Metadata And Data Brokering Online Tool

Publish your research data FAIR.

Madbot connects the places your data already lives (a cluster, Galaxy, S3, OMERO) to the repositories that publish it. Describe it once, with help to keep to standard metadata, check it against the requirements of the repository you choose (today ENA, Recherche Data Gouv and Zenodo), and submit it in one click. Or simply keep your project’s metadata in madbot as it comes, share it with your collaborators and core facilities, and be ready to submit when the time comes.

Animated example in four steps. 1: you tell madbot where your data is stored, and nothing is copied. 2: you describe it with metadata fields that are checked as you type. 3: madbot translates each field into the format of the repository you chose. 4: madbot fetches the files, checks them and submits them, and the repository returns an identifier. Four use cases: sequencing reads from a cluster to ENA, microscopy images from S3 to Recherche Data Gouv, RNA-seq counts from Galaxy to Zenodo, and an ongoing soil survey with no submission yet, whose metadata stays in madbot, shared with the team and ready when you are.

madbot todayas of 19 novembre 2026

189users
158workspaces
2Kdatasets associated with madbot
6.9Kmetadata associated with the datasets

How it works

Four steps from lab bench to repository

Research data goes through a life cycle, from planning to reuse. Madbot joins it as soon as you collect your data: it links the data you collect, process and analyse where it is stored, helps you describe it as you go, and publishes it to a repository where it is preserved and shared, ready to be reused.

Without it, describing and publishing data usually waits until the end of a project, and is done from memory. With madbot, it takes four steps: open your workspace, link your data, describe it, publish it. Save time: you never enter the same information twice.

The research data life cycle: plan, collect, process, analyse, preserve, share, reuse. Madbot sits in the middle: it links the data you collect, process and analyse, and publishes it so that it is preserved and shared.
Where madbot fits in the data life cycle. Adapted from the RDMkit data life cycle (ELIXIR), CC BY 4.0.

01

Open a workspace for your project

Your project lives in your team’s workspace, a shared space where the colleagues you invite can read and complete its information. Shape it like a directory tree, in a way that makes sense to you: name each level and nest it as deep as you need, following the ISA model, your field’s conventions or simply your lab’s habits. Each level holds its own description, samples, datalinks and metadata.

Project tree in madbot: S. salivarius conjugative elements, with a literature review and methods split into data analysis, data collection and a study design with lab experiments and a strain collection
madbot / Oral streptococci lab

02

Link data where it already lives

Link madbot to your files, whether they are stored on a computing cluster, a NAS storage system, Galaxy, OMERO or an S3 bucket. Madbot doesn’t store your files: it creates “datalinks” to where each dataset is stored, and checks that it is still accessible.

Datalinks table listing eight paired fastq.gz files; the details of B50_R1.fastq.gz give its path on cluster.my-institute.fr and show it is accessible
madbot / S. salivarius conjugative elements

03

Describe once, inherit everywhere

Describe your data with fields from community standards such as Dublin Core, Darwin Core or ENA checklists, or from madbot’s own referential. Every value is checked against the referential as you type, and mapped to the repositories’ formats. Save time: a value set on a parent level flows down to its children, so you don’t type it again for each of your datasets, and you can still override it.

Metadata of the Lab experiments node: an isolation source set on the node itself, then a collection date, geographic location, host scientific name and subject inherited from the Methods node, greyed out with an Inherited value from node Methods tooltip
madbot / S. salivarius conjugative elements / Methods / Study design / Lab experiments

04

Publish in one click

Choose the datalinks to publish and the repository. Madbot fetches the files, maps your metadata to the repository’s format, validates it and submits it. The DOI or accession number comes back into madbot, linked to your data.

Submissions list with ENA, Recherche Data Gouv and Zenodo submissions, all marked Deposited
madbot / Submissions

Across disciplines

Not just genomics

Madbot was born in bioinformatics, but its workflow isn’t tied to DNA. Plugins bring each field its own storage, standards and repositories.

Each use case below works with today’s plugins.

Genomics

Sequencing runs

Paired-end reads described with ENA checklists and submitted as samples and runs.

Data from SSHFS Described with ENA checklists Madbot assembles data and metadata into an ENA submission. Published to ENA

Bioimaging

Microscopy screens

Images stay in OMERO; madbot describes the experiment and publishes it.

Data from OMERO Described with Dublin Core Madbot translates Dublin Core to Recherche Data Gouv’s format, then assembles the deposit. Published to Recherche Data Gouv

Omics analyses

Workflow outputs

Results from a Galaxy history, described once and published with a DOI.

Data from Galaxy Described with madbot referential Madbot translates its own fields to Zenodo’s format, then assembles the deposit. Published to Zenodo

Biodiversity

Occurrence data

Species observations described with Darwin Core terms, the standard of biodiversity data.

Data from S3 Described with Darwin Core Madbot translates Darwin Core to Zenodo’s format, then assembles the deposit. Published to Zenodo

Research software

Code and notebooks

A release from your repository, archived with its authors and license.

Data from GitLab Described with Dublin Core Madbot translates Dublin Core to Zenodo’s format, then assembles the deposit. Published to Zenodo

Your field

Build it with your community

Occurrence data bound for GBIF? Microscopy images for the BioImage Archive? Each new use case is a plugin, and the next ones are best built with the people who need them.

Plugins

Plugs into what you already use

Every connection is a small Python plugin. There are three kinds, and you can contribute by writing a community plugin for your needs.

Describe with

Field sets that extend the metadata catalogue.

Publish to

Repositories madbot can submit to directly.

Preview Still evolving: it works, and your feedback helps us stabilize it.Community Built and maintained by partners outside the core development team.

Metadata

Metadata, finally compatible

Each piece of metadata is a field and its value. Madbot is built on one shared catalogue of fields, drawn from standards across disciplines, and each “Publish to” plugin maps them to its repository’s own format. Values link to the authorities researchers already trust, so a name stays a person and a place stays a place. Every entry is checked as you type: change any field of the example, or paste your own ORCID iD.

ORCID
people
ROR
institutions
DOI
datasets and papers
PubMed
publications
NCBI Taxonomy
organisms
EDAM · OLS
ontology terms
Try it! Every field can be edited
Geographic location (country and/or sea)
enasample
↳ inherited from Methods
Collection date
enasample
✓ valid ISO 8601 date
Scientific name
enasample
✓ NCBI Taxonomy 1304
Dataset contacts
recherche data gouvgeneral
Type an ORCID iD, e.g. 0000-0002-1825-0097

Workspaces

As open as possible, as closed as necessary

Each team gets its own workspace. Invite people such as your collaborators or the core facilities that produce your data, give them a role on each level, and keep restricted what must not be disclosed while the rest stays visible to collaborators.

Role on a level
ownerFull access to the level, can delete it and can manage members
managerFull access to the level, and can manage members
contributorFull access to the level, but cannot delete it
collaboratorCan view the level, but not modify any content

For developers and institutions

Built to fit your infrastructure

Madbot is open source and designed to work with your own tools, not only through its interface.

API

A documented REST API

An OpenAPI 3 API, documented with Swagger, lets scripts and lab information systems (LIMS) work with madbot directly.

Plugins

Extend it in Python

The plugin API and its development guide let you add a connection without touching madbot’s core. Partners already maintain their own.

Security

Credentials in a vault

The credentials madbot uses to reach your storage are kept in HashiCorp Vault.

Quality

Checked on every change

Each change to madbot’s code goes through automated style and quality checks, dependency and security scans, and tests before it is released.

License

Open source, BSD-3-Clause

The code is public on GitLab. Read it, reuse it, deploy it or contribute to it.

Archive

Preserved on Software Heritage

Madbot’s source code is archived by Software Heritage, the universal archive of software source code.

Roadmap 2026

What’s next for madbot

Q1

  • Import metadata when creating a datalink
  • Batch metadata editing
  • Duplicate and move levels
  • Personal access tokens for the API

Q2

  • SSH key authentication for SSHFS
  • Recherche Data Gouv submission plugin
  • Accession numbers and DOIs shown on submissions, data and metadata
  • Suggested metadata values and sample–data associations
  • Easier sample–data associations, with guidance for paired data

Q3

  • Faster lists, project trees and metadata pages
  • Plugin maturity shown in the app (Stable or Preview)
  • A profile with ORCID and ROR identifiers
  • Smoother editing: visible multi-selection, accurate unsaved-changes warnings, fixed rich-text editors
  • Clear error messages when a plugin connection fails

Q4

  • Batch editing everywhere, in one fast table view
  • A faster application start-up
  • Metadata entry: autosave, clearer validation, a single window to add fields
  • Create datalinks and samples in bulk, with clear errors when something fails
  • Typed data associations (paired, assembly…)
  • A clearer interface: guided discovery of madbot’s vocabulary, separate user and workspace settings
  • A documented API path for data-producing platforms

Want the details? Follow the epics on Taiga ↗

Team

Meet the people behind madbot

Developers and project leads at the Institut Français de Bioinformatique (IFB/ELIXIR-FR), in Paris and Strasbourg, working to fit FAIR research data into researchers’ busy schedules.

Thomas Denecker

Project owner
Research Engineer, CNRS

IFB-core, Hub Pasteur · Paris

“This is awesome! Let’s try it!”

Julien Seiler

Project owner
Research Engineer, CNRS

IFB-core, BigEst · Strasbourg

“It’s great but…”

Baptiste Rousseau

Full-stack developer
Research Engineer, CNRS

IFB-core, Hub Pasteur · Paris

“I may have gone down a rabbit hole… but look what I found!”

Elora Vigo

Full-stack developer
Research Engineer, CNRS

IFB-core, BigEst · Strasbourg

“Isn’t this a bit off-center?”

National working group

Shaped with institutes across France

Researchers, data-producing platforms and national infrastructures meet in madbot’s working group. Its members try madbot on their own use cases, tell us what to fix, and take it back to their communities.

  • Institut Français de Bioinformatique ELIXIR France
  • CNRS
  • INRAE
  • Inserm
  • France Génomique
  • France-BioImaging
  • GenomiqueENS
  • ProFI, proteomics
  • IGBMC
  • CIRAD
  • Université Clermont Auvergne
  • Aix-Marseille Université

Link your first dataset in minutes.

Madbot is free and open source, and runs on French public research infrastructure.

One email a month. Unsubscribe anytime.