10. Shotgun Metagenomic Assembly & MAG Reconstruction
De novo assembly of microbial communities from complex environmental or host samples, multi-algorithm binning (MetaBAT2, MaxBin2), CheckM2 quality validation, and GTDB-Tk taxonomic resolution.
| Metric / Requirement | Specification | Target Configuration |
|---|---|---|
| Core Engine / Framework | MEGAHIT + MetaBAT2 + CheckM2 + GTDB-Tk | Nextflow / Snakemake Compatible |
| Benchmark Runtime | ~8 – 14 hours (60M PE reads) | 32 CPU cores / 64 GB RAM |
| Standard Container | quay.io/biocontainers/checkm2:1.0.2--pyh7cba7a3_0 | Docker • Singularity • Conda |
| Input → Deliverable | Metagenome FASTQ → Metagenome-Assembled Genomes (MAGs) FASTA, Taxonomic Tree | |
Step 1: Host DNA Depletion with Bowtie2
Filters human/host-derived sequencing reads to prevent host genomic contamination in microbial assemblies.
bowtie2 -p 16 -x /references/human_hg38_index \
-1 raw_R1.fastq.gz -2 raw_R2.fastq.gz \
--un-conc-gz clean_microbe_R%.fastq.gz -S /dev/null--un-conc-gz writes only genuinely unmapped microbial paired reads to output FASTQs.Step 2: De Novo Metagenome Co-Assembly with MEGAHIT
Builds De Bruijn graph contigs across iterated k-mer values (k=21 to k=141) with minimal memory overhead.
megahit -1 clean_microbe_R1.fastq.gz -2 clean_microbe_R2.fastq.gz \
-t 32 --min-contig-len 1500 -o megahit_out/--min-contig-len 1500 discards fragmented contigs that degrade downstream binning accuracy.Step 3: Contig Coverage Depth Profiling
Calculates contig mean coverage across multiple biological samples for differential abundance binning.
bwa-mem2 index megahit_out/final.contigs.fa
bwa-mem2 mem -t 32 megahit_out/final.contigs.fa clean_microbe_R1.fastq.gz clean_microbe_R2.fastq.gz | \
samtools sort -@ 8 -o mapped.sorted.bam -
jgi_summarize_bam_contig_depths --outputDepth depth.txt mapped.sorted.bamStep 4: Metagenomic Binning with MetaBAT2
Clusters contigs into discrete microbial genomes using tetranucleotide frequencies and coverage depth correlations.
metabat2 -i megahit_out/final.contigs.fa -a depth.txt -o bins/bin -t 32Step 5: Machine Learning MAG Quality Audit with CheckM2
Uses gradient-boosted decision trees to evaluate MAG completeness and contamination levels.
checkm2 predict --threads 16 --input bins/ -x fa --output-directory checkm2_results/Completeness > 90% and Contamination < 5%.Step 6: Phylogenetic Taxonomic Classification with GTDB-Tk
Assigns species and genus nomenclature based on 120 bacterial and 53 archaeal marker gene alignments.
gtdbtk classify_wf --genome_dir bins/ --out_dir gtdbtk_out/ --cpus 16 -x fa🚀 Automated Nextflow (nf-core) Production Execution
Execute this complete workflow across any SLURM, AWS Batch, or Google Cloud HPC cluster using the verified community container pipeline:
nextflow run nf-core/mag -r 3.1.0 \
--input samplesheet.csv \
--outdir ./results_mag/ \
-profile docker