Comparative performance of reference-based metagenomic tools to identify species-level taxa among families of bacteria: benchmarking Mycobacteriaceae and Neisseriaceae.

Metapresence and sylph outperform other tools in detecting closely related bacterial species in simulated metagenomic samples from Mycobacteriaceae and Neisseriaceae families.

• Why it matters: Accurate identification of low-abundance, species-level bacteria in metagenomic data is crucial for microbial ecology, evolution, and clinical diagnostics, but remains challenging due to closely related genomes and reference database complexities.
• What they did: The study evaluated five reference-based methods—YACHT, Kraken2/bracken, Metapresence, MetaPhlAn4, and sylph—using simulated metagenomic samples from two bacterial families with closely related species, focusing on detection limits and robustness.
• The result: Metapresence and sylph demonstrated the best overall performance, with sylph requiring coverage greater than approximately 0.1× for detection and Metapresence capable of detecting hundreds of reads, enabling more accurate species identification in diverse sample types.