Skip to content

MetaBAT2

MetaBAT2 is a metagenome binning tool that groups assembled contigs into bins representing individual genomes (metagenome-assembled genomes, MAGs). It uses tetranucleotide frequency (TNF) and coverage depth across samples to cluster contigs.

Keywords: metagenomics, binning, MAG, assembly, microbial ecology

Available modules

---------------------------- /software/modulefiles -----------------------------
metabat2/2.18(default)

To see what a module sets up: 

module show metabat2/2.18

For built-in documentation: 

module help metabat2/2.18

Executables

Loading the module adds the following to your PATH:

Executable Description
metabat2 Main binning tool
metabat1 Legacy ensemble binning (MetaBAT 1)
jgi_summarize_bam_contig_depths Computes per-contig coverage depth from sorted BAM files
runMetaBat.sh Convenience wrapper: runs depth summarization + metabat2 in one step

Quick start

MetaBAT2 requires two inputs:

  1. A contig assembly (FASTA)
  2. Coverage depth information derived from sorted BAM file(s)

Step 1 — Generate depth file

module load metabat2/2.18
module load samtools/1.22.1

# Sort and index your BAM file if not already done
samtools sort -o reads_sorted.bam reads.bam
samtools index reads_sorted.bam

# Compute per-contig depth
jgi_summarize_bam_contig_depths \
    --outputDepth depth.txt \
    reads_sorted.bam

For multiple samples (recommended — improves binning accuracy):

jgi_summarize_bam_contig_depths \
    --outputDepth depth.txt \
    sample1_sorted.bam sample2_sorted.bam sample3_sorted.bam

Step 2 — Run MetaBAT2

mkdir -p bins/
metabat2 \
    -i assembly.fa \
    -a depth.txt \
    -o bins/bin \
    --minContig 2500 \
    --numThreads 4

Output: bins/bin.1.fa, bins/bin.2.fa, … — one FASTA per bin.

Example job script (Midway3)

#!/bin/bash
#SBATCH --job-name=metabat2
#SBATCH --account=pi-[cnetid]
#SBATCH --partition=caslake
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=8
#SBATCH --mem=32G
#SBATCH --time=02:00:00

module load metabat2/2.18
module load samtools/1.22.1

cd /scratch/$USER/my_metagenome

jgi_summarize_bam_contig_depths \
    --outputDepth depth.txt \
    sample1_sorted.bam sample2_sorted.bam

mkdir -p bins/
metabat2 \
    -i assembly.fa \
    -a depth.txt \
    -o bins/bin \
    --minContig 2500 \
    --numThreads 8

Notes

  • Multiple samples: Using BAM files from multiple samples substantially improves bin quality. A single-sample run may produce few or no bins.
  • Minimum contig length: The default --minContig 2500 skips contigs shorter than 2500 bp. Lowering this recovers more sequence but reduces bin purity.
  • runMetaBat.sh wrapper: Combines both steps in one command — useful for simple single-sample runs: runMetaBat.sh assembly.fa reads_sorted.bam
  • Memory: Large assemblies (>500 Mbp) may require 64 GB or more. Adjust --mem accordingly.