Comparative performance of reference-based metagenomic tools to identify species-level taxa among families of bacteria: benchmarking <i>Mycobacteriaceae</i> and <i>Neisseriaceae</i>
Bibliographic record
Abstract
Abstract Hypotheses concerning the ecology and evolution of bacteria commonly relate to the presence and abundance of species in various settings and conditions. Shotgun metagenomics may address these hypotheses, which previously relied on PCR or culture. However, the problem of determining the presence/absence of a given species of interest is not trivial, particularly when closely related species are present in the reference database or metagenomic sample. Reference-based methods to detect species-level taxa mostly rely on thresholding of aligned reads or mapped k-mers or derivative metrics like genomic coverage, and create a trade-off between recall/completeness and precision/purity. New methods for species-level profiling (YACHT, metapresence and sylph) have recently been published. Here we test the performance of these methods, along with Kraken2/bracken and MetaPhlAn4, to detect related species of interest using simulated metagenomic samples from genomes in the families Mycobacteriaceae and Neisseriaceae , which contain closely related genomes. Among methods tested, metapresence, when used with an alignment quality filter, and sylph offer the best overall performance. Sylph maintains high precision but requires a depth of coverage greater than approximately 0.1x to reliably detect a genome’s presence. Metapresence has a lower limit of detection of hundreds of reads but this is balanced against relatively lower precision. Both methods are relatively robust to the presence of reads from genomes outside the groups of interest. We demonstrate the application of these methods in two real-world datasets: a mycobacterial community in a drinking water system and the community of Neisseriaceae present in the human oral cavity. Importance Detecting which bacterial species of interest are present in a given sample is fundamental to studies of microbial ecology and evolution, and to applied microbiology (e.g. clinical diagnostics). Culture-dependent and independent (e.g. PCR) approaches are increasingly complemented by metagenomic approaches, but methods to accurately identify specific low-abundance species-level genomes in a shotgun metagenomic sample are still being refined. Here we comprehensively test YACHT, Kraken2/bracken, metapresence, MetaPhlAn4, and sylph using two simulated datasets of bacterial families, Mycobacteriaceae and Neisseriaceae that contain closely related species. Our simulations exploit natural genomic diversity to create a challenging benchmark. We demonstrate that metapresence and sylph perform best, with the former being well-suited to low-biomass host-associated datasets, and the latter with environmental metagenomic samples. This study is the first extensive benchmark of these methods for this use case, and demonstrates these methods can accurately identify closely related species of interest.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".