| name | metabuli |
| description | Metabuli — metagenomic taxonomic classification via joint DNA + amino-acid (metamer) analysis for sensitive homology and specific close-taxon resolution. Use when: (1) Choosing a classifier beyond DNA-only (Kraken2) or AA-only (Kaiju), (2) Profiling understudied clades where DNA k-mers fail, (3) CAMI- style or plant/environment datasets needing both sensitivity and specificity, (4) Writing Methods for joint DNA/AA classification. Upstream: https://github.com/steineggerlab/Metabuli. Complementary to metaphlan / kraken2 / kaiju — do not merge tables unlabeled.
|
| license | MIT |
| category | analysis-tools |
| tags | ["taxonomy","Metabuli","DNA","amino-acid","classification"] |
| upstream | https://github.com/steineggerlab/Metabuli |
| stage | taxonomy |
Metabuli
Upstream: steineggerlab/Metabuli ·
https://metabuli.steineggerlab.com
Citation
Kim, J. & Steinegger, M. Metabuli: sensitive and specific metagenomic
classification via joint analysis of amino acid and DNA. Nat. Methods 21,
971–973 (2024). https://doi.org/10.1038/s41592-024-02273-y
See also docs/references.md.
Analytical thinking
Classical dilemma:
| Signal | Strength | Weakness |
|---|
| DNA k-mers (Kraken-like) | Distinguishes close, well-studied taxa | Misses remote homology |
| AA / 6-frame (Kaiju-like) | Sensitive to understudied clades | Weaker close-species separation |
Metabuli jointly analyzes DNA and AA via metamers, aiming to keep both
sensitivity and specificity. In CAMI2 plant-associated benchmarks it recovered
nearly all classifications of strong DNA- and AA-only tools.
Still database-bound: unclassified reads ≠ empty community. Prefer MetaPhlAn
when you need clade-specific marker abundances for pathways (HUMAnN); prefer
Metabuli/Kraken/Kaiju when you need read-level classification against large
nucleotide/protein DBs.
Prerequisites
| Requirement | Notes |
|---|
| DB | Metabuli database matching software version |
| RAM / disk | Large DBs — plan like Kraken-scale resources |
| Input | Preferably host-depleted FASTQ |
How to run
metabuli classify \
$METABULI_DB \
temp/hr/${id}_1.fastq temp/hr/${id}_2.fastq \
temp/metabuli/${id} \
--threads 16
Decision tree
Taxonomic classification goal?
├─ Marker-gene relative abundance → metaphlan / motus
├─ Fast DNA k-mer + Bracken abundance → kraken2 + bracken
├─ Protein-level sensitivity → kaiju
├─ Joint DNA+AA (sensitivity + specificity) → metabuli
└─ Sketch containment / ANI-aware presence → sylph
Methods caveats
- Do not average Metabuli with MetaPhlAn/Kraken without labeling method.
- Pin Metabuli + DB release; report paired-end mode and filters.
- Species labels for novel lineages remain reference-dependent.
Related skills
kraken2 · bracken · kaiju · metaphlan · sylph · motus ·
tool-selection · metagenomics-workflow
