scverse: a consortium for community-driven software development for single-cell analysis

Software infrastructures for bioinformatics

Since 2001, >2000 packages
Distributed, open development model
Tens of thousands of users
Broad coverage of bioinformatics; Biostatistics

Single-cell + adjacent communities (e.g., image analysis)
Link to modern AI/ML via Python
Acceleration via Rust and CUDA

Workshops/conferences/hackathons

2nd scverse conference, Nov 2025, Stanford (US)

Software from the scverse community

107

ecosystem
packages

939

dependent
packages

4,650

dependent
repositories

A community that learns and grows together

343

Unique contributors
for core packages

1,708

Pull requests (last year)
for core packages

1,230

Zulip members
(discussion platform)

https://scverse.org/

scverse is a consortium for community-driven software development for single-cell analysis, addressing use cases such as single-cell atlases, data repositories, data exchange, multi-omics and spatial integration, and model deployment. In the broader landscape of software infrastructures for bioinformatics, Bioconductor and R have been established for 25 years with over 2,000 packages, tens of thousands of users, a broad bioinformatics scope, and strong biostatistics foundations. scverse complements this with Python, providing a link to modern AI/ML, connections to adjacent communities such as image analysis, and a bridge to emerging technologies like Rust and CUDA. On the community side, we have organized workshops, conferences, and hackathons, including the 2nd scverse conference in November 2025 at Stanford. The scverse community contributes 107 ecosystem packages, with 939 dependent packages and 4,650 dependent repositories on GitHub. We have 1,230 members on our Zulip discussion platform, 1,708 pull requests in the last 365 days, and around 72,569 daily downloads.

scverse: bridging user and developer experiences for single-cell biology

Users want: do science, analyse data, discover things in nature

Easy installation, good documentation, unit and integration tests

Developers want: low friction, easy maintainance, little overhead

Templates (scverse cookiecutter), dependency management, CI/CD, automation

Bridge users ↔︎ developers

Online support, in-person events
Let users become developers
Only when you teach a tool, you realize how it should be
Development driven by real scientific needs
Don’t reinvent the wheel: interoperability

scverse bridges users and developers. For users, the goal is to let scientists focus on scientific discovery — with easy installation and comprehensive documentation. For developers, we remove friction through automation: CI/CD, the scverse cookiecutter template, and dependency management. The bridge between these two worlds is built through online support, in-person events, and a teaching philosophy where developers help scientists become developers themselves, and gather crucial feedback in the process. Software interoperability — for instance Python and R/Bioconductor via AnnDataR — ensures broad accessibility. As the operations figure shows, this pipeline flows from the core team maintaining 12 core packages, through hackathons and workshops with hundreds of participants, to thousands of developers producing tutorials and documentation, ultimately advancing scientific discovery.

Networking and national infrastructure

How can your project benefit from networking with other projects?

Optimally exploit shifting paradigms and workflows from AI: e.g., hard and solid core methods, LLM-enhanced glue and user interfacing
Share and learn from each other: what works, what doesn’t

How can your project contribute to a national integrated structure for research software?

Broader collaboration on modular, reusable components for shared problems

What should a national integrated structure for research software achieve?

European digital sovereignty to have the tools for doing the science we want to do
Set the environment for sustainable scientific software development by both, professional RSEs and amateur developers who are PhD or postdocs in their main jobs — through training, good incentives, research performance assessment, careers

Three questions to address. First, networking: we want to share and learn from each other — what works, what doesn’t — and collectively navigate the shift that AI is bringing to our workflows, for instance by building tools for AI agents and making documentation LLM-ready. Second, contribution: we believe in reusing and contributing to modular, reusable components across projects, so we can support and advance each other’s efforts rather than reinventing redundant tools that increase the maintenance burden. Third, the vision: a national integrated structure should enable us to work together towards end-to-end open, free scientific workflows. Concretely: for teaching and outreach, scverse will partner with de.NBI and Bioconductor; for reproducible workflows, with nf-core and Galaxy; and for bridging adjacent scientific communities such as bioimage analysis, with German BioImaging, IO-FAST and BioImage Archive.

The DFG funding will enable establishing a robust full-time core for scverse

scverse team
(part-time or on voluntary basis)

Philipp Angerer	Pau Badia i Mompel	Danila Bredikhin	Can Ergen-Behr	Emma Dann
Severin Dicks	Jennifer Foltz	Ilan Gold	Lukas Heumos	Sara Jimenez
Mikaela Koutrouli	Ori Kronfeld	Luca Marconato	Giovanni Palla	Roshan Sharma
Gregor Sturm	Tim Treis	Wouter-Michiel Vierdag	Isaac Virshup	Kai Zhang