An open-source MS/MS spectral library for fungal natural products

MycoMSBase brings together high-resolution tandem mass spectra alongside ion mobility data and compound metadata — biosynthetic class, fungal producer, and literature references — to support the dereplication and discovery of fungal secondary metabolites in metabolomics studies.

Work in progress — MycoMSBase is actively being developed. Features and data coverage are growing. Feedback, suggestions, and contributions are very welcome (see Contributing).

---

Contents

• Features
• Architecture
• Prerequisites
• Quick start
• Configuration reference
• Data format
• Development
• Contributing
• Citing

---

Features

Spectral library search

MycoMSBase supports three complementary search modes:

• Spectral similarity search — powered by matchms, a Python library for MS/MS spectrum processing and cosine similarity scoring. Query with a peak list and a similarity threshold to retrieve the closest library matches.
• Library matching via SpecReboot — spectra in MycoMSBase are compatible with SpecReboot, a tool for large-scale spectral library matching and dereplication in fungal metabolomics workflows.
• Fragment peak search — search for spectra containing specific fragment ions within a configurable mass tolerance.
• Substructure search — filter by chemical substructure using SMILES queries, powered by the Bingo extension for PostgreSQL.

Ion mobility data

MycoMSBase is one of the few fungal MS/MS libraries to include collisional cross-section (CCS) values from ion mobility measurements. CCS values are stored per spectrum and are searchable, enabling multi-dimensional dereplication that combines retention time, exact mass, MS/MS fragmentation, and ion shape.

Taxonomy integration

Fungal producer metadata is enriched with full NCBI taxonomy — genus, family, order, class, and phylum — fetched automatically via the NCBI Entrez API. The web interface provides:

• Interactive taxonomy tree — a cladogram showing all fungal producers with biosynthetic class distributions as pie charts at each node, with rank-level zoom (Kingdom / Family / Genus).
• Taxonomy filters — filter the library by genus or species directly in the search panel.
• NCBI Taxonomy links — each record links out to the NCBI Taxonomy Browser entry for its producer species.

Compound metadata

Every record is annotated with:

• Biosynthetic class (Polyketide, Terpene, NRPS-like, PKS-NRPS) — drives the charts and taxonomy tree visualisation.
• Fungal producer with full taxonomic lineage.
• Literature reference (DOI) linking to the original publication.

---

Architecture

MycoMSBase is deployed as a set of Docker services orchestrated with Compose:

Browser
  └── nginx (reverse proxy :8080)
        ├── mb3frontend   React/TypeScript web app
        ├── mb3server     Go REST API
        ├── similarity-service   Python spectral search (matchms)
        └── export-service       Java bulk export (MGF/MSP)
              └── postgres   PostgreSQL + Bingo (substructure search)

Service	Technology	Role
postgres	PostgreSQL + Bingo	Stores records; enables substructure search
mb3server	Go	REST API backend
similarity-service	Python / FastAPI / matchms	Cosine spectral similarity search
export-service	Java	Bulk MGF/MSP export
mb3frontend	React / TypeScript	Web interface
nginx	nginx	Routes all services under one port
mb3tool	Go	One-shot database initialisation from a data repo

---

Prerequisites

• Docker ≥ 24 with Compose v2
• Git (for loading data from a repository)
• ~4 GB RAM for the full stack

---

Quick start

1. Clone the repository

git clone https://github.com/ECharria/mycomsbase.git
cd mycomsbase

2. Set up the environment

Copy the annotated template and edit the key variables:

cp compose/env.dist compose/.env

Minimum required changes in compose/.env:

# Where PostgreSQL data will be stored on disk
DB_LOCAL_PATH=./../data/postgres-data

# Git repository containing the MassBank .txt record files
MB_GIT_REPO="https://github.com/<your-org>/<your-data-repo>"
MB_GIT_BRANCH=main

# Local path to the record files (used by the similarity and export services)
MB_DATA_DIRECTORY="./../data/mycomsbase-data"

# Hostname or IP of the server (use localhost for local deployment)
MB3_API_HOST=localhost
MB3_FRONTEND_HOST=localhost

3. Build and launch

cd compose
docker compose build
docker compose up -d

4. Load the spectral library into the database

docker compose run --rm mb3tool

The mb3tool service clones the data repository and imports all records into PostgreSQL.

Once all services are healthy, open http://localhost:8080/MycoMSBase in your browser.

---

Configuration reference

All settings live in compose/.env. The fully annotated template is compose/env.dist.

Variable	Default	Description
DB_USER / DB_PASSWORD / DB_NAME	mycomsbase / mycomsbasepassword / mycomsbase	PostgreSQL credentials
DB_LOCAL_PATH	./../data/postgres-data	Host path for database storage
MB3_API_BASE_URL	/MycoMSBase-api	API base path
MB3_FRONTEND_BASE_URL	/MycoMSBase	Frontend base path
MB_GIT_REPO	—	URL of the MassBank-format data repository
MB_DATA_DIRECTORY	./../data/mycomsbase-data	Local path to record .txt files
COSINE_TOLERANCE	0.05	Fragment mass tolerance in Da for similarity search
SIMILARITY_SERVICE_VERBOSE	false	Verbose logging in similarity service
DISTRIBUTOR_TEXT	—	Institution name shown on the About page

---

Data format

Records follow the MassBank record format. MycoMSBase adds three fields to each record:

PUBLICATION: doi:10.xxxx/xxxxx          ← literature reference
CH$COMPOUND_CLASS: Polyketide           ← biosynthetic class
SP$SCIENTIFIC_NAME: Hypoxylon rickii    ← fungal producer

A reference table of all compounds (mycomsbase_unique_compounds.csv) is included at the repository root with the columns:

inchikey · compound_name · compound_class · fungal_producer · n_spectra · doi · example_accession

---

Development

Rebuild a single service

docker compose build mb3server           # Go backend
docker compose build mb3frontend         # React frontend
docker compose build similarity-service  # Python similarity service
docker compose up -d <service>

API

The REST API is served at http://localhost:8080/MycoMSBase-api/.
The full OpenAPI specification is at config-openapi.yaml.

Similarity service endpoints

Endpoint	Method	Description
/similarity	POST	Cosine similarity search against the library
/export/mgf	POST	Export selected records as MGF
/version	GET	Service version and loaded library size

---

Contributing

MycoMSBase is under active development and we warmly welcome contributions of all kinds — bug reports, feature ideas, or code improvements.

• Open a pull request — for code changes or new records, fork the repo and submit a PR
• Open an issue — for bug reports or feature requests
• Send us your samples — if you have fungal samples or extracts that could expand the library, we would love to hear from you
• Get in touch — for questions, sample contributions, or collaboration: esteban.charriagiron@wur.nl

---

Citing

If you use MycoMSBase, please cite the underlying MassBank3 infrastructure:

Neumann S. et al. MassBank3: the spectral reference library's next generation software product.
DOI: 10.5281/zenodo.16923315

---

License

Distributed under the GPL-3.0 license — see LICENSE for details.