Overview
MS2DeepScore uses a Siamese neural network to predict fingerprint-based similarity and generate spectral embeddings. Model versions differ in supported ionization modes.
What can this tool do?
Common uses
- Prioritize potentially related molecules from MS/MS spectra.
- Explore spectral collections through learned embeddings.
Scientific fit
Primary function
Spectral processing and comparison
Data compatibility
Inputs
Mass spectra
Outputs
Statistical results
Input formats
MGF, MSP
Output formats
Other, Predicted similarity matrices and embeddings
Scope and considerations
Designed for
- Prioritize potentially related molecules from MS/MS spectra.
Not designed for
- Interpreting predicted similarity as an identification probability.
Important requirements
- Python, cleaned spectra and a compatible model checkpoint.
Considerations
- Use a model appropriate to the ion mode and chemical domain; report model version and preprocessing.
Access and documentation
Apache-2.0 software; trained models and training datasets may carry additional terms.
Evidence and review
Documentation-based editorial review on 7 September 2026. Scope and limitations summarize the linked official sources; software execution and developer verification were not assessed.
Core publications
Resource papers, direct updates and tool-specific methods. Selection policy.
- Huber F, van der Burg S, van der Hooft JJJ, et al. MS2DeepScore: a novel deep learning similarity measure to compare tandem mass spectra. J Cheminform 13, 84 (2021). doi:10.1186/s13321-021-00558-4
Related tools
Entry history
Changes to this catalogue entry, not the tool's software release history. Acceptance and editorial review are separate.
- 2026-09-08 · Entry update
Audit core publications across existing tools
View change · Entry at this revision - 2026-09-07 · First committed entry
Add 50 documented metabolomics resources
View change · Entry at this revision