Required
Precursor m/z & Ion Mode
This defines the precursor ion mass-to-charge ratio (m/z).
This defines polarity and ensures correct model selection.
Example: PRECURSOR_MZ=517.22098
IONMODE=positiveIONMODE=negative
DM
DeepMASS2 Documentation
Input Format Reference
Data Input Specifications for DeepMASS2
To ensure DeepMASS2 accurately identifies metabolites and correctly names output files, your input data should include the required metadata tags. Examples below use .mgf format.
Batch Annotation
For large-scale annotation, use the local version.
Recommended local repository
https://github.com/hcji/DeepMASS2_GUI
Metadata Tags
These tags help DeepMASS2 parse spectra correctly, select the proper polarity-specific model, and assign meaningful output filenames.
Keep the input simple: each spectrum should include precursor m/z, ion mode, and the corresponding peak list.
Recommended
Automatic Naming & Adduct
Controls exported result filename and ion adduct. Defaults: [M+H]+ for positive, [M-H]- for negative.
DeepMASS2 uses this tag to define the output filename for the exported semantic similarity analysis.
If provided, the results will be saved as <COMPOUND_NAME>.csv.
Example: ADDUCT=[M-H]-
Example: COMPOUND_NAME=challenge_0
Optional
Molecular Formula
Adding the molecular formula helps the semantic similarity engine constrain potential chemical space, significantly improving the ranking accuracy of structurally related metabolites.
Example: FORMULA=C25H38O9
BEGIN IONS
PRECURSOR_MZ=194.04588176800002
COMPOUND_NAME=challenge_255
FORMULA=C9H9NO4
ADDUCT=[M-H]-
IONMODE=negative
132.0454 0.006008828855732822
134.0245 0.0022295306804225633
149.0483 0.00297960480724686
150.056 1.0
194.0459 0.36915136378467533
END IONS
DeepMASS2 Results Explanation
Formula Finder
Annotates possible chemical formulas for unknown spectra.
- Formula: Chemical formula, e.g.,
C9H9NO4 - Mass: Theoretical mass
- Error (mDa): Deviation between experimental and theoretical mass in millidaltons
Structure Finder
Predicts potential molecular structures with SMILES and structure scores.
- Title: Compound name
- MolecularFormula: Molecular formula
- CanonicalSMILES: Standardized SMILES
- InChIKey: Unique structure identifier
- Database IDs: Reference database IDs
- Formula Score, Structure Score, Consensus Score: Ranked from high to low, indicating likelihood
Reference Spectrums
- Name: Reference spectrum name
- Adduct: Ion form
- SMILES: Corresponding SMILES
- Parent Mass: Precursor ion mass
- Database: Source database
- Sorted by similarity, highest first
Spectrum Plot
Shows normalized intensity vs m/z, including unknown spectrum peaks and reference spectrum peaks for visual comparison.
