Skip to content

BinSanity

BinSanity is a metagenomics contig binning tool that recovers Metagenome-Assembled Genomes (MAGs) from assembled contigs using coverage-based Affinity Propagation (AP) clustering, optionally refined with GC content and CheckM quality assessment.

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

Available modules

---------------------------- /software/modulefiles -----------------------------
binsanity/0.5.4(default)

To see what a module sets up: 

module show binsanity/0.5.4

For built-in documentation: 

module help binsanity/0.5.4

Executables

Loading the module adds the following to your PATH:

Executable Description
Binsanity Core coverage-based AP clustering
Binsanity-wf Full workflow: coverage + GC content + CheckM quality assessment
Binsanity-lc Low-coverage variant of the core workflow
Binsanity-refine Bin refinement using coverage, GC content, and k-mer composition
Binsanity2-beta Beta workflow variant
Binsanity-profile Per-contig coverage profiling from sorted BAM files (featureCounts)

Quick start

BinSanity requires two inputs:

  1. An assembled contig FASTA file
  2. Per-contig coverage depth (from sorted BAM files)

Step 1 — Compute coverage depth

module load binsanity/0.5.4
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

# Place sorted BAM(s) in a directory, then compute per-contig coverage
mkdir bam_files/
mv reads_sorted.bam bam_files/
Binsanity-profile -i contigs.fa -s bam_files/ -c coverage

-s takes a directory of sorted BAM files (not a single file path). -c is a basename — the tool appends .cov automatically, producing coverage.cov.

For multiple samples (improves binning accuracy):

mkdir bam_files/
mv sample1_sorted.bam sample2_sorted.bam sample3_sorted.bam bam_files/
Binsanity-profile -i contigs.fa -s bam_files/ -c coverage

Step 2 — Bin contigs by coverage

Binsanity -f <contig_dir> -l contigs.fa -c coverage.cov -o bins/

<contig_dir> is the directory containing the contig FASTA file.

Step 3 (optional) — Refine bins with GC content and k-mer composition

Binsanity-refine -f <contig_dir> -l contigs.fa -c coverage.cov -o bins_refined/

-f must point to the directory containing the original contig FASTA (same as Step 2), not the bins/ output directory.

Example job script (Midway3)

This example runs the core binning step. For the full workflow (Binsanity-wf), increase --mem to at least 40 GB and consider --partition=bigmem.

#!/bin/bash
#SBATCH --job-name=binsanity
#SBATCH --account=pi-[cnetid]
#SBATCH --partition=caslake
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=4
#SBATCH --mem=16G
#SBATCH --time=02:00:00

module load binsanity/0.5.4
module load samtools/1.22.1

cd /scratch/$USER/my_metagenome

# Step 1 — coverage profiling (BAMs must be in a directory)
mkdir -p bam_files/
mv sample1_sorted.bam sample2_sorted.bam bam_files/
Binsanity-profile -i contigs.fa -s bam_files/ -c coverage

# Step 2 — bin contigs
Binsanity -f . -l contigs.fa -c coverage.cov -o bins/

Notes

  • Memory — CheckM steps: Binsanity-wf invokes CheckM internally, which requires approximately 40 GB RAM for the full reference tree (or ~16 GB with --reduced_tree). Always run it on a compute partition: --partition=bigmem --mem=48G (full tree) or --partition=caslake --mem=20G --reduced_tree (reduced tree). The core Binsanity and Binsanity-refine commands do not invoke CheckM and run comfortably with 8–16 GB.
  • Memory — large assemblies: BinSanity's Affinity Propagation builds a pairwise distance matrix that scales O(N²) with contig count. Very large assemblies (hundreds of thousands of contigs) may require substantially more RAM. If memory is a concern, MetaBAT2 is a faster alternative with much lower memory requirements.
  • Multiple samples: Using BAM files from multiple samples improves binning quality, particularly for low-abundance organisms.
  • Minimum contig length: Short contigs contribute noise. Filter to ≥2500 bp before binning (e.g., with seqkit or awk).
  • CHECKM_DATA_PATH: The module sets this automatically — no manual configuration needed.