[link to article]
DartUniFrac rethinks UniFrac computation by connecting phylogeny-aware microbiome distances to weighted Jaccard similarity and using fast weighted MinHash sketching to make large-scale analysis practical.
The method is implemented in Rust with CPU parallelization and GPU acceleration, enabling UniFrac-scale analyses across millions of samples and billions of taxa. This opens the door to routine microbiome meta-analysis, large-scale resampling, and phylogeny-aware comparisons at dataset sizes that were previously computationally out of reach. Check the paper details and code below.
Congratulations to Jianshu and co-authors!
Paper: https://www.nature.com/articles/s41587-026-03260-8
Code: https://github.com/jianshu93/DartUniFrac
