Binette is a fast and accurate binning refinement tool designed to construct high-quality MAGs from the output of multiple binning tools.

From the input bin sets, Binette constructs new hybrid bins. A bin is a set of contigs. When at least two bins overlap (i.e., share at least one contig), Binette applies fundamental set operations to generate new bins:

• Intersection Bin: Contains contigs that are shared by overlapping bins.
• Difference Bin: Includes contigs that are unique to one bin and not found in others.
• Union Bin: Encompasses all contigs from the overlapping bins.

Binette then evaluates bin quality using CheckM2, ensuring the selection of the best possible bins.

Binette is inspired by the metaWRAP bin-refinement tool but effectively addresses its limitations. Key improvements include:

• Enhanced Speed
  Binette significantly accelerates the refinement process by running CheckM2's initial steps (e.g., Prodigal and Diamond) only once for all contigs. These intermediate results are then reused for bin quality assessment, eliminating redundant computations.

• No Limit on Input Bin Sets
  Unlike metaWRAP, Binette supports any number of input bin sets, allowing seamless processing of multiple binning outputs.

Binette can be easily installed with conda

conda create -c bioconda -c defaults -c conda-forge -n binette binette
conda activate binette

Binette should be able to run :

binette -h

Clone this repository:

git clone https://github.com/genotoul-bioinfo/Binette
cd Binette

Then create a Conda environment using the binette.yaml file:

conda env create -n binette -f binette.yaml
conda activate binette 

Finally install Binette with pip

pip install .

Binette should be able to run :

binette -h

Before using Binette, it is necessary to download the CheckM2 database:

checkm2 database --download --path <checkm2/database/>

Make sure to replace <checkm2/database/> with the desired path where you want to store the CheckM2 database.

Binette supports two input formats for bin sets:

1. Contig2bin Tables: You can provide bin sets using contig2bin tables, which establish the relationship between each contig and its corresponding bin. In this format, you need to specify the --contig2bin_tables argument.

For example, consider the following two contig2bin_tables:

• bin_set1.tsv:

  contig_1   binA
  contig_8   binA
  contig_15  binB
  contig_9   binC

• bin_set2.tsv:

  contig_1   bin.0
  contig_8   bin.0
  contig_15  bin.1
  contig_9   bin.2
  contig_10  bin.0

  The binette command to process this input would be:

  binette --contig2bin_tables bin_set1.tsv bin_set2.tsv --contigs assembly.fasta

2. Bin Directories: Alternatively, you can use bin directories, where each bin is represented by a separate FASTA file. For this format, you need to provide the --bin_dirs argument. Here's an example of two bin directories:

   bin_set1/
   ├── binA.fa: contains sequences of contig_1, contig_8
   ├── binB.fa: contains sequences of contig_15
   └── binC.fa: contains sequences of contig_9
   

   bin_set2/
   ├── binA.fa: contains sequences of contig_1, contig_8, contig_10
   ├── binB.fa: contains sequences of contig_15
   └── binC.fa: contains sequences of contig_9
   

   The binette command to process this input would be:

   binette --bin_dirs bin_set1 bin_set2 --contigs assembly.fasta

In both formats, the --contigs argument should specify a FASTA file containing all the contigs found in the bins. Typically, this file would be the assembly FASTA file used to generate the bins. In these examples the assembly.fasta file should contain at least the five contigs mentioned in the contig2bin_tables files or in the bin fasta files: contig_1, contig_8, contig_15, contig_9, and contig_10.

Binette results are stored in the results directory. You can specify a different directory using the --outdir option.

In this directory you will find:

• final_bins_quality_reports.tsv: This is a TSV (tab-separated values) file containing quality information about the final selected bins.
• final_bins/: This directory stores all the selected bins in fasta format. Can be skipped with --no-write-fasta-bins.
• final_contig_to_bin.tsv: A headerless TSV file mapping each contig to its assigned bin. This format is much lighter than the fasta output to describe the final Binette bins.
• input_bins_quality_reports/: A directory storing quality reports for the input bin sets, with files following the same structure as final_bins_quality_reports.tsv.
• temporary_files/: This directory contains intermediate files. If you choose to use the --resume option, Binette will utilize files in this directory to prevent the recomputation of time-consuming steps.

The final_bins_quality_reports.tsv file contains the following columns:

To report bugs, request new features, or seek help and support, please open an issue.

This tool is released as open source software under the terms of the GNU General Public Licence.

Binette is a scientific software tool with a published paper in the Journal of Open Source Software. If you use Binette in academic research, please cite:

Binette: a fast and accurate bin refinement tool to construct high-quality Metagenome Assembled Genomes.
Mainguy et al., (2024).
Journal of Open Source Software, 9(102), 6782.
doi: 10.21105/joss.06782

Binette extensively uses CheckM2. If your work relies on Binette, consider citing:

CheckM2: a rapid, scalable, and accurate tool for assessing microbial genome quality using machine learning.
Chklovski, Alex, et al. (2023).
Nature Methods, 20(8), 1203-1212.
doi: 10.1038/s41592-023-01940-w