The paper introduces Bracken, a new statistical method for estimating species abundance in metagenomics data. Bracken addresses the challenge of accurately estimating species abundance when many reads cannot be unambiguously assigned to a single strain or species due to high sequence similarity and ongoing taxonomic revisions. The method probabilistically redistributes reads from higher taxonomic levels (e.g., genus) to the species level, using probabilities derived from the Kraken database. This approach significantly improves the accuracy of species-level abundance estimates compared to Kraken's raw read assignments, which often underestimate species abundance. The authors demonstrate the effectiveness of Bracken through experiments on simulated and real metagenomics datasets, showing that it can produce accurate estimates even in cases with multiple near-identical species. The method is freely available and can be applied to various taxonomic levels, from species to phylum.The paper introduces Bracken, a new statistical method for estimating species abundance in metagenomics data. Bracken addresses the challenge of accurately estimating species abundance when many reads cannot be unambiguously assigned to a single strain or species due to high sequence similarity and ongoing taxonomic revisions. The method probabilistically redistributes reads from higher taxonomic levels (e.g., genus) to the species level, using probabilities derived from the Kraken database. This approach significantly improves the accuracy of species-level abundance estimates compared to Kraken's raw read assignments, which often underestimate species abundance. The authors demonstrate the effectiveness of Bracken through experiments on simulated and real metagenomics datasets, showing that it can produce accurate estimates even in cases with multiple near-identical species. The method is freely available and can be applied to various taxonomic levels, from species to phylum.