metagenomics-workflow

name	metagenomics-workflow
description	End-to-end shotgun metagenomics: quality control, assembly, mapping, binning, taxonomic annotation, functional annotation, microbial mining (BGCs, AMR, MGEs, element cycling, DOM, defense), and research analysis (stats/figures). Use when starting a project or ordering stages. For tool choice use tool-selection. For mining use microbial-mining; for paper stats use research-analysis. For public DBs use public-databases.
license	MIT
category	orchestration
tags	["guidance","pipeline","workflow","QC","assembly","mapping","binning","taxonomy","function","mining","research-analysis"]

Metagenomics Workflow

Organize by scientific stage. Mapping sits between assembly and binning. Microbial mining is an optional discovery track on contigs/MAGs.

Path

study-design + setup (+ public-databases)
        │
        v
1 QC      fastp → kneaddata
        │
        ├─► 5 taxonomy / 6 function (reads) ─► research-analysis
        │
        v
2 Assembly   megahit | metaspades | metaflye | metamdbg
             (± binchicken for targeted coassembly)
        │
        v
3 Mapping    strobealign | bowtie2 | bwa-mem2 | minimap2 → samtools
        │
        ├─► 6 function (genes)
        │
        v
4 Binning    single binners → basalt ★ | metawrap | dastool
             → (± anvio refine | deepurify clean)
             → drep → coverm → checkm2 / gunc / mag-qc
        │
        ├─► 5 taxonomy (genomes): gtdbtk
        ├─► 6 function (genomes): bakta / prokka / eggnog-mapper / rgi
        └─► 7 microbial mining: antismash / genomad→checkv→vcontact / dram / …
        │
        v
   research-analysis (multiqc · microeco · maaslin2 · …) → Methods archive

Common paths

Goal	Route
Composition only	QC → metaphlan / singlem / kraken2 → research-analysis
Genome-resolved catalogue	QC → assembly → mapping → basalt ★ → mag-qc → gtdbtk
Interactive MAG polish	After auto-bins → anvio → re-QC
Environmental unknowns	QC → singlem (± binchicken coassembly)
Viral ecology	contigs → genomad/virsorter2 → checkv → vcontact

Checklist

• Design & DB plan (study-design, public-databases)
• QC + PE order
• Assembly + QUAST glance
• Sorted BAM coverage (multi-coverage when related samples exist)
• Ensemble binning route declared (basalt recommended)
• Optional refine/clean: anvio / deepurify then re-QC
• Taxonomy / function layers not merged unlabeled
• If mining: route via microbial-mining; pin DBs; state hypothesis limits
• Homology search paradigm declared (homology-search) when annotating dark genes
• If trees/evolution: route via phylogenomics (not ANI-as-tree)
• Downstream stats/figures via research-analysis (pin transforms + FDR)
• Provenance: software and database releases

Related skills

tool-selection · research-analysis · microbial-mining · homology-search · phylogenomics · metagenomics-llm · public-databases · study-design · mag-qc · stage skills under skills/