Spectral Match Interpretation

spectral-match-interpretation

Summary

Interprets ranked spectral matches returned from domain-specific MASST tools by aggregating results across microbe, plant, tissue, microbiome, and food domains, then visualizing and consolidating hits with associated metadata to enable cross-domain metabolite identification.

When to use

You have obtained search results from one or more domain-specific MASST web applications (microbeMASST, plantMASST, tissueMASST, microbiomeMASST, foodMASST) for a query mass spectrum and need to consolidate, rank, and visualize those matches to infer the identity and biological source of the metabolite.

When NOT to use

• Input is a single query spectrum without prior MASST search results — run domain-MASST search first
• Goal is to retrieve new spectral matches rather than interpret existing results — use domain-MASST web applications or Fast Search API directly
• You need Level 1 (confirmed structure) or Level 3 (bioactive compound) annotation that metadataMASST cannot provide; metadataMASST generates Level 2 (probable structure) consensus

Inputs

• ranked spectral match lists with cosine similarity scores from microbeMASST
• ranked spectral match lists with cosine similarity scores from plantMASST
• ranked spectral match lists with cosine similarity scores from tissueMASST
• ranked spectral match lists with cosine similarity scores from microbiomeMASST
• ranked spectral match lists with cosine similarity scores from foodMASST
• library spectrum metadata (organism, taxonomy, collision energy, ionization mode)
• query spectrum metadata (precursor m/z, ionization mode, collision energy)

Outputs

• consolidated match table (CSV or TSV with merged hits, similarity scores, metadata)
• interactive HTML tree visualizations per domain (_microbe.html, _plant.html, etc.)
• JSON tree files for programmatic downstream analysis
• _matches.tsv file containing all matching spectra with scores
• _library.tsv file containing GNPS library matches (Level 2 annotation)
• _datasets.tsv file with count of unique samples per dataset
• domain-specific count tables (_count_domain.tsv)
• aggregated visualization showing cross-domain hit distribution and score rankings

How to apply

Load aggregated search outputs (ranked match lists with cosine similarity scores and library metadata) from domain-specific MASSTs into metadataMASST. Process and consolidate the results by merging hits, similarity scores, and taxonomy/organism metadata across domains. Generate interactive visualizations (e.g., hit distribution plots, score rankings, domain overlap) to identify high-confidence matches and cross-domain consensus. Export the consolidated results with embedded visualizations in structured format (HTML trees, JSON, or TSV tables). Use cosine similarity thresholds and minimum peak-matching criteria (configurable during batch search, typically adjusted via parameters in jobs.py) to filter low-confidence matches, then compare the ranked scores and metadata consistency across domains to judge annotation confidence.

Related tools

• metadataMASST (web application that processes and consolidates aggregated domain-MASST search outputs and generates cross-domain visualizations) — https://masst.gnps2.org/metadatamasst/
• microbeMASST (domain-specific MASST for microbial metabolomics that returns ranked spectral matches against microbial reference database) — https://masst.gnps2.org/microbemasst/
• plantMASST (domain-specific MASST for plant metabolomics that returns ranked spectral matches against plant reference database) — https://masst.gnps2.org/plantmasst/
• tissueMASST (domain-specific MASST for tissue metabolomics that returns ranked spectral matches against tissue reference database) — https://masst.gnps2.org/tissuemasst/
• microbiomeMASST (domain-specific MASST for microbiome metabolomics that returns ranked spectral matches against microbiome reference database) — https://masst.gnps2.org/microbiomemasst/
• foodMASST (domain-specific MASST for food metabolomics that returns ranked spectral matches against food reference database) — https://masst.gnps2.org/foodmasst2/
• GNPS_MASST (code repository containing implementation of all domain-specific standalone web applications and batch search infrastructure) — https://github.com/mwang87/GNPS_MASST
• jobs.py (batch execution script for searching multiple spectra against all domainMASSTs via Fast Search API and generating multi-domain outputs) — https://github.com/robinschmid/microbe_masst
• Fast Search API (backend API enabling batch search of multiple MS/MS spectra against indexed data in GNPS/MassIVE, Metabolomics Workbench, MetaboLights, and NORMAN) — https://fasst.gnps2.org/fastsearch/

Examples

python jobs.py  # After configuring input_file, output_prefix, and parameters (min_cosine_score, mz_tolerance, min_matching_peaks) in jobs.py, execute batch search against all domainMASSTs; re-run until skip_existing=True produces no new files

Evaluation signals

• Consolidated match table contains all input domain results merged with no duplicate rows; verify by counting rows and checking for duplicate (spectrum_id, match_score) pairs
• Cosine similarity scores in output fall within expected range (0–1) and are consistent with input domain-MASST scores; spot-check top 10 matches
• HTML tree visualizations render without errors and display hierarchical hit distribution with domain labels and scores visible
• GNPS library matches achieve Level 2 annotation status (probable structure) with recorded cosine similarity ≥ minimum threshold set in batch parameters
• Cross-domain consensus: metabolites appearing in ≥2 domains with consistent top-ranked match should indicate high-confidence annotation; check for agreement in library ID and organism taxonomy

Limitations

• metadataMASST requires pre-computed aggregated outputs from individual domain MASSTs; it does not perform the initial spectral search itself
• Batch search via jobs.py may require multiple re-runs to catch all Fast Search API failures; the README notes 'sequent re-runs should catch all the possible matches'
• Python 3.10 is required; compatibility with other Python versions is not documented
• Visualization quality and cross-domain consensus depend on the completeness and curation of underlying domain-specific reference databases; some domains (e.g., microbiomeMASST) have fewer indexed species than others (microbeMASST: 1379 species; plantMASST: 3712 species)
• Results exclude matches outside currently indexed datasets (GNPS, MassIVE, Metabolomics Workbench, MetaboLights, NORMAN); newer or proprietary datasets are not covered

Evidence

• [intro] metadataMASST takes aggregated search outputs from domain-specific MASSTs (microbeMASST, plantMASST, tissueMASST, microbiomeMASST, and foodMASST) and generates visualized aggregated results: "metadataMASST takes aggregated search outputs from domain-specific MASSTs (microbeMASST, plantMASST, tissueMASST, microbiomeMASST, and foodMASST) and generates visualized aggregated results as its"
• [other] workflow for metadataMASST result interpretation: "Load aggregated domain-MASST search outputs (from one or more domain-specific MASST tools: microbeMASST, plantMASST, tissueMASST, microbiomeMASST, foodMASST) as input. 2. Process and consolidate"
• [readme] batch search output types: "A series of interactive HTML trees files will be generated for each domain-specific MASST ending with _domain.html (e.g., _microbe.html); A series of JSON files for the different trees will be"
• [readme] library annotation level and metadata: "A _library.tsv file will be generated. This contains a list of spectra from the GNPS libraries found to match your spectrum of interest. This enables a Level 2 annotation according the Metabolomics"
• [readme] requirement for multiple batch search runs: "Make sure to run jobs.py a couple of times, until no new output is generated by having the option: skip_existing=True. Due to the Fast Search API some of the entries will fail."
• [readme] Python version requirement: "Please make user to use Python 3.10"