Skip to content

SemiBin2 Specifications

Process Details

  • Name: SemiBin2
  • Process UUID: f931uhzokx2ir4nqhpu43yzynpdjod
  • Process Group: Binning

Overview

SemiBin2 is a binning tool designed for generating metagenome-assembled genomes (MAGs) from short-read sequencing data. This process performs single-sample binning using contig sequences and alignment information to group related contigs into genomic bins representing individual microbial genomes. The tool is specifically optimized for various environmental and host-associated microbiome samples.

This process is implemented in Bash, which invokes the SemiBin2 Python-based tool for metagenomic binning analysis.

Key Functionality

  • Contig Binning: Groups assembled contigs into bins representing individual microbial genomes based on sequence composition and coverage patterns
  • Environment-Specific Models: Utilizes pre-trained models optimized for different sample types including human gut, oral, mouse gut, and global environments
  • Quality Filtering: Applies minimum contig length thresholds to ensure high-quality bin generation
  • Automated Processing: Performs single-sample easy binning with automated parameter optimization

Input/Output Specification

Inputs

Required Inputs

  • Input Directory
    • Description: Directory containing assembled contigs (FASTA format) and corresponding sorted BAM alignment files
    • Format: Directory

Outputs

  • Log File

    • Description: Detailed log of the SemiBin2 binning process including parameters used and bin statistics
    • Format: Log file

  • Output Directory

    • Description: Directory containing the generated genomic bins as individual FASTA files, numbered sequentially (bin.0.fa, bin.1.fa, etc.)
    • Format: Directory

Parameters & Settings

These parameters can be adjusted in the Foundry UI when running this process.

  • Minimum Contig Length

    • Description: Write the minimum expected contig length for SemiBin2 to work on. The value should be an integer!
    • Default value: 2500

  • Data Type

    • Description: Choose what the sequencing data is based on. For instance, choose human_gut for human fecal samples, or mouse_gut for mouse fecal samples, build_environment should be the model for the environmental data other than water, wastewater and soil.
    • Available options: human_gut (default), human_oral, mouse_gut, global

References & Resources

  • Tool Documentation: Contact the team for details on SemiBin2 implementation
  • Related Papers: Pan, S., Zhu, C., Zhao, XM. et al. A deep siamese neural network improves metagenome-assembled genomes in microbiome datasets across different environments. Nat Commun 13, 2326 (2022). https://doi.org/10.1038/s41467-022-29843-y