Mash

Mash estimates genome and metagenome distances using a MinHash dimensionality-reduction technique, computing pairwise mutation distances with P value significance testing to enable rapid comparison of large sequence collections.

Key Features:

• MinHash sketching: Transforms genomic sequences into compact representative sketches using a MinHash dimensionality-reduction approach.
• Pairwise mutation distance calculation: Computes global mutation distance estimates between sequences from their sketches.
• P value significance testing: Assigns statistical significance to distance estimates via P value testing.
• Sequencing technology support: Operates on assembled or unassembled Illumina reads, Pacific Biosciences long-reads, and Oxford Nanopore sequences.
• Scalable clustering and search: Enables rapid clustering and real-time database searches across large collections (e.g., 54,118 NCBI RefSeq genomes clustered in ~33 CPU hours).
• Metagenome compositional clustering: Scales to cluster hundreds of metagenomic samples based on compositional similarity.

Scientific Applications:

• Genome clustering: Groups large sets of genomes by estimated global mutation distances.
• Real-time sequence database querying: Performs rapid searches using sketches derived from assembled or unassembled Illumina, Pacific Biosciences, and Oxford Nanopore reads.
• Metagenome analysis: Compares and clusters metagenomic samples by compositional similarity for community profiling.
• Large-scale comparative genomics: Facilitates comparative analyses across thousands of genomes such as NCBI RefSeq.

Methodology:

Mash constructs compact MinHash sketches from input sequences, computes pairwise mutation distances between sketches with P value significance testing, and uses those distances for rapid clustering and database searches.