pku-embl/metagenomics-skills/bracken

Bracken

Upstream: Bracken

Citation

Lu, J., Breitwieser, F. P., Thielen, P. & Salzberg, S. L. Bracken: estimating species abundance in metagenomics data. PeerJ Comput. Sci. 3, e104 (2017). https://doi.org/10.7717/peerj-cs.104

See also docs/references.md.

Analytical thinking

Bracken redistributes Kraken-classified reads to a target taxonomic level using a Bayesian model conditioned on genome lengths / k-mer uniqueness in the same database build. It answers: “given Kraken’s read assignments, what are smoothed abundances at rank L?” — not “is species X truly present?”

Caveats:

• Must use a Bracken index built from the same Kraken2 DB.
• -r read length should match sequencing.
• Rare taxa remain false-positive-prone; MetaPhlAn papers have criticized high

FP rates for some Bracken setups — report confidence/thresholds carefully.

• Never paste Bracken and MetaPhlAn columns into one “species abundance” matrix

without a method factor.

How to run

bracken -d $KRAKEN2_DB -i temp/kraken/${id}.report \
  -o temp/kraken/${id}.bracken -r 150 -l S

Key parameters

Validation, recovery, and provenance

Require a non-empty Bracken table at the requested rank and verify its input report came from the identical Kraken2 database build. Confirm the database contains Bracken distribution files for the exact -r read length and k-mer settings. Reconcile redistributed reads, inspect zero/no-read samples and controls, and retain both original Kraken and Bracken outputs.

Record Bracken/Kraken versions, rank, threshold, read length, database manifest/checksum, commands, and output schema. Rerun all samples after a database or read-length change rather than combining tables. Bracken estimates conditional rank abundance; it does not add independent evidence for taxon presence.

Related skills

kraken2 · metaphlan · metabuli · visualization · tool-selection