A Snakemake pipeline for end-to-end processing of metagenomic shotgun sequencing data (paired-end or single-end). MAGmaker takes raw FASTQ files through quality control, host read removal, assembly, and taxonomic profiling, then — in a second stage — through mapping, binning, bin consolidation, and MAG QC to produce metagenome-assembled genomes (MAGs). An optional viral/plasmid discovery track (geNomad + CheckV) runs alongside the MAG track.
Developed by the Moeller Lab at Cornell University and Princeton University, and maintained by the Sprockett Lab at Wake Forest University School of Medicine.
MAGmaker runs in two stages, each driven by a separate Snakefile. Both stages can be chained automatically using the run_magmaker.sh wrapper.
Stage 1 — Main pipeline (Snakefile)
raw reads → FastQC → fastp / Cutadapt → host removal (Bowtie2) → FastQC → MultiQC
→ MEGAHIT / metaSPAdes → QUAST → MultiQC
→ MetaPhlAn 4 → merged abundance table (optional)
→ Kraken 2 → Bracken → per-sample profiles (optional)
→ sourmash sketch / compare → prototype selection
Stage 2 — Binning pipeline (Snakefile-bin)
contigs + prototype reads → minimap2 / Bowtie2 → sorted BAMs (differential coverage)
BAMs + contigs → MetaBAT2 + MaxBin2 + CONCOCT + SemiBin2
→ Binette (default) or DAS_Tool → one MAG set per sample
MAGs → CheckM2 + GUNC + GTDB-Tk + CMSeq → mag_summary.tsv → renamed_mags/
Viral / plasmid track (opt-in, `virus_all` target):
contigs → geNomad → CheckV → virus_summary.tsv
A generate_binning_config rule bridges the two stages: it reads
selected_prototypes.yaml, picks params.prototypes.n representative samples,
and writes auto_binning.txt. Every sample's assembly is binned; the
prototype samples supply the reads that form the differential-coverage basis.
# Clone and install
git clone https://github.com/SprockettLab/MAGmaker.git
cd MAGmaker
mamba env create -n snakemake -f resources/env/snakemake.yaml
conda activate snakemake
# Edit resources/config/config.yaml and metadata.txt
# Run Stage 1 + Stage 2 through MAG renaming — passes all args through to snakemake
./run_magmaker.sh --cores 8 --use-conda
# On an HPC cluster with a Snakemake SLURM profile
./run_magmaker.sh --profile resources/profiles/your_clusterrun_magmaker.sh runs Stage 1 and Stage 2 through MAG renaming. Read-level
profiling (MetaPhlAn, Kraken2) is toggled by the biobakery / profilers
config lists or per-run flags, and the viral track is opt-in — see
Running the pipeline.
WFUSM users on DEMON: See Running on DEMON for a self-contained setup guide with all databases and environments pre-configured.
See the documentation below for details on each step.
| Page | Contents |
|---|---|
| Pipeline overview | What each stage does and why, tool choices, the two-Snakefile structure |
| Installation | Requirements, conda setup, test data |
| Configuration | config.yaml, metadata.txt, binning.txt |
| Database setup | CheckM2, GUNC, GTDB-Tk, MetaPhlAn, Kraken2, geNomad, CheckV, host genome |
| Running the pipeline | Local, HPC/SLURM, run_magmaker.sh wrapper, per-run flags, stage-by-stage |
| Output | Directory layout, MAG summary table, renaming workflow, GTDB-Tk stages |
| Viral / plasmid track | geNomad + CheckV, virus_summary.tsv |
| Running on DEMON (WFUSM) | DEMON cluster setup, pre-configured databases and environments |
If you use MAGmaker in your research, please cite:
Sanders JG, Sprockett DD, Li Y, Mjungu D, Lonsdorf EV, Ndjango JN, Georgiev AV, Hart JA, Sanz CM, Morgan DB, Peeters M, Hahn BH, Moeller AH. Widespread extinctions of co-diversified primate gut bacterial symbionts from humans. Nat Microbiol. 2023 Jun;8(6):1039-1050. doi: 10.1038/s41564-023-01388-w. PMID: 37169918.