microbiomeDASim: Simulating longitudinal differential abundance for microbiome data.

Area of Science:

• Microbiology
• Bioinformatics
• Statistical Modeling

Background:

• Longitudinal metagenomic studies are increasing to understand microbial community dynamics.
• Existing data types (whole metagenome shotgun, marker-gene surveys) necessitate advanced statistical methods for time-interval analysis.
• Study design, including sampling frequency, impacts the ability to detect biological effects.

Purpose of the Study:

• To present a novel simulation paradigm for longitudinal microbiome data analysis.
• To introduce the R package microbiomeDASim for simulating differential abundance.
• To evaluate statistical methods for analyzing time-series microbiome data.

Main Methods:

• Development of a simulation paradigm implemented in the R package microbiomeDASim.
• Simulation of longitudinal differential abundant microbiome features with controllable functional forms and signal-to-noise ratios.
• Evaluation of statistical method performance using defined metrics, focusing on metaSplines.

Main Results:

• microbiomeDASim enables flexible simulation of complex microbiome dynamics.
• The package allows for the generation of synthetic datasets to test statistical approaches.
• Performance metrics were established to assess the success of differential abundance detection methods.

Conclusions:

• microbiomeDASim provides a valuable tool for researchers designing and analyzing longitudinal microbiome studies.
• The simulation framework supports the development and validation of statistical methodologies for time-series microbiome data.
• This approach aids in optimizing study design and understanding microbial community temporal changes.