pku-embl/metagenomics-skills/antismash

antiSMASH

Upstream: antismash/antismash

Citation

Blin, K. et al. antiSMASH 7.0: new and improved predictions for detection, regulation, chemical structures and visualisation. Nucleic Acids Res. 51, W46–W50 (2023). https://doi.org/10.1093/nar/gkad344

Analytical thinking

antiSMASH finds BGC regions (PKS, NRPS, RiPP, terpene, …). It does not prove a compound is produced. Fragmented MAGs truncate clusters — prefer near-complete bins (checkm2 / mag-qc) and note contig breaks inside regions.

Run on per-genome FASTA (or GenBank from bakta/prokka), not mixed community contigs, unless you explicitly want contig-level screens and will not assign clusters to species without binning.

Record antiSMASH version and database download date. For comparative BGC families across MAGs, follow with BiG-SCAPE (bigscape).

Prerequisites

How to run

Install databases in a pinned environment before the study. This example is for one bacterial MAG FASTA; use a separate output per genome.

set -Eeuo pipefail
id="${1:?usage: run-antismash.sh GENOME_ID GENOME_FASTA OUTPUT_DIR}"
genome="${2:?usage: run-antismash.sh GENOME_ID GENOME_FASTA OUTPUT_DIR}"
out="${3:?usage: run-antismash.sh GENOME_ID GENOME_FASTA OUTPUT_DIR}"
threads="${THREADS:-16}"
taxon="${ANTISMASH_TAXON:-bacteria}"
[[ "$id" =~ ^[A-Za-z0-9][A-Za-z0-9._-]*$ ]] || exit 2
case "$taxon" in bacteria|fungi) ;; *) exit 2 ;; esac
test -s "$genome" || exit 2
log_dir="${out}.logs"
test ! -e "$out" && test ! -e "$log_dir" || {
  echo "use new output and log paths" >&2; exit 2;
}
mkdir -p "$(dirname "$out")" "$log_dir"
 
antismash "$genome" --output-dir "$out" --html-title "$id" \
  --cpus "$threads" --genefinding-tool prodigal --taxon "$taxon" \
  >"$log_dir/antismash.stdout.log" 2>"$log_dir/antismash.stderr.log"
test -s "$out/index.html"
find "$out" -maxdepth 1 -type f -name '*.json' -print -quit | grep -q .

For Bakta/Prokka GenBank input, omit or verify gene-finding options so existing annotations are handled as intended. Do not mix outputs from different module or database configurations in one comparative set.

Key parameters

Outputs to archive

Decision tree

BGC mining goal?
├─ Single HQ MAG → bakta/prokka → antismash → (optional) bigscape
├─ Many MAGs catalogue → filter by checkm2 first, then batch antismash
└─ Need AMP smORFs not BGCs → macrel (different claim)

Troubleshooting

Validation, recovery, and provenance

Require HTML, JSON, annotated GenBank and region files, completed logs, unique record IDs, and an input-to-region audit including zero-region genomes. Inspect cluster boundaries, contig-edge truncation, product/class confidence, gene finding, known-cluster similarity, and MAG quality. Cross-genome family analysis must use compatible antiSMASH/module/database versions.

Use a fresh directory after changing genome, database, modules, or release. Record antiSMASH and dependency versions, database/MIBiG snapshot and checksums, input/annotation checksums, taxon/gene-finding/module flags, commands, logs, region manifest, and output checksums. BGC predictions indicate biosynthetic potential, not expression or compound production.

Related skills

microbial-mining · bakta · prokka · bigscape · dram · checkm2 · mag-qc