Impact of Data and Study Characteristics on Microbiome Volatility Estimates

Area of Science:

• Microbiology
• Bioinformatics
• Genomics

Background:

• The human microbiome, comprising diverse microorganisms, plays a crucial role in health.
• Microbiome composition and temporal variability (volatility) are associated with human health outcomes.
• The impact of data acquisition factors like sampling intervals and sequencing depth on microbiome volatility estimates remains unclear.

Purpose of the Study:

• To assess the effect of unbalanced sampling intervals and differential read depth on microbiome volatility estimations.
• To compare various volatility measures under different data subsampling conditions.
• To provide recommendations for analyzing longitudinal microbiome data.

Main Methods:

• Utilized four public gut and vaginal microbiome time-series datasets.
• Subsampled datasets to simulate varying sampling intervals and read depths.
• Compared additive, multiplicative, centered log ratio (CLR)-based, qualitative, and distance-based volatility measures.

Main Results:

• Longer sampling intervals increased quantitative volatility measures, especially for common microbes, but not qualitative or distance-based measures.
• Lower sequencing read depth decreased multiplicative, CLR-based, and qualitative volatility measures, particularly for less common microbes.
• Findings highlight the influence of data processing choices on microbiome volatility metrics.

Conclusions:

• Sampling intervals and read depth significantly influence microbiome volatility estimates.
• Strategic subsampling can be a valuable sensitivity analysis tool for longitudinal microbiome studies.
• Accurate assessment of microbiome volatility requires careful consideration of data acquisition and processing parameters.