In heavy development. Features change without notice, and stored data may need migrating between versions. Keep your own copy of anything you cannot afford to lose.

Metadata that meets the standard

Describe research data against MIAPPE, ISA, Darwin Core, ENA and more — or build your own standard when none of them fits. Validated as you work, and exported in the form a repository expects.

Standards: MIAPPE · ISA · Darwin Core · ENA · PRIDE · MetaboLights · DiSSCo
Your own: Build a specification when no standard fits, and publish it here
Validation: Continuous, against the specification itself
Deposit: Export in the shape the archive accepts
Access: Your institutional account, via SRAM

Research metadata is judged against standards — MIAPPE, ISA, Darwin Core, ENA — and meeting them by hand is slow and easy to get wrong. Metaseed turns a standard into a form that knows its own rules, tells you what is missing while you work, and exports what a repository expects.

Manage datasets

Choose a standard and fill in the entities it defines. Every change is checked against the specification, so you always know what is still missing rather than finding out at submission.

Import an existing record from ENA, PRIDE, MetaboLights or a BrAPI server
Export to the format a repository expects
Full version history — nothing is lost, anything can be restored

Create a dataset

Build specifications

A specification defines a standard: its entity types, their fields, and the rules a dataset has to satisfy. Use a built-in one, or build your own when your work does not fit an existing standard.

Start from a built-in standard or from scratch
Validate as you design, before anyone fills it in
Publish to share it with everyone here — reversible

Open Builder

Explore and compare

Read any specification before committing to it: which entities exist, how they nest, and what each field means.

Compare two standards side by side
Understand a specification somebody else published
See the ontology terms behind each field

Open Explorer

Collaborate

Share a dataset or a draft specification with named colleagues as Owner, Curator or Viewer, and discuss it in threaded comments on the record itself. Publish a specification when it is ready and everyone here can build against it — reversibly, so a mistake is not permanent.

Bring an AI assistant

Claude, or any client speaking the Model Context Protocol, can read and fill in your datasets alongside you. It sees the same specification rules the forms do, so it knows what each field means and what is still missing. Setting it up takes one command, and the instructions are waiting for you once you sign in.

Your work is private to you until you choose to share it.