Metastats

Metastats identifies differentially abundant features in metagenomic count data from sequencing technologies to compare two treatment populations in clinical and environmental studies.

Key Features:

• False Discovery Rate (FDR) Application: Metastats applies false discovery rate (FDR) correction to control type I error when testing multiple features in high-complexity datasets.
• Handling Sparse Data with Fisher’s Exact Test: For sparsely sampled features, Metastats uses Fisher's exact test for significance assessment.
• Robust Performance Across Simulations: Simulations demonstrated improved performance over previously used methods, particularly for identifying features with sparse counts.
• Versatility Across Datasets: Applicable to metagenomic and digital gene expression datasets such as Serial Analysis of Gene Expression (SAGE), and robust across varied complexity and sampling levels.

Scientific Applications:

• 16S rRNA Survey: Applied to 16S rRNA surveys comparing obese and lean human gut microbiomes, revealing differences not reported in the original analysis.
• COG Functional Profiles: Provided a statistically rigorous assessment of differences between infant and mature gut microbiomes using COG functional profiles.
• Bacterial and Viral Metabolic Subsystem Data: Analyzed bacterial and viral metabolic subsystem data inferred from random sequencing of 85 metagenomes to evaluate metabolic subsystems across populations.

Methodology:

Metastats compares count data from two treatment groups across samples from multiple subjects to detect differentially abundant features, applying Fisher's exact test for sparsely sampled features and false discovery rate (FDR) correction for multiple testing.