Where this came from

This is not a new observation, and the framework described here is the result of a long series of conversations rather than a single design session.

Two of those conversations were decisive. At the ELIXIR All Hands Meeting in Lyon in early June 2026, and again at the Galaxy Community Conference in Clermont-Ferrand later that month, discussions between Bioconductor and Galaxy people kept converging on the same idea from different directions. There is real and growing appetite for automatically wrapping Bioconductor tools for Galaxy, provided it can be done in a high-quality, developer-driven way rather than as a lowest-common-denominator scrape of function signatures. That qualifier is the whole design constraint. A generated wrapper is only worth having if it is as good as a careful hand-written one, and the way to get there is to have the package author declare the interface deliberately, not have automation scrape it from functions.

The scope widened during those same discussions. Once an author has declared a job precisely enough to generate a good Galaxy tool, that same declaration should carry most of what a general workflow dispatcher needs. It seemed wasteful to spend the effort and get only Galaxy out of it.

Why the GA4GH task model became the goal

The design settled on the GA4GH Task Execution Service (TES) task model as the common denominator.

TES is small. A task is: some input files staged in, a short sequence of executors, each one a container image plus a command run one after another, some resource requirements, and some output files collected out. That sequence is the only structure TES has. No branching, no fan-out, no data flow between tasks; orchestration is explicitly somebody else’s job. That minimalism is what makes it a good target for package developers. If a unit of analysis can be expressed as a TES task, it can be projected onto a Galaxy tool, a Nextflow process, a WDL task, or a cloud batch submission without rewriting.

So BiocJobs declarations are shaped around that model, and Galaxy became one target among several rather than the only one.

What a job looks like

A package opts in by adding two files under inst/biocjobs/. Nothing else about the package changes: no new imports, no code changes, no build-system requirements. Packages that are inherently interactive simply do not add the directory.

The first file is the declaration: what the job consumes, produces, and exposes. Abridged here from the example DESeq2 spec, which declares two inputs, three outputs and nine options:

biocjobs: "1.0"
name: deseq2-differential-expression
package: DESeq2
title: DESeq2 differential expression
description: >
  Runs the canonical DESeq2 workflow on a raw count matrix: size factor and
  dispersion estimation, negative-binomial GLM fitting and a Wald test for
  one pairwise contrast.
version: "1.0.0"
script: scripts/deseq2-differential-expression.R
depends: [apeglm, ashr]

inputs:
  - name: counts
    format: tsv
    label: Raw count matrix
    help: >
      Tab-separated matrix of raw (un-normalized) integer read counts.

outputs:
  - name: results
    format: tsv
    label: Differential expression results

options:
  - name: shrinkage
    type: choice
    choices: [apeglm, ashr, normal, none]
    default: apeglm
    label: Log2 fold change shrinkage
  - name: alpha
    type: float
    default: 0.1
    min: 0
    max: 1
    label: FDR threshold

resources:
  cpus: 1
  memory_gb: 4

citations:
  - doi: 10.1186/s13059-014-0550-8

The second file is the script: plain R, around 130 lines for DESeq2, effectively the analysis script to dispatch. Its first line hands the entire interface over to the declaration:

params <- BiocJobs::jobParams("DESeq2", "deseq2-differential-expression")

That call parses the command line against the declaration, applying type coercion, defaults, numeric bounds, enumerated choices, required-parameter checks and output directory creation. What is notably absent from the script is any argument parsing, any type checking, and any usage message, handled by the BiocJobs framework. The rest of the file is ordinary analysis code reading params$counts, params$alpha and so on.

What comes out

From that one declaration, BiocJobs generates:

Every target launches the same self-locating command, so the artifacts carry no absolute paths and do not drift against the installed package:

Rscript -e 'BiocJobs::execJob("DESeq2", "deseq2-differential-expression")' \
    --counts counts.tsv --coldata coldata.tsv \
    --contrast_factor condition --contrast_numerator treated \
    --contrast_denominator control --alpha 0.05

First implementation, at the BioC2026 hackathon

The first working implementation was built at the BioC2026 hackathon in Seattle in August 2026, spearheaded by Alexandru Mahmoud, and taken far enough to get a first working example.

DESeq2 was chosen for this purpose. It is a popular package, batch-shaped, and is already used in many workflow engines, hence having something to compare against after generating the wrappers. A fork of DESeq2 carries the two inst/biocjobs/ files a maintainer would add.

The job itself was run end to end in R against simulated data, 600 genes by 6 samples with 60 planted differentially expressed genes, recovering the planted signal; the same run through the generated command-line path produced byte-identical results. The generated artifacts were then checked with the tooling each ecosystem uses. The Nextflow module passes nextflow lint with zero findings and executes under -stub-run with correct channel and emit: wiring. The WDL task passes miniwdl check. The Galaxy wrapper validates against Galaxy’s official tool XML schema (XSD), and the TES task against the GA4GH TES 1.1 tesTask schema.

What none of that establishes is whether a generated artifact survives a real workflow run against real data, which is where the next section comes in.

Independent evaluations

The most useful outcome from the Hackathon collaboration was an evaluation by Nextflow and WDL users. The WDL evaluation was documented in the WILDS WDL Library which added a run_deseq2_biocjobs task to the ww-deseq2 module, calling BiocJobs::execJob() as an alternative to the module’s existing hand-written R script, and ran it against real data. It produced valid results tables, normalised counts and plots.

The PR was closed rather than merged, with more testing to come in the future.

Feedback, in the form of GitHub issues, was provided regarding the formatting of the Nextflow module files generated by BiocJobs. Topics to consider include:

• The extent to which we should strive for compatibility with nf-core
• Separate input items vs inputs grouped into tuples, with the latter being useful in multi-sample processing
• Use of the tag directive
• Naming of output files

As a test from the Bioconductor package developer perspective, an example job was successfully developed for the VariantAnnotation package. The job takes as inputs an indexed VCF, a BED indicating regions of interest, and a list of sample names, and produces a TSV of genotypes reformatted as alternative allele counts.

A subproject: per-package containers

Making a job dispatchable exposes a second problem immediately. A generated wrapper needs an environment containing R, the host package, and the job’s declared dependencies, and the generic Bioconductor container ships none of the analysis packages.

That pushed out a parallel subproject: a pipeline to automatically build and host a container per Bioconductor package, or per group of packages, or per BiocJobs script. Each image would carry one package plus everything it declares (Depends, Imports, LinkingTo and Suggests) so that vignettes, examples and the package’s own tests all run inside it.

This is not a replacement for the container infrastructure Bioconductor and BioContainers already provide; it builds directly on top of it. Images are layered on the existing Bioconductor base stacks, both the familiar bioconductor_docker images and the newer bioc2u stack, which installs packages as Debian binaries and so builds far faster. The distinction from what exists today is granularity. Bioconductor publishes broad base images, and BioContainers publishes per-package images built from the Bioconda recipes; what a dispatched job wants is an image scoped to exactly one package and its full declared dependency closure, tracking the Bioconductor release directly. Longer term, the ambition is to work with BioContainers so that these images are published in their Quay repository alongside the Bioconda-derived ones, since that is where workflow authors already look. Per-package images on GHCR are simply the first target, because they can be built and iterated on without coordination.

The one image that exists so far, ghcr.io/almahmoud/deseq2:devel, was built ad hoc from the DESeq2 fork, and is what the WDL evaluation described above actually ran against. The work in progress for building all packages lives at almahmoud/biocpkgcontainers.

How this was built

The design of the specification, meaning what a job declaration contains and what the runtime contract is, came out of the conversations described above and out of a much wider set of them over a longer period.

In the interest of transparency: the first implementation of the generators, and a first draft of this post, were written with substantial assistance from Claude, in order to get something runnable in front of others quickly. The result is a functioning prototype rather than a finished product, and it still needs a great deal of refinement by human hands.

This is early, and here is what would help if you want to contribute

BiocJobs is a work in progress. The spec is marked version 1.0 but should not be considered stable and should still be expected to change before an actual v1 release. Today it supports single-file inputs only, five option types, and one analysis command per job. Collections, multi-file inputs and a CWL generator are on the roadmap. Nothing here is set in stone, and that is deliberate, so any and all feedback is welcomed.

Two groups of people could help enormously right now.

Bioconductor package developers. The single most valuable contribution is adding a job declaration to your own package. The framework has been validated on exactly one package so far, which is not enough to know whether the specification is expressive enough, whether the format vocabulary covers real use cases, or whether the runtime contract survives contact with analyses structured differently from DESeq2. Every additional package is a test of the design. If a job in your package cannot be expressed in the current spec, that is precisely the feedback needed before a first release.

Workflow developers. If you maintain Galaxy tools, Nextflow modules, WDL tasks or nf-core pipelines, generated wrappers need to hold up against the standards you already apply by hand. The WILDS evaluation above is a great model: take a generated artifact, try to use it in a real pipeline, and say plainly where it falls short.

Before a first version is stabilised, the aim is to have job declarations in ten or so packages of different shapes, with the generated artifacts reviewed manually to validate their correctness. If that describes you or your package, open an issue on the BiocJobs repository and say which package you have in mind!

There is also a good opportunity to work on this together in person or remotely. BiocJobs is one of the projects at the BioFAIR 2026 Workflow Interoperability Sprint, a hybrid event running 15 to 17 September 2026 in Milton Keynes, United Kingdom, which brings together developers from across the Bioconductor, Galaxy, Nextflow, nf-core and WDL ecosystems. If you would like to join, in person or remotely, the sprint repository has the details, and the #biofair2026-workflow-sprint channel on Bioconductor Zulip is where planning happens.

Links

• BiocJobs on GitHub, the first implementation, and where to leave feedback as issues
• The DESeq2 fork, used to validate the framework on a first example
• The WILDS WDL Library evaluation of the generated WDL

Acknowledgements

The design of this framework came out of a large collaborative network and a great many conversations among researchers worldwide, particularly across the Bioconductor, Galaxy and OpenWDL communities. It would not exist without the people who kept raising these ideas.

Meet the Council members

• Carissa Chen (Postdoctoral Researcher at University of Padova, Italy)

I am a computational biologist with a background in molecular biology and bioinformatics. My current research combines deep learning and statistical methods to jointly analyse sub-cellular spatial transcriptomics and histology images. I have been a long-time user of Bioconductor packages and I am now taking my first steps into computational method development. By joining the Student-ECR Council, I hope to support interdisciplinary researchers like myself to navigate the Bioconductor community.

• Guillaume Deflandre (PhD student at UCLouvain in Brussels, Belgium)

I am a computational biologist with a background in bio-engineering. Initially out of touch with the biomedical field, I took a leap of faith during my Master’s thesis and absolutely loved it! Since then, I have been doing research on how to optimise peptidoform identification in single-cell proteomics. As soon as I delved into this world of bioinformatics, I was introduced to Bioconductor, its packages and its community. They have helped me so much so far, and by joining the Student-ECR Council, I hope to help other researchers in the same way.

• Jasper Spitzer (Postdoctoral Researcher at the University of Bonn, Germany)

I am a computational biologist with a background in immunology. Throughout my PhD, I’ve focused increasingly on computational work and am now 100% on the computational side. Currently, I am working on large-scale CRISPR screen data as well as single-cell projects and am generally interested in how structure in data can be translated into biology. I’ve been a long-time user of Bioconductor packages and have recently tried to be a more active member of the community. With the positive experience I’ve had in the community, I want to share that fun and excitement by encouraging others to be a part of the community as well.

• Sophie-Marie Wind (PhD student at University of Münster, Germany)

I am a bioinformatician with a focus on high-throughput data analysis and method development. Currently, I am developing an analysis framework for 4C-seq data. My first contact with Bioconductor was during my master’s studies, when I used Bioconductor packages for data analyses. During my PhD, I became more involved in package development and came to appreciate the supportive and inspiring Bioconductor community. This experience motivated me to become more actively involved. As a member of the Student-ECR Council, I hope to share my enthusiasm for bioinformatics, contribute to the community, and help support other early-career researchers.

• Kateřina Matějková (PhD student at Charles University in Prague, Czech Republic)

I am a computational biologist with a background in molecular biology and genetics. My research focuses on splicing analysis in Mendelian diseases. I was first introduced to Bioconductor packages during my Master’s, and during my PhD, I joined EuroBioC. I deeply appreciate the supportive and collaborative nature of this community. Currently, we are working to establish a Czech BioC community and create educational initiatives to support early career researchers working at the intersection of biology and informatics.

• Nyasita Ondari (Research Associate at International Institute of Tropical Agriculture (IITA), Nairobi, Kenya)

I am a bioinformatician focused on building pipelines and workflows for the analysis of large-scale biological data. I have a background in bioinformatics and medical biochemistry. My first introduction to Bioconductor came during my MSc, where I used its packages to analyse my data in R. Since then, my involvement has grown. In March 2025, I instructed the first Bioconductor workshop ever held in Africa, in Nairobi. Today, I lead the Bioconductor Africa Seminar Series and support community engagement and social media across the Bioconductor project.

Our vision

We aim to create an inclusive and diverse environment for students and early-career researchers in the Bioconductor community, from those discovering Bioconductor for the first time to long-time users and developers of Bioconductor packages.

Using this platform, our goal is to create opportunities for early-career researchers (ECRs) to connect with one another, cultivating a positive and supportive community that embraces open science and reproducible research practices. We also aim to empower researchers to contribute to Bioconductor projects, strengthening transparency and encouraging community-driven software development. At the same time, we hope to strengthen the feedback loop between users and developers, encouraging collaboration and helping shape and continuously improve the Bioconductor ecosystem.

A few of our proposed initiatives include:

• Mentorship programme providing peer support alongside technical guidance on areas such as package development, coding best practices, and contributing to Bioconductor
• Targeted seminars covering dissemination of the latest research and career panels
• Workshops such as community-contributed package demonstrations
• “Good First Issue” hackathons to help new contributors gain experience by working on beginner-friendly issues

…and more!

As we’re just getting started, we would also love to hear your ideas for future events and initiatives! If you have any suggestions, please take a few minutes to fill in our poll.

To be notified about upcoming activities, please subscribe to our Zulip channel!

Bioconductor Mentorship Programme

We are launching the Bioconductor Mentorship Programme and are looking for both mentees and mentors for the upcoming intake (tentatively October/November). The proposed focus areas may include:

• Package development: developing R scripts to a fully fledged Bioconductor package
• Technical development: guidance on coding best practices, contributing to existing packages, and navigating the Bioconductor development community.
• Peer mentoring: managing a PhD, career advice and more. Other areas TBC depending on interest, so please let us know if there are any specific areas you would benefit from!

If you are interested in participating in the Bioconductor Mentorship Programme, please express your interest here.

Join us!

We welcome anyone from all career stages and scientific backgrounds to join the Council. Specifically, Council members are people who choose to take an active role, for example by attending meetings when they can and helping to plan or deliver activities according to their interests and availability.

Feel free to pop by our regular Council meetings on the third Tuesday of every month at 9am CET to learn more. As we are welcoming new members who may be from different timezones, the meeting time is amenable to change so please don’t hesitate to contact us.

If you are interested in joining the Student-ECR Council, please connect with us on our dedicated Zulip channel, or reach out to us via email.

Or just say hi and introduce yourself on Zulip! (We need more friends!)

BioC2026 in Seattle

We will be holding a live Q&A and casual Student-ECR gathering during BioC2026 on Tuesday, August 11, from 8:30 to 9:00 am PT. Members of the Council will join by Zoom to meet attendees and answer questions following their pre-recorded lightning talk on Monday. Come along to meet other ECRs in the Bioconductor community and learn more about how to get involved with the Student-ECR Council!

Participants by country

Participants travelled to Turku from across Europe and beyond, reflecting the increasingly global nature of the Bioconductor community. While Finland represented the largest delegation, attendees also joined from Italy, Belgium, Germany, Switzerland, the United States, Sweden, the United Kingdom, Ireland, Spain, Kenya, South Korea, Australia, and several other countries.

Preconference

Ahead of the main conference, EuroBioC2026 hosted two preconference events on June 1-2. These were delivered in collaboration with the University of Turku, CompLifeSci, the Finnish Society for Bioinformatics, and members of the Bioconductor community. Running in parallel over two days, the events allowed participants to either strengthen their analytical skills through hands-on training or contribute directly to the development of Bioconductor software through collaborative coding projects.

Workshop: Orchestrating Microbiome Analysis with Bioconductor

The preconference workshop focused on microbiome data analysis using Bioconductor and followed the Bioconductor Carpentry model, combining interactive instruction with practical exercises. Over two days, participants learned how to import, process, and analyse microbiome datasets using the established Bioconductor workflow; Orchestrating Microbiome Analysis (OMA). The workshop covered diversity analyses, differential abundance testing, and approaches for integrating microbiome data with other omics data types. A key component of the workshop was the use of cloud computing resources from CSC (Finnish IT Centre for Science) using Noppe, which provided participants with immediate access to all required datasets, software, and computing resources. By removing installation and configuration barriers, instructors were able to begin teaching immediately and spend more time focusing on the workshop content rather than troubleshooting technical issues. The platform also ensured that all participants worked within the same environment, creating a smoother learning experience for everyone involved. The workshop was hands-on throughout: participants asked questions, worked through exercises, and discussed how the workflows related to their own projects. This practical, open, instructor-led format aligns well with the approach shared by both Bioconductor and The Carpentries. The workshop was led by Leo Lahti, Himel Mallick, Thomaz Bastiaanssen, Tuomas Borman, and Giulio Benedetti.

Participants during the pre-conference microbiome workshop.

Hackathon

We held the first of our series of hackathons attached to Bioconductor conferences this June at EuroBioC2026 in Turku, Finland. Eighteen in-person attendees worked on four projects focused on interoperability, and at least three of those are now being prepared for submission to BioHackrXiv.

A big congratulations to all the participants for their effort. You can read more about the projects from the EuroBioC2026 Hackathon. We’re looking forward to building on this work at the North American Bioconductor conference, BioC2026, in Seattle this August. See the BioC2026 Hackathon for more details.

Participants during the pre-conference hackathon.

Programme overview

EuroBioC2026 covered both established and emerging areas of computational biology, with a consistent focus on reproducible and open-source research.

Poster sessions

The 68 posters presented at EuroBioC2026 covered a broad mix of biological applications and software development. Topics included spatial omics, microbiome research, proteomics, metabolomics, disease modelling, and machine learning, as well as new packages and infrastructure projects from across the Bioconductor ecosystem. The poster sessions encouraged interactions between package developers, researchers, students, and first-time conference attendees, helping strengthen collaborations across the community.

EuroBioC2026 participants during a poster session.

Celebrating 25 years of Bioconductor

A major highlight of EuroBioC2026 was the celebration of Bioconductor’s 25th anniversary. Since its founding in 2001, Bioconductor has grown from a small collection of software packages into a global open-source community used by thousands of researchers worldwide. Over the past quarter-century, it has become a central resource for reproducible computational biology, providing infrastructure, software, training, and community support across numerous biological disciplines.

One of the highlights of the celebration was a retrospective presented by Maria Doyle, Bioconductor Community Manager, who took attendees through the history of Bioconductor, from the earliest contribution on the Bioconductor support site to the project’s growth into the global community it is today. The presentation highlighted how the project has evolved over the past 25 years and its impact on computational biology. The celebrations continued at the conference dinner, where attendees marked the occasion with a special anniversary cake. During the evening, Levi Waldron, one of Bioconductor’s Principal Investigators, shared a personal reflection on his journey with Bioconductor, from first encountering the project through his collaborations with Martin Morgan to becoming part of its leadership.

Levi Waldron shares a personal reflection on his journey with Bioconductor during the 25th anniversary celebrations.

Infrastructure and tools

Sticker Hexwall

The sticker hexwall returned for EuroBioC2026 following its successful introduction in 2025. The display showcased Bioconductor package stickers contributed by Bioconductor community members and served as a visual representation of the diversity of software projects within the ecosystem.

The hexwall quickly became a popular gathering point and photo location throughout the conference.

The hexwall at EuroBioC2026.

Social interactions and networking

Conference materials

Conference recordings will be available on the Bioconductor YouTube channel in the coming weeks. Auditorium sessions were also live streamed, and Slido was used to facilitate audience questions from both in-person and remote participants, alongside traditional in-room discussion. Presenters were encouraged to upload their slides, posters, and supplementary materials to the Bioconductor Zenodo Community, making conference outputs openly available and citable through persistent digital object identifiers (DOIs).

These resources make conference outputs available to those who could not attend and support continued learning across the community. Additional photos from EuroBioC2026, including talks, workshops, posters, social events, and the conference dinner, are available in the conference photo gallery. A short recap video capturing moments from the conference is also available on YouTube.

Coming up…

The 25th anniversary year will continue when the Bioconductor community gather next at BioC2026, which will take place from August 10-12, 2026 at the Fred Hutch Cancer Center in Seattle, Washington. The conference will continue the tradition of bringing together developers, researchers, and educators to share new software, methods, and applications in computational biology.

Later in the year, the community will head to Melbourne, Australia, for BioCAsia2026, taking place on November 19-20, 2026, immediately following the ABACBS conference. BioCAsia brings together researchers across the Asia-Pacific region for scientific exchange, training, and community building. The BioCAsia Seminar Series has also expanded to a bi-monthly schedule. alongside growing regional initiatives such as the Bioconductor Africa Seminar Series and the Bioconductor Latin America seminar series. Stay connected with the community through dedicated Zulip channels.

EuroBioC2026 concluded with an invitation to Basel, Switzerland, where EuroBioC2027 will take place from September 8–10, 2027. See you there.

Acknowledgements

Access to Information

The information is in a publicly accessible database. We do not recommend connecting directly to the webservice but instead using the accompanied Bioconductor R package BiocMaintainerApp. It provides a Shiny application interface for querying Bioconductor package maintainers’ information.

Thank you

We appreciate maintainers’ cooperation moving forward.

Information

Role responsibilities

The mandate of this role is somewhat broad. Bioconductor, and academic computing generally face a myriad of distinct and interrelated challenges as hardware, computing paradigms, and education environments change rapidly. Improving developer resources for tackling new and existing challenges, modernizing Bioconductor developer onboarding materials (particularly for early career researchers), and improving recognition mechanisms for community members who volunteer time and effort to the Bioconductor project are all general themes within the role scope.

Some Specific Efforts

A few of the specific efforts I’ll be working on in this role include:

How to get in touch

For developer discussions and ideas, the Bioconductor Zulip is the best place to connect.

A Shared Mission: Tooling for Managed Repositories

R-universe was designed as a next-generation package distribution and build system for R. It provides:

• Continuous building and checking of R packages across platforms
  
• Binary packages for Windows, macOS, Linux, and WebAssembly
  
• Transparent and reproducible build environments managed via GitHub actions
  
• Dashboards and metadata APIs for monitoring ecosystem health and activity
  
• CRAN-like package repositories with discoverable metrics and documentation

From the outset, a key objective has been to support curated and reviewed communities — such as rOpenSci and Bioconductor — by offering modern infrastructure without requiring them to redesign their governance model or review processes.

For Bioconductor, this means incrementally introducing piece-wise functionality, with consideration for established release cycles and quality control mechanisms:

1. Setting up independent build and dashboard tooling, replicating processes from the current Bioconductor build systems on R-universe infrastructure
2. Mirroring Windows and macOS binaries produced on R-universe to Bioconductor
3. Exploring further integration of results and metadata produced by R-universe for Bioconductor health/activity monitoring and aiding the curation processes
4. Potential future steps toward deeper automation and harmonization

By taking small gradual steps towards adopting R-universe components, everyone gets the opportunity to experiment with new tooling and evaluate where adjustments may be needed in order to minimize disruption to existing practices.

An important milestone in this venture is that Bioconductor now uses R-universe to build the Windows and macOS binaries, which significantly reduces costs and the maintenance load on the Bioconductor team. Beyond binary distribution, we are currently exploring deeper integration of R-universe’s continuous check results into Bioconductor’s quality control and release processes.

Two Universes: Release and Development

Bioconductor maintains two distinct repositories:

• A release branch for stable packages
  
• A devel branch for ongoing development and the next release cycle

To mirror this structure, we currently operate two dedicated R-universe instances:

• Development branch: https://bioc.r-universe.dev
  
• Release branch: https://bioc-release.r-universe.dev

These universes integrate directly with Bioconductor’s existing Git infrastructure and provide continuous builds for packages in both branches.

Through the R-universe dashboard, package maintainers and users can:

• Inspect cross-platform check results
  
• Review extended BiocCheck diagnostics
  
• Monitor build logs and dependency graphs
  
• Explore rich package metadata and metrics
  
• Publish binary packages for Windows, macOS, and Linux

This provides a familiar yet modern interface for Bioconductor contributors, aligned with what users increasingly expect from contemporary R package infrastructure.

Information about each package is available on https://bioc.r-universe.dev/{pkgname}. For example, https://bioc.r-universe.dev/DESeq2 provides details on the DESeq2 package as shown below:

If this is your first time visiting R-universe, we recommend clicking the “Website Tour” button which will walk you through the most important information in 1 or 2 minutes.

Technical Documentation for Bioconductor Maintainers

The R-universe project maintains comprehensive technical documentation at https://docs.r-universe.dev. For Bioconductor specifically, we created a dedicated section summarizing the most relevant topics for developers to get started with R-universe: https://docs.r-universe.dev/bioconductor/

As the collaboration evolves and new components get introduced, the documentation will continue to be expanded. The goal is to provide Bioconductor maintainers with a clear reference point for understanding how R-universe fits into their development workflow, while maintaining compatibility with the established practices that have made Bioconductor a successful project within the R community.

Looking Ahead

Adopting new infrastructure inevitably involves adjustments. For Bioconductor developers, integrating with a new build and distribution system will likely require some changes to workflows, and time to become familiar with new or different package checks, build diagnostics, and binary distribution.

However, by gradually moving toward common infrastructure, the Bioconductor project will benefit from improvements that are being continuously developed and maintained for the broader R ecosystem. A system based on modern continuous integration (CI) will provide developers with improved tooling, and will give the core team more time to focus on community coordination and quality control, rather than on maintaining costly infrastructure. At the same time, the shared platform provided by R-universe can help to increase the visibility and accessibility of Bioconductor software to the greater R community.

We look forward to continuing this alliance and to working with the Bioconductor community to ensure that the next generation of infrastructure supports the project for many years to come.

Outreachy

Outreachy is a paid, remote internship program whose main goal is to support people from groups underrepresented in tech. They help newcomers to free software and open source make their first contributions.

Interns work remotely for 3 months with experienced mentors from open source communities, gaining hands-on experience, building real-world skills, and contributing to meaningful projects. They receive guidance, feedback, and the support of a diverse and welcoming community.

Our Journeys

Upcoming Opportunities

One of the core values of open source is Community, because it ensures projects remain active, relevant, and adaptable over time, preventing stagnation. The Bioconductor community is participating in the current Outreachy internship round (December 2025 to March 2026).

If you’re inspired by our journey and would like to contribute to the BugSigDB project, either through Outreachy or simply to share knowledge, feel free to join us on the Bioconductor Zulip channel.

Applications for the next internship round (May 2026 – August 2026) will open in early February 2026, so keep an eye out for announcements here.

What we taught and what was learned

During the week-long Bioconductor course at the University of Abomey-Calavi, participants were introduced to the foundations of R and RStudio, good coding practices, tidyverse workflows, data handling, and reproducible research, before progressing into hands-on Bioconductor analyses for bulk RNA-seq using DESeq2 and SummarizedExperiment. Participant feedback reflected exactly these learning goals. Many highlighted how practical and hands-on the sessions were, sharing that “the course was very practical with a lot of explanations” and that practicing directly on their laptops made the material easy to apply. The step-by-step guidance on dplyr, tidyverse, ggplot2, and RNA-seq workflows resonated strongly, with one participant noting “the step-by-step explanation about dplyr and tidyverse was highly useful,” while another said they especially appreciated “good step-by-step guidance on RNA-seq data analysis.” Participants repeatedly praised the teaching approach and the supportive environment, noting that “the trainers always came to help us when we had difficulties without judgment,” and “even when we didn’t understand at first, they always took the time to explain clearly.” Others valued the inclusive, interactive format: “the course was inclusive, with a participatory approach,” and “we all practiced directly with the trainers, who were highly attentive to our concerns.” Many also appreciated the organisation and materials, calling the course “well organized,” with “great interaction,” “well-accessible documents and scripts,” and resources they could return to later.

Agenda and instructor list are on the workshop page.

Language accessibility and bilingual support

This was the first Bioconductor Africa course where French translations of introductory materials were provided. Participants strongly appreciated having French-speaking instructors available throughout the week. Over half of the attendees (52%) found this support very helpful, with another 32% rating it helpful, emphasising that bilingual facilitation made it easier to follow explanations, ask questions, and stay engaged. While most participants felt comfortable with English as the primary teaching language (56% reported that English was sufficient), a significant portion noted that French support enhanced their understanding of the material. The French-translated Intro to R materials were also well received: one-third of participants used them during the workshop, and another third planned to consult them later.

Link to French translated materials here

Highlights

GBioS and URMAPha graciously offered us a tour of their laboratory facilities, providing the instructors and participants with an opportunity to learn more about the impactful research they lead, ranging from safeguarding seed diversity to identifying pharmaceutically active compounds.

Below is a photo of the instructors, hosts, and participants taken after the lab tours.

To close the workshop, the team also hosted a beautiful certificate ceremony, creating a memorable and celebratory end to an impactful week.

Carpentries Instructor Training and Capacity Building

Through support from the CZI EOSS6 grant, two researchers from local institutions in Benin, Kevin Sintondji (URMAPha) and Dédéou Tchokponhoue (GBioS), are participating in Carpentries instructor training to become certified instructors. This initiative is helping build local capacity to deliver high-quality training and support reproducible research practices within their teams and the wider community. As they shared: “The Carpentries training has been extremely valuable for strengthening our skills in data management and reproducible analysis. It helped us adopt more structured workflows, improve collaboration within our teams, and apply these methods directly in our AMR and environmental microbiology projects. This capacity building is already having a concrete impact on the quality of our research.”

In addition to the Carpentries training supported by the CZI EOSS6 grant, two of the other instructors at the Benin workshop, Marie Hidjo (Benin) and Amal Boukteb (Tunisia), were previously certified through the Bioconductor Carpentries program funded by the CZI EOSS4 grant (see summary blog post). This workshop was also an opportunity for these instructors to teach locally and share their expertise within the African bioinformatics community.

Impact of Bioconductor Training across East and West Africa

From surveys conducted across Kenya, Ethiopia, and Benin, 100% of participants reported that they would recommend the Bioconductor training course to colleagues, indicating a strong and meaningful impact. The impact was also evident in skills development: every participant reported a significant improvement in their R abilities. In Benin, 60% of the participants rated their learning gains as “very significant” (scale 5) and 40% as “significant” (scale 4), a clear indication that the training effectively strengthened practical, hands-on bioinformatics skills.

Looking ahead: what participants want

Across Kenya, Ethiopia, and Benin, participants expressed a strong interest in more advanced and diverse bioinformatics training that builds on the foundations of R and RNA-seq. The most recurring areas of interest were:

• Advanced transcriptomics: advanced RNA-seq, transcript-level workflows, and single-cell RNA-seq.

• Genomics & population genetics: variant calling, SNP/CNV analysis, and GWAS/genomic selection.

• NGS & sequence analysis: raw sequence handling, QC, genome assembly, and phylogenetics.

• Metagenomics & microbiome analysis using Bioconductor tools.

• Multi-omics integration: combining genomics, transcriptomics, proteomics, and metabolomics.

• Computational skills: Linux basics, command-line workflows, and version control (Git/GitHub).

• Reproducible research & visualisation: R Markdown/Quarto and more advanced data visualisation.

We’re exploring ways to make Bioconductor training more accessible, including webinars and other online formats. One example is our new online seminar series, which recently featured a talk on “Deep-learning-based Gene Perturbation Effect Prediction Does Not Yet Outperform Simple Linear Baselines” and attracted participants who attended our workshops in Africa this year.

Collaborators & Acknowledgements

This workshop was co-hosted by GBioS and URMAPha. We are deeply grateful for their leadership and commitment to strengthening bioinformatics capacity in Benin. Notably, there was a special link between the Kenya and Benin workshops: the Kenya training went so well that one of its participants, Abdou Mouizz Salaou, helped bring Bioconductor training to Benin, further expanding the impact of Bioconductor capacity-building efforts across the region.

The event was co-organised with the International Institute of Tropical Agriculture (IITA) and the University of Limerick, with funding support from the Chan Zuckerberg Initiative (CZI) EOSS6 grant.

Our organising team: Abdou Mouizz Salaou (GBioS, University of Abomey-Calavi), Aristide Carlos Houdegbe (GBioS, University of Abomey-Calavi), Prof. Enoch G. Achigan-Dako (GBioS, University of Abomey-Calavi), Prof. Victorien Dougnon (URMAPha, University of Abomey-Calavi), Trushar Shah (International Institute of Tropical Agriculture), Laurah Ondari (International Institute of Tropical Agriculture), Maria Doyle (University of Limerick / Bioconductor).

Get involved

• Workshop page
• About Bioconductor training
• Join the Bioconductor Africa mailing list or #bioc_africa channel in Bioconductor Chat