Community

Community & Roadmap

Track project progress, explore what is coming next, and meet the team behind BioChef.

Status

Project status

The web app, Hub CI, and Registry are live today. Agent mode is in active development; workflow aggregation is next. Federated learning is on the longer roadmap.

Live

BioChef web app

Tools page + Workflow page; bioinformatics tools (GTO, minimap2, samtools, bcftools, bedtools, seqtk, muscle, kalign, fasttree, ivar, prodigal, mummer4, fastp, and more) running in WebAssembly via biowasm.

Live

biochef-hub + Registry

Reusable GitHub workflow signs and publishes recipes to registry.biochef.app.

In development

Agent mode

BioChef Agent runs signed bundles outside the browser, next to the data. Already working: steps can run in containers (Apptainer); the agent can be set to refuse any bundle whose signature, SBOM, SLSA provenance or blob digests do not check out, and to require a bearer token; runs can be stopped and outlive the request; logs stream live; large files stream instead of being held in memory; each run records how it was produced. Workflow aggregation comes next.

Roadmap

What is next

Planned milestones and current progress.

Complete

BioChef web app: Tools + Workflow pages

A catalogue of 34 signed bioinformatics recipes (GTO, minimap2, bcftools, bedtools, samtools, seqtk, muscle, kalign, fasttree, ivar, prodigal, mummer4, fastp, and more) compiled to WebAssembly, in-browser DAG editor, workflow export and import. Live at biochef.app.

Complete

biochef-hub CI + signed Registry

Reusable GitHub workflow that validates, builds and tests recipes, generates and checks an SBOM, publishes bundles to registry.biochef.app, signs them with cosign, attaches SLSA provenance, and publishes a signed catalogue that the web app verifies.

In Progress

biochef-recipes catalog growth

Expand the public recipe catalogue with more tools, more contributors, and richer metadata. 15 of the 34 recipes also build natively, so the agent can run them outside the browser.

In Progress

Agent mode

Runtime that executes signed bundles outside the browser, next to the data, with the same provenance guarantees. Working today: optional containerised steps (Apptainer), an optional strict mode that refuses any bundle whose signature, SBOM, SLSA provenance or blob digests do not check out, optional bearer-token auth, stoppable runs, live logs, streamed large files and a provenance record per run.

Planned

Workflow aggregation

Package multi-step workflows as first-class signed artifacts in the registry, alongside individual tools, so a whole pipeline becomes a single shareable recipe.

In Progress

GA4GH standards and audit in the agent

The agent speaks GA4GH: runs submitted through the WES API, researchers identified by GA4GH Passports, data fetched by DRS URI from hosts named in advance, plus an audit log of who ran what over which data. Implemented and in review.

Planned

Federated learning + server runtimes

Federated Aggregator + Silo Agents, TES, htsget/refget connectors, MCP gateway, observability. Designed, not yet built.

Team

Who builds BioChef

A research team at IEETA, University of Aveiro.

Project leadership

José Luis Oliveira

José Luis Oliveira

Supervisor · IEETA

Active development

RC

Rafael Chuva

MSc student · Federated learning in TREs (IST)

Past contributors

Cite this work

Published in BMC Bioinformatics · 2026

Rosa, J., Andrade, J., Silva, J. M., & Oliveira, J. L. (2026).Biochef: a client-side WebAssembly-based workflow builder for genomic data analysis.BMC Bioinformatics. DOI:10.1186/s12859-026-06431-1

Open access. BioChef code is released under the MIT License. Registered in bio.tools as biotools:biochef.

Funding: BioChef is Work Package 4 of the ELIXIR Commissioned Service on Human Data and Translational Research (HDTR) 2027/28, after HDTR 2024 (WP6, Fed-A-Crate). Also supported by BioData.pt | ELIXIR Portugal, the IEETA strategic project Browser-First Federated Genomics Toolkit, FCT (UID/PRR/00127/2025, 2023.14342.CPCA.A1) and the European Commission (Genomic Data Infrastructure, grant 101081813).

Get involved

Join the project

Open issues, discuss ideas, or open a pull request. Everything happens on GitHub.