully identified key microbial players associated with BMI, offering insights into obesity's microbiome links.
• WQS_RSRH is a valuable tool for advancing microbiome research and understanding its impact on human health.