name: dastool description: > DAS Tool (Dereplication, Aggregation and Scoring) integrates multiple binning predictions into one non-redundant MAG set for a single assembly. Use this skill when working with DAS_Tool: preparing contig2bin tables, choosing score thresholds, comparing ensemble binning options, or writing Methods. For suite-style MetaBAT/MaxBin refinement use metawrap. For recommended multi-assembly refined recovery use basalt. For routing, use tool-selection. license: MIT metadata: version: "1.0.1" metagenomics.category: "analysis-tools" metagenomics.tags: "MAG, ensemble-binning, DAS_Tool, refinement" metagenomics.upstream: "https://github.com/cmks/DAS_Tool" metagenomics.stage: "binning" metagenomics.maturity: "reviewed" metagenomics.last-reviewed: "2026-10-24" metagenomics.command-status: "DAS Tool and conversion CLI checked against upstream usage; not executed here"

DAS Tool

Upstream: cmks/DAS_Tool

Citation

Sieber, C. M. K. et al. Recovery of genomes from metagenomes via a dereplication, aggregation and scoring strategy. Nat. Microbiol. 3, 836–843 (2018). https://doi.org/10.1038/s41564-018-0171-1

See also docs/references.md.

Analytical thinking

DAS Tool sits in the ensemble binning family with MetaWRAP and BASALT: run several independent binners (MetaBAT2, MaxBin2, CONCOCT, …) on the same assembly, then select an optimized non-redundant set.

DAS Tool scores candidates with single-copy marker genes and aggregates across binning predictions. It does not invent new contig membership from cross-assembly connectivity or deep learning — that is BASALT’s role.

Choose DAS Tool when you already have multiple contig2bin tables and want a lightweight, classical consensus. Prefer `basalt` as the default ensemble route for related samples / multi-assembly work; use `metawrap` when you want a packaged binning + CheckM-style refinement suite.

Prerequisites

How to run

Use only predictions generated from the same immutable assembly and namespace.

set -Eeuo pipefail
assembly="${1:?usage: run-dastool.sh ASSEMBLY METABAT_BINS MAXBIN_BINS OUTPUT_DIR}"
metabat_bins="${2:?usage: run-dastool.sh ASSEMBLY METABAT_BINS MAXBIN_BINS OUTPUT_DIR}"
maxbin_bins="${3:?usage: run-dastool.sh ASSEMBLY METABAT_BINS MAXBIN_BINS OUTPUT_DIR}"
out="${4:?usage: run-dastool.sh ASSEMBLY METABAT_BINS MAXBIN_BINS OUTPUT_DIR}"
threads="${THREADS:-8}"
test -s "$assembly" && test -d "$metabat_bins" && test -d "$maxbin_bins" || exit 2
test ! -e "$out" || { echo "use a new output directory: $out" >&2; exit 2; }
mkdir -p "$out"
 
Fasta_to_Contigs2Bin.sh -i "$metabat_bins" -e fa \
  >"$out/metabat.contigs2bin.tsv" 2>"$out/metabat.convert.stderr.log"
Fasta_to_Contigs2Bin.sh -i "$maxbin_bins" -e fasta \
  >"$out/maxbin.contigs2bin.tsv" 2>"$out/maxbin.convert.stderr.log"
test -s "$out/metabat.contigs2bin.tsv"
test -s "$out/maxbin.contigs2bin.tsv"
 
DAS_Tool \
  -i "$out/metabat.contigs2bin.tsv,$out/maxbin.contigs2bin.tsv" \
  -l metabat,maxbin -c "$assembly" -o "$out/run" -t "$threads" \
  --write_bins --write_bin_evals --search_engine diamond \
  >"$out/dastool.stdout.log" 2>"$out/dastool.stderr.log"

Key parameters

Output

*_DASTool_summary.tsv
*_DASTool_contigs2bin.tsv
DASTool_bins/          # if --write_bins
*_allBins.eval         # if --write_bin_evals

Decision tree

Ensemble MAG recovery?
├─ Recommended default (related samples / multi-assembly) → basalt
├─ Packaged multi-binner + CheckM refinement suite → metawrap
└─ Classical SCG aggregation of existing contig2bin sets → dastool

Literature notes (Nat. Microbiol. 2018)

Sieber et al. motivate DAS Tool with observations still true operationally:

• No single binner wins on all ecosystems; tools reconstruct different genomes

at different completeness — combining predictions recovers more near-complete genomes than any single method.

• Early work showed manual/curated multi-binner merges help; naive automation can

reduce contamination and completeness — DAS Tool’s SCG scoring aims for an optimized non-redundant set from one assembly.

• Input is contig2bin tables on one assembly, not cross-assembly DL refinement

(contrast basalt).

Practical rule from the paper’s logic: diversify binners (MetaBAT2 + MaxBin2 + CONCOCT/SemiBin2/VAMB/COMEBin), then aggregate — do not run DAS Tool on a single contig2bin table expecting magic.

Validation, recovery, and provenance

Require non-empty summary, final contig2bin map, bin-evaluation table, and FASTA bin directory. Verify every assigned contig exists exactly once in the source assembly, input labels match table order, and duplicate contig assignments within an input set are resolved before running. Reassess exported bins with checkm2 and gunc; SCG score is selection evidence, not a publication-quality metric.

Use --resume only with the exact prior proteins, inputs, database, and release. Otherwise keep the old prefix and start a new one. Archive DAS Tool/dependency versions, SCG database/search-engine version, assembly and table checksums, commands, logs, score parameters, input evaluations, selected-bin lineage, and final checksums.

Related skills

metawrap · basalt · megahit · metaspades · checkm2 · gunc · drep · mag-qc · tool-selection