name: abricate description: > ABRicate — mass screening of contigs for AMR/virulence/plasmid markers against bundled databases (ResFinder, CARD, NCBI, VFDB, PlasmidFinder, EcoLI_VF, ARG-ANNOT, MEGARES, …). Use for fast multi-DB isolate or MAG screens and database concordance checks. Route via cge-finders / microbial-mining; official CGE CLI still via resfinder/plasmidfinder when phenotype mapping or PointFinder is required. license: MIT metadata: version: "1.0.1" metagenomics.category: "analysis-tools" metagenomics.tags: "ABRicate, AMR, virulence, PlasmidFinder, ResFinder, CARD, VFDB, screening" metagenomics.upstream: "https://github.com/tseemann/abricate" metagenomics.stage: "mining" metagenomics.maturity: "reviewed-legacy" metagenomics.last-reviewed: "2026-10-24" metagenomics.command-status: "ABRicate screen/summary interface checked against upstream usage; not executed here"

ABRicate

Upstream: tseemann/abricate

Citation

Seemann, T. ABRicate: mass screening of contigs for antimicrobial and virulence genes. https://github.com/tseemann/abricate

Also cite the database papers for each --db used (ResFinder, CARD, PlasmidFinder, VFDB, … — see resfinder, rgi, plasmidfinder, virulencefinder).

See also docs/references.md and cge-finders.

Analytical thinking

ABRicate is a BLASTN mass-screen wrapper over multiple curated gene DBs. Ideal for: (1) quick multi-DB presence/absence, (2) cross-checking ResFinder vs CARD vs NCBI AMRFinder-style sets, (3) plasmid replicon + virulence gene panels on many assemblies.

Does not replace PointFinder chromosomal SNPs, ResFinder phenotype summaries, or CARD protein models (rgi). Thresholds (%ID / coverage) drive sensitivity — pin them in Methods. Metagenome contigs: expect fragmented hits.

How to run

Freeze the installed databases before screening; do not update them in the middle of a cohort.

set -Eeuo pipefail
id="${1:?usage: screen-abricate.sh SAMPLE_ID ASSEMBLY OUTPUT_DIR}"
assembly="${2:?usage: screen-abricate.sh SAMPLE_ID ASSEMBLY OUTPUT_DIR}"
out="${3:?usage: screen-abricate.sh SAMPLE_ID ASSEMBLY OUTPUT_DIR}"
[[ "$id" =~ ^[A-Za-z0-9][A-Za-z0-9._-]*$ ]] || exit 2
test -s "$assembly" || exit 2
test ! -e "$out" || { echo "use a new output directory: $out" >&2; exit 2; }
mkdir -p "$out"
abricate --list >"$out/database-manifest.tsv"
 
IFS=, read -r -a databases <<<"${ABRICATE_DATABASES:-resfinder,plasmidfinder,vfdb}"
reports=()
for db in "${databases[@]}"; do
  [[ "$db" =~ ^[A-Za-z0-9._-]+$ ]] || exit 2
  report="$out/${id}.${db}.tsv"
  abricate --db "$db" "$assembly" >"$report" 2>"$out/${id}.${db}.stderr.log"
  grep -q '^#FILE' "$report"
  reports+=("$report")
done
abricate --summary "${reports[@]}" >"$out/${id}.summary.tsv" \
  2>"$out/${id}.summary.stderr.log"
test -s "$out/${id}.summary.tsv"
test -s "$out/database-manifest.tsv"

Decision tree

Multi-DB gene screen on assemblies?
├─ Fast concordance / batch TSV → abricate
├─ CGE ResFinder + phenotype / PointFinder → resfinder (± pointfinder)
├─ CGE DisinFinder → disinfinder via resfinder
├─ CARD protein models → rgi
├─ Official PlasmidFinder CLI → plasmidfinder
└─ Plasmid ST schemes → pmlst

Validation, recovery, and provenance

Require raw per-database tables with headers, a summary table, and the database manifest. Retain no-hit samples. Inspect identity, coverage, coordinates, fragmentation, overlapping/duplicate hits, database discordance, and whether a hit lies on a plasmid/MGE-supported sequence. Do not infer phenotype, gene expression, or complete plasmids from an ABRicate row.

Use fresh outputs after changing assemblies, databases, or thresholds. Record ABRicate/BLAST versions, every database name/date/sequence count and checksum, assembly checksum, command, logs, raw/summary checksums, and filtering rules.

Related skills

cge-finders · resfinder · plasmidfinder · disinfinder · virulencefinder · staramr · rgi · pmlst · mlst · microbial-mining · tool-selection