datalab

datalab is a user-friendly, open-source platform that can capture all the experimental data and metadata produced in a scientific lab, targeted (broadly) at materials chemistry but with customisability and extensability in mind. datalab records data and metadata securely and makes it accessible and reusable by both humans and machines via the web UI and API, respectively. datalab can be self-hosted and managed deployments are also available.

You can try the demo deployment at demo.datalab-org.io and read the online documentation at docs.datalab-org.io with release notes and changelog available on GitHub and online.

Features:

• Capture and store sample and device metadata
• Connect and sync raw data directly and from laboratory instruments
• Built-in support for multiple characterisation techniques (XRD, NMR, echem, TEM, TGA, Mass Spec, Raman and more).
• Capture scientific context: store the graph of relationships between research objects.
• Python API for programmatic access to your lab's data enabling custom analysis and automation.
• Join the datalab federation: you can add your datalab to the federation for additional shared features.
• Plugin ecosystem allowing for custom data blocks, AI integration and other instance-specific code.
• Deployment and infrastructure automation via Ansible playbooks.

[!NOTE] You may be looking for the identically named project DataLab for signal processing, which also has plugins, clients and other similar concepts!

Getting started

The full documentation lives at docs.datalab-org.io, split by audience:

• For users — using an instance that someone else runs: recording samples and cells, attaching data, and getting it back out again via the Python API.
• For developers — running datalab locally, writing plugins, and contributing to the core.
• For deployments — standing up and configuring an instance for your group.

Note that datalab is self-hosted and plugin-extensible, so no two instances are alike: the item types, data blocks and metadata fields available to you depend on how your instance has been configured and which plugins it has installed.

We can also provide paid managed deployments via datalab industries ltd.: contact us at hello@datalab.industries.

Design philosophy and architecture

The datalab architecture is shown below:

graph TD
classDef actor fill:#0066CC,fill-opacity:0.3,stroke:#333,stroke-width:2px,color:#000;
classDef clientInterface fill:#00AA44,fill-opacity:0.3,stroke:#333,stroke-width:2px,color:#000;
classDef coreComponent fill:#FF6600,fill-opacity:0.3,stroke:#333,stroke-width:2px,color:#000;
classDef umbrellaLabel fill:#666666,fill-opacity:0.3,stroke:#666,stroke-width:1px,color:#000,rx:5,ry:5,text-align:center;
classDef subgraphStyle fill:#f9f9f9,fill-opacity:0.1,stroke:#ccc,stroke-width:1px;

    subgraph ExternalActors [External actors]
        direction TB
        User[User]
        Machine[Machine]
    end
    class User,Machine actor;
    class ExternalActors subgraphStyle;

    UmbrellaDesc["Raw instrument data,<br>annotations, connections"]
    class UmbrellaDesc umbrellaLabel;

    subgraph ClientInterfaces [Client interfaces]
        direction TB
        BrowserApp[_datalab_<br>Browser app]
        PythonAPI[_datalab_<br>Python API]
    end
    class BrowserApp,PythonAPI clientInterface;
    class ClientInterfaces subgraphStyle;

    subgraph Backend
        direction TB
        RESTAPI[_datalab_<br>REST API]
        MongoDB[MongoDB Database]
        DataLake[Data Lake]
    end
    class RESTAPI,MongoDB,DataLake coreComponent;
    class Backend subgraphStyle;

    User      <-- "User data I/O" --> UmbrellaDesc;
    Machine   <-- "Machine data I/O" --> UmbrellaDesc;

    UmbrellaDesc <-- "_via_ GUI" --> BrowserApp;
    UmbrellaDesc <-- "_via_ scripts" --> PythonAPI;

    BrowserApp  <-- "HTTP (Data exchange)" --> RESTAPI;
    PythonAPI   <-- "API calls (Data exchange)" --> RESTAPI;

    RESTAPI <-- "Annotations, connections" --> MongoDB;
    RESTAPI <-- "Raw and structured characterisation data" --> DataLake;

    linkStyle 0 stroke:#666,stroke-width:3px
    linkStyle 1 stroke:#666,stroke-width:3px
    linkStyle 2 stroke:#666,stroke-width:3px
    linkStyle 3 stroke:#666,stroke-width:3px
    linkStyle 4 stroke:#666,stroke-width:3px
    linkStyle 5 stroke:#666,stroke-width:3px
    linkStyle 6 stroke:#666,stroke-width:3px
    linkStyle 7 stroke:#666,stroke-width:3px

    click PythonAPI "https://github.com/datalab-org/datalab-api" "datalab Python API on GitHub" _blank
    click BrowserApp "https://github.com/datalab-org/datalab/tree/main/webapp" "datalab Browser App on GitHub" _blank
    click RESTAPI "https://github.com/datalab-org/datalab/tree/main/pydatalab" "pydatalab REST API on GitHub" _blank

The main aim of datalab is to provide a platform for capturing the significant amounts of long-tail experimental data and metadata produced in a typical lab, and enable storage, filtering and future data re-use by humans and machines. datalab is targeted (broadly) at materials chemistry labs but with customisability and extensability in mind.

The platform provides researchers with a way to record sample- and cell-specific metadata, attach and sync raw data from instruments, and perform analysis and visualisation of many characterisation techniques in the browser (XRD, NMR, electrochemical cycling, TEM, TGA, Mass Spec, Raman).

Importantly, datalab stores a network of interconnected research objects in the lab, such that individual pieces of data are stored with the context needed to make them scientifically useful.

License

This software is released under the conditions of the MIT license. Please see LICENSE for the full text of the license.

Citation

Should you use datalab in your research, please consider citing the following preprint:

Matthew L. Evans, Joshua D. Bocarsly, Benjamin Charmes, Ben E. Smith, Gian-Marco Rignanese, David Waroquiers, Clare P. Grey, datalab: Federated data management infrastructure for materials chemistry and beyond, ChemRxiv (2026). DOI: 10.26434/chemrxiv.15001945/v1

Contact

We are available for consultations on setting up and managing datalab deployments, as well as collaborating on or sponsoring additions of new features and techniques. Please contact Josh or Matthew on their academic emails, or join the public datalab Slack workspace.

Contributions

Bug reports, feature requests and pull requests are all welcome. Please see CONTRIBUTING.md for how to get started, what we expect from a contribution, and our policy on AI-assisted changes.

This software was conceived and developed by:

• Prof Joshua Bocarsly (Department of Chemistry, University of Houston, previously Department of Chemistry, University of Cambridge)
• Dr Matthew Evans (Department of Chemistry, University of Cambridge, previously MODL-IMCN, UCLouvain & Matgenix)

with support from the group of Professor Clare Grey (University of Cambridge), and major contributions from:

• Benjamin Charmes
• Dr Ben Smith
• Dr Yue Wu

plus many contributions, feedback and testing performed by other members of the community, in particular, the groups of Prof Matt Cliffe (University of Cambridge) and Dr Peter Kraus (TUBerlin) and the company Matgenix SRL.

A full list of code contributions can be found on GitHub.

Funding

Contributions to datalab have been supported by a mixture of academic funding and consultancy work through datalab industries ltd.

In particular, the developers thank:

• Initial proof-of-concept funding from the European Union's Horizon 2020 research and innovation programme under grant agreement 957189 (DOI: 10.3030/957189), the Battery Interface Genome - Materials Acceleration Platform (BIG-MAP), as an external stakeholder project.
• The Faraday Institution CATMAT project (FIRG016) for support of Dr Joshua Bocarsly during initial development of datalab.
• The Leverhulme Trust and Isaac Newton Trust for support provided by an early career fellowship for Dr Matthew Evans.