Statistical methods for detecting differentially abundant features in clinical metagenomic samples.
James Robert White1, Niranjan Nagarajan, Mihai Pop
1Applied Mathematics and Scientific Computation Program, Center for Bioinformatics and Computational Biology, University of Maryland, College Park, Maryland, United States of America.
Plos Computational Biology
|April 11, 2009
Summary
Metastats is a new statistical method for comparing complex microbial communities. It accurately identifies differences in large metagenomic datasets, outperforming existing methods, especially for sparse data.
Area of Science:
- Microbiology
- Bioinformatics
- Computational Biology
Background:
- Characterizing microbial communities is crucial for understanding symbiotic relationships.
- Accurate computational tools are needed to compare large, complex metagenomic datasets.
- Existing methods struggle with high-complexity environments and sparsely sampled features.
Purpose of the Study:
- To develop a robust statistical method for comparing clinical metagenomic samples.
- To identify differentially abundant features in complex microbial communities.
- To provide a tool for analyzing large-scale microbiome data.
Main Methods:
- Developed Metastats, a statistical method using false discovery rate and Fisher's exact test.
- Applied Metastats to compare metagenomic datasets including 16S rRNA, COG profiles, and metabolic subsystems.
- Validated performance through simulations against existing methods.
Main Results:
- Metastats demonstrates strong performance, outperforming other methods for sparse counts.
- Identified novel differences in human gut microbiomes of obese vs. lean subjects.
- Provided the first statistically rigorous assessment of COG and subsystem differences in infant vs. mature gut microbiomes.
Conclusions:
- Metastats is a robust and accurate tool for analyzing clinical metagenomic data.
- The method is effective across diverse datasets with varying complexity and sampling levels.
- Metastats can also be applied to digital gene expression studies, offering broad utility.
