Classifies shotgun metagenomic reads to taxa with Kraken2's minimizer/LCA matching against a chosen reference database, then hands off to Bracken for abundance re-estimation. Covers why the database (not the algorithm) decides what can be detected, the --confidence and --minimum-hit-groups precision levers, unique-minimizer false-positive control, host-read removal, and why raw Kraken2 read counts are not abundances. Use when profiling who-is-there from shotgun reads, choosing a Kraken2 database, setting a confidence threshold, controlling false positives, or feeding reports to Bracken. For marker-gene profiling see metaphlan-profiling; for abundance mechanics see abundance-estimation; for assembly/MAG recovery see genome-assembly/metagenome-assembly.
Classifies shotgun metagenomic reads to taxa with Kraken2's minimizer/LCA matching against a chosen reference database, then hands off to Bracken for abundance re-estimation. Covers why the database (not the algorithm) decides what can be detected, the --confidence and --minimum-hit-groups precision levers, unique-minimizer false-positive control, host-read removal, and why raw Kraken2 read counts are not abundances. Use when profiling who-is-there from shotgun reads, choosing a Kraken2 database, setting a confidence threshold, controlling false positives, or feeding reports to Bracken. For marker-gene profiling see metaphlan-profiling; for abundance mechanics see abundance-estimation; for assembly/MAG recovery see genome-assembly/metagenome-assembly.
Before using code patterns, verify installed versions match. If versions differ:
CLI: kraken2 --version, bracken -h, kraken2 --help to confirm flags and defaults
Python: pip show <package> then help(module.function) to check signatures
If code throws ImportError, AttributeError, or TypeError, introspect the installed
package and adapt the example to match the actual API rather than retrying.
The DATABASE is the version that matters most. A taxon absent from the database is invisible no matter how the binary is configured. Record the Kraken2 database build (Standard / PlusPF / PlusPFP / nt / GTDB, and whether capped to 8/16 GB) and the Bracken read length, which must match both the Kraken2 database and the actual read length. enforcement varied across older 2.x point releases; confirm defaults with on the installed build.
databaseRLENmers.kmer_distrib
--minimum-hit-groups
kraken2 --help
Kraken Classification
"What's in my metagenome?" -> Match each read's k-mers to a reference database by lowest-common-ancestor, then re-estimate abundance with Bracken - because the database, not the algorithm, decides what can be found.
Scope: read-based, assembly-free taxonomic classification of shotgun reads, plus the Bracken handoff. Marker-gene profiling -> metaphlan-profiling. Bracken command mechanics -> abundance-estimation. Genome/MAG recovery -> genome-assembly/metagenome-assembly. Host removal and read QC -> read-qc/contamination-screening, contamination-controls. Amplicon/16S -> the microbiome category.
The Single Most Important Modern Insight -- A Kraken Report Is a Database-Conditioned Similarity Ledger, Not a Sample Inventory
Kraken reports what each read most resembles in THIS database - never what is truly present, and never how much. Every number is hostage to three choices made before the run: the database, the confidence threshold, and the assumption that read count means abundance. Three corollaries each common misuse violates:
Classification is not presence. At --confidence 0 with the LCA rule, one shared k-mer labels a read with its nearest database relative even when the true organism is absent. Absence-from-database becomes a confident wrong species.
Read count is not abundance. Counts scale with genome length and copy number, so the report percentage is a fragment fraction, not a cell fraction. Bracken fixes the wrong-rank problem; it does not fix this.
A taxon at the bottom of the report is a hypothesis, not a finding. Single-region hits, hash collisions, and contaminated references populate the long tail. Unique-minimizer coverage separates a real low-abundance organism from a phantom.
Organize the analysis around defending against these three, not around listing flags. Kraken2 at defaults over-classifies; Kraken2 tuned (right database + confidence + hit-groups + a unique-k-mer floor + host removal) is competitive with any classifier.
Why "Exact K-mer" Is Wrong for Kraken2
Kraken1 (Wood & Salzberg 2014 Genome Biol 15:R46) stored every exact k-mer. Kraken2 (Wood 2019 Genome Biol 20:257) replaced that with three ideas that make it fast and lean but PROBABILISTIC: (a) minimizers collapse each k-mer (default k=35) to the smallest hashed l=31-mer in its window; (b) a spaced seed (s=7 masked positions) tolerates errors at "don't care" positions; (c) a compact hash table stores only high bits of each key. The compact hash can return a wrong or spurious LCA on collision - which is exactly why the precision levers below exist. Calling Kraken2 "exact k-mer" hides where false positives come from.
Tool Taxonomy
Tool
Citation
Mechanism / role
When
Kraken2
Wood 2019 Genome Biol 20:257
minimizer + spaced-seed + compact-hash LCA
fast read classification; database-bound; the default choice
Bracken
Lu 2017 PeerJ Comput Sci 3:e104
Bayesian redistribution of reads stranded at higher ranks
always run after Kraken2 for species/genus estimates
KrakenUniq
Breitwieser 2018 Genome Biol 19:198
HyperLogLog count of unique k-mers per taxon
false-positive control on hits of interest
KMCP
Shen 2023 Bioinformatics 39:btac845
genome-coverage pseudo-mapping
low-depth/clinical/viral where conserved-region FPs hurt
sourmash gather
Pierce 2019 F1000Res 8:1006
FracMinHash containment, min-set-cover
"which genomes are present" with calibrated containment
the FP tail is the enemy; one unique-k-mer filter cuts phantoms
Reads not host-depleted / unQC'd
-> contamination-controls, read-qc/contamination-screening first
host reads swamp the profile; references carry human fragments
Want abundance comparable across studies
state the database + confidence; do not merge with MetaPhlAn percentages
read fraction != cell fraction; different tools = different quantities
Recover genomes / novel taxa / MAGs
-> genome-assembly/metagenome-assembly
classification is assembly-free and database-bound
16S amplicon reads
-> microbiome category
Kraken-on-16S works but amplicon analysis lives there
Basic Classification
# Paired-end, with the precision levers that defaults omit
kraken2 --db "$KRAKEN_DB" \
--paired --gzip-compressed --threads 8 \
--confidence 0.1 \ # raise from default 0 to suppress single-k-mer false positives
--minimum-hit-groups 2 \ # require >=2 distinct hit regions (default 2; raise to 3 for clinical)
--report out.kreport \
--output out.kraken \
R1.fq.gz R2.fq.gz
--paired joins mates with a k-mer-breaking N and classifies the pair as one fragment, raising specificity. The per-read --output (large) can be dropped to /dev/null once the .kreport is what feeds Bracken. --memory-mapping runs without loading the database into RAM (slower; for low-memory hosts).
False-Positive Control: Unique Minimizers
Goal: Separate a real low-abundance organism from a single-region phantom before believing any tail taxon.
Approach: Enable --report-minimizer-data so the report carries distinct-minimizer counts; a taxon with many reads but few distinct minimizers is hitting one conserved region and is a red flag. KrakenUniq's HyperLogLog unique-k-mer count is the heavier-weight version of the same signal.
kraken2 --db "$KRAKEN_DB" --paired --confidence 0.1 \
--report-minimizer-data \ # inserts 2 columns: total + DISTINCT minimizers (shifts later columns)
--report out.kreport --output /dev/null \
R1.fq.gz R2.fq.gz
# A species with high reads but low distinct-minimizers = false positive (one region lit up repeatedly).
Calibrate a unique-k-mer floor against negative controls rather than hard-coding one; the clinical ">=1024 unique k-mers" cutoff is dataset/database-specific folklore, not a constant.
Build a Custom Database
Goal: Build a database whose contents define exactly the detectable universe (and include human for host capture).
Approach: Download taxonomy, add the libraries the question needs (including human), build the minimizer index, then build the matching Bracken distributions at the actual read length.
kraken2-build --download-taxonomy --db custom_db
for lib in bacteria archaea viral human UniVec_Core; do
kraken2-build --download-library "$lib" --db custom_db
done
kraken2-build --build --db custom_db --threads 16 # writes hash.k2d, opts.k2d, taxo.k2d
kraken2-build --clean --db custom_db # drop library/ + taxonomy/ to shrink
bracken-build -d custom_db -t 16 -k 35 -l 150 # -k MUST equal the Kraken2 k (35); -l = read length
kraken2-build --special gtdb builds a GTDB-taxonomy database (curated; the greengenes/silva/rdp special downloads have rotted). --max-db-size randomly downsamples k-mers to fit a cap - this is how the prebuilt 8gb/16gb databases are made, and the reason confidence collapses classification on them.
Hand Off to Bracken
Kraken strands reads at the shared genus when species share k-mers; Bracken redistributes them down using genome-derived priors. It fixes the wrong-rank problem only - never genome-size bias, and never false positives (it can amplify or even invent a species by reassigning an absent organism's reads to its nearest congener). Run FP control first. Command mechanics live in abundance-estimation:
bracken -d "$KRAKEN_DB" -i out.kreport -o out.bracken -w out.bracken.kreport \
-r 150 \ # MUST match a built databaseRLENmers.kmer_distrib AND the actual read length
-l S -t 10 # species level; -t is a redistribution floor (drops taxa with fewer than 10 clade-level reads, strict <), not a confidence
Per-Method Failure Modes
Over-classification at default confidence
Trigger: running --confidence 0 and reporting the species list. Mechanism: one shared k-mer can classify a read; the LCA labels it with its nearest database relative. Symptom: hundreds of low-abundance species, many biologically implausible. Fix:--confidence 0.1-0.4 (database-dependent) plus --minimum-hit-groups >=2; verify the tail with unique minimizers.
Counts read as abundance
Trigger: using the .kreport percentage column as relative abundance. Mechanism: read count is proportional to abundance x genome length x copy number. Symptom: large-genome taxa overstated; downstream diversity/ordination on a non-cell-fraction. Fix: run Bracken for the rank problem; treat even Bracken fraction_total_reads as a read fraction and hand off genome-size/copy-number caveats to abundance-estimation.
Bracken read-length mismatch
Trigger:-r 100 on 150 bp reads, or a database built only for a different length. Mechanism: the redistribution model is fragment-length specific. Symptom: biased abundances with no error (silent) if the .kmer_distrib exists, hard crash if it does not. Fix:-r = actual read length AND a matching databaseRLENmers.kmer_distrib must exist (build it or pick a prebuilt database shipping that length).
Capped database plus high confidence
Trigger: a Standard-8/16 database with confidence cranked up. Mechanism: capped databases are random k-mer subsamples; few reads can clear a high threshold. Symptom: classification collapses toward zero; "my sample is mostly novel." Fix: use a full/large database for high confidence, or lower confidence on a capped database and accept lower precision.
Host reads and contaminated references
Trigger: classifying without host depletion, then trusting human and low-level hits. Mechanism: host reads dominate low-biomass samples, and >2 million GenBank entries carry mislabeled human/vector sequence (Steinegger & Salzberg 2020 Genome Biol 21:115). Symptom: confident Homo sapiens plus a long artifactual tail. Fix: include human in the database and/or host-deplete upstream; scrutinize any taxon co-varying with host load; consider Recentrifuge negative-control subtraction.
Quantitative Thresholds
Threshold
Source
Rationale
--confidence 0.0 default; use 0.2-0.4 on a comprehensive DB
Liu 2024 aBIOTECH 5:465; Lu 2022 Nat Protoc 17:2815
species precision rose from ~0.16 to ~0.76 at CS 0.2; default over-classifies
--minimum-hit-groups 2 (raise to 3 for clinical)
Kraken2 manual
a single lucky minimizer/collision cannot make a call
Bracken -k = 35
Lu 2017 PeerJ Comput Sci 3:e104
must equal the Kraken2 database k-mer length
Bracken -r = actual read length
Bracken docs
redistribution priors are fragment-length specific
Bracken -t 10 default
Bracken docs
redistribution floor; too high deletes real rare taxa, not a confidence
Classification rate 30-70% (environmental)
community
low rate = novel taxa OR host contamination OR wrong database - diagnose which
run bracken-build -l <readlen> or use a prebuilt DB shipping that length
Report parsing columns misaligned
--report-minimizer-data inserted 2 columns
parse 8-column layout when the flag is on
Near-zero classification on a small DB
capped (downsampled) DB + high confidence
larger DB, or lower confidence on the capped DB
Confident human + odd tail
host reads + contaminated references
host-deplete first; treat human/tail as suspect
Bash example silently truncates flags
inline # comment after a \ line continuation
put comments on their own line
References
Wood DE, Lu J, Langmead B. 2019. Improved metagenomic analysis with Kraken 2. Genome Biol 20:257.
Wood DE, Salzberg SL. 2014. Kraken: ultrafast metagenomic sequence classification using exact alignments. Genome Biol 15:R46.
Lu J, Breitwieser FP, Thielen P, Salzberg SL. 2017. Bracken: estimating species abundance in metagenomics data. PeerJ Comput Sci 3:e104.
Breitwieser FP, Baker DN, Salzberg SL. 2018. KrakenUniq: confident and fast metagenomics classification using unique k-mer counts. Genome Biol 19:198.
Lu J, Rincon N, Wood DE, Breitwieser FP, Pockrandt C, Langmead B, Salzberg SL, Steinegger M. 2022. Metagenome analysis using the Kraken software suite. Nat Protoc 17:2815-2839.
Liu Y, Ghaffari MH, Ma T, Tu Y. 2024. Impact of database choice and confidence score on the performance of taxonomic classification using Kraken2. aBIOTECH 5:465-475.
Steinegger M, Salzberg SL. 2020. Terminating contamination: large-scale search identifies more than 2,000,000 contaminated entries in GenBank. Genome Biol 21:115.
Shen W, Xiang H, Huang T, et al. 2023. KMCP: accurate metagenomic profiling of both prokaryotic and viral populations by pseudo-mapping. Bioinformatics 39:btac845.
Related Skills
abundance-estimation - Bracken command mechanics and read-count-to-abundance conversion
