Skip to main content
MaAsLin3 is designed to determine associations between microbial features (e.g., taxonomic compositions, functional profiles) and metadata (e.g., clinical or environmental variables). MaAsLin3 offers enhanced performance, greater flexibility in modeling, and improved stability. If you use the MaAsLin3 tool, please cite the reference pre-print manuscript:
Nickols WA, Kuntz T, Shen J, Maharjan S, Mallick H, Franzosa EA, Thompson KN, Nearing JT, Huttenhower C. MaAsLin 3: refining and extending generalized multivariable linear models for meta-omic association discovery. Nat Methods. 2026 Mar;23(3):554-564. doi: 10.1038/s41592-025-02923-9. Epub 2026 Jan 15. PMID: 41540124; PMCID: PMC12982127.

​
Microbiome Association Detection with MaAsLin3

Relative to MaAsLin 2, MaAsLin 3 introduces the ability to quantify and test for both abundance and prevalence associations while better accounting for compositionality. By incorporating generalized linear models, MaAsLin 3 accommodates most modern study designs, including cross-sectional and longitudinal studies. it efficiently identifies meaningful associations while controlling for confounding factors and handling complex data structures. Through the Cosmos-Hub, researchers can easily upload their raw fastq data for profiling and configure MaAsLin3 through the Comparative Analysis module. Results can be interpreted via automatically generated tables, heatmaps, and summary files—ultimately streamlining the discovery of significant microbiome–metadata relationships. MaAsLin3 workflow two inputs, which are generated in the Cosmos-Hub:

​
Profiling Data (feature table)

A table where rows represent samples and columns represent microbial features (e.g., species, genes, pathways). Each cell in the table provides the abundance of that feature in that sample. MaAsLin3 can take raw counts or relative abundances as input. This is the default output of a comparative analysis within the Cosmos-Hub.

​
Metadata

Metadata that describes is required for this analysis, and tutorials on metadata input can be found here.

Running the MaAsLin3 Workflow

Interpreting MaAsLin3 Results