Description

MS-Net is a KNIME workflow for confident metabolite annotation in LC-MS/MS untargeted metabolomics. It combines mass spectral similarity networks, molecular structure similarity, and taxonomic knowledge to prioritize structural annotations within large candidate spaces.

Key Features

• Network-based annotation propagation from high-confidence seeds (spectral library matches, authentic standards)
• Multi-level confidence scoring system (Levels 1-5) with taxonomic enrichment
• Composite Link Score integrating Tanimoto similarities (full-molecule + scaffold), MS/MS cosine similarity, and in silico confidence
• Taxonomic filtering via Coconut 2.0 natural products database
• Rich metadata output with ClassyFire and NPClassifier chemical ontologies and Database identifiers (CID, Kegg...)
• Compatible with LC-MS processing tools: MZmine, MS-DIAL, Sirius-CSI, MS-Finder

Requirements

Software

• KNIME Analytics Platform ≥ 5.2 (https://www.knime.com)

Input Data Sources

• MZmine or MS-DIAL: Feature detection, MS/MS spectral library matching, molecular networking
• Sirius ≥ 5.8 or MS-Finder: In silico structure prediction with CSI:FingerID and MSNovelist

Quick Start

1. Read the tutorial
2. Install KNIME 5.2+ and required extensions 
3. Import workflow into KNIME (File → Import KNIME Workflow)
   1. For MZmine/MSDial and Sirius-based annotation: MSNet-Sirius.knwf
   2. For MSDial and MSFinder-based annotation: MSNet-MSFinder.knwf
4. Configure input file paths (feature table, MSS network, Sirius annotations)
5. Adjust parameters (taxonomic filters, α weighting, fingerprint type)
6. Execute workflow

Typical runtime: 5-30 minutes, depending on dataset size

Outputs

• Annotated feature table with confidence levels and chemical metadata
• Feature intensity matrix for statistical analysis
• Mass spectral similarity network with annotations
• Tanimoto structural similarity network
• Summary statistics and annotation quality metrics

Citation

If you use MS-Net in your research, please cite:

Pereira Francisco V, Duthen S, Crossay E, et al. (2025)
Pereira Francisco, Duthen, Crossay et al. MS-Net: Multi-Similarity based network annotation for untargeted metabolomics, 27 November 2025, PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-8174529/v1]

Workflow DOI: 10.5281/zenodo.17669288

Documentation

Full documentation, including detailed installation instructions, parameter optimization guides, troubleshooting tips, and example datasets is provided in the accompanying README.md file.

• Contact: guillaume.marti@utoulouse.fr

License

CeCILL-B Free Software License Agreement

Version

v1.0.0 (Initial release)

Compatible with KNIME ≥ 5.2

Keywords: metabolomics, mass spectrometry, annotation, molecular networking, Tanimoto similarity, natural products, KNIME, workflow

Upload Contents:

• MS-Net_workflow.knwf - Main KNIME workflow
• Quick reference.md - Short documentation
• TutoMS-Net: Complete step-by-step guide based on MZ-Mine/Sirius annotation processing
• MZbatch files: Import in MZmine for raw LC-MS data processing