Mar 19, 2026

CAMI III launches with the release of a new benchmark dataset, named CAMI III toy human gut, prior to the opening of the challenges. It is designed to support method development, testing of scalability, and familiarization with CAMI formats ahead of the upcoming benchmarking challenges.

The metagenome dataset represents human gut microbiomes of 10 individuals, with two metagenome samples per individual. The microbiomes span multiple domains of life and include host, viruses, and plasmids. It offers both short- and long-read shotgun metagenomic data together with gold standard assemblies, genome and taxonomic bins, and taxonomic profiles.

We invite the research community to explore the dataset and share any feedback or observations you may have. For assessing tool results on these data, the new CAMI portal offers web-based evaluations of single- and cross-sample assemblies via its “Submit” function. Support for genome binning, taxonomic binning, and profiling evaluations — using the new NCBI taxonomic ranks — will be enabled soon. You can also share your thoughts via the contact form.

Note! This dataset was created from public microbial genomes only (different from the challenge data sets, which include unpublished data). Assessing reference-based tools on it using reference data collections that include the same data will lead to exaggerated performance estimates.

Technical details and download instructions are available at https://cami-challenge.org/datasets/Toy Human Gut.

The CAMI Team