Sylph samples 1/c of the k -mers for metagenomes and reference genomes, resulting in sketches. Black reads are from organisms not present in the database at the species level. Sylph computes the containment of each reference genome’s sketch within the metagenome’s sketch. (2) The coverage distribution of genome A's k -mers within metagenome sample B is modeled with a zero-inflated Poisson distribution, where zero inflation is because of divergent k -mers between the genome and the metagenome having 0 coverage. Sylph infers the effective coverage λ and estimates the coverage-adjusted ANI instead of the commonly used naive ANI, which is ignorant of read-sampling effects. This is repeated for each reference genome against each metagenome sample. (3) Sylph can output the ANI estimates directly (sylph query) or a metagenomic profile (sylph profile), which reassigns shared k -mers and reports abundances for genomes with >95% ANI.