Mash Screen measures how much of a reference genome appears in mixed sequencing data
Mash Screen estimated how much of a genome was represented in sequencing data containing genetic material from a mixture of organisms. Brian Ondov and colleagues described the tool in Genome Biology on 5 November 2019, setting out uses in contamination screening and searches through previously collected data for novel genomes. The tool addressed a limitation of MinHash, an algorithm that estimates how closely two genomes or mixtures of genomes resemble each other.
A mixture of genetic material from different organisms is called a metagenome. MinHash can estimate resemblance between two genomes or metagenomes, but it cannot reliably measure how much of a particular genome is represented within a metagenome. Mash Screen was designed to estimate that measure, known as containment.
The tool works with sequencing reads, the fragments produced by sequencing, whether they remain separate or have been assembled into longer sequences. It can measure containment for genomes and for proteomes, the sets of proteins associated with organisms. This lets researchers estimate representation within a mixed dataset rather than rely on an estimate of overall resemblance.
The authors described contamination screening as one use: looking for unwanted genomes in sequencing data. They also described searching metagenomes collected earlier for novel genomes, allowing existing datasets to be examined again for material of interest.
They used Mash Screen to provide containment estimates for every genome in NCBI RefSeq, a curated reference-genome database, within every metagenome in SRA, the Sequence Read Archive of public raw sequencing data. As a demonstration of genome discovery, they identified a novel species of polyomavirus, a member of a family of small DNA viruses, in a public metagenome.