Software Competence, Self-Sufficiency, and Sovereignty through Open Science
Abstract
A talk given at the Max Planck IMPRS in Leipzig, Germany: https://imprs-coni.mpg.de/events/42858/34550 Abstract: To funders, society, and aspiring early career researchers, open science is a strict necessity in research conduct - and with good reason. But when you’re buried in the everyday workload in science, inches away from a last-minute deadline, or within a rather traditional research environment, open science can feel like an additional burden stacked on top of a large existing todo list. This talk focuses on open science from the perspective of open source research software, which at times may feel particularly cumbersome: Why open tools rather than the integrated analysis suite there is a license for anyway? Why script, when there’s a point and click interface? Why publish my code, or even maintain it, when others may use it to publish faster without any benefit to me? In this talk, I’ll argue for the concrete benefits of these decisions, and explore how open practices and their byproducts empower researchers to take control over their outputs, ensure sustainability, and contribute to digital sovereignty – both for the individual scientist and science and society at large. A rendering of these slides is available at https://files.inm7.de/adina/imprs/
Full text
SOFTWARE COMPETENCE, SELF-SUFFICIENCY, AND SOVEREIGNTY THROUGH OPEN SCIENCE Dr. Adina Wagner Institute of Neuroscience and Medicine, Brain & Behavior (INM-7), Research Centre Jülich DOI: @[email protected] doi.org/10.5281/zenodo.17249652 1
WHO AM I? Free and open source software RSE Open (Neuro-) science Handbook AI RDM me! 2
Science: Its building blocks: Image credit: CC-BY Scriberia and The Turing Way 3
Science: Its building blocks: Image credit: CC-BY Scriberia and The Turing Way 3.1
Science: Its building blocks: Image credit: CC-BY Scriberia and The Turing Way 3.2
Science: Its building blocks: Image credit: CC-BY Scriberia and The Turing Way 3.3
Science: Its building blocks: Image credit: CC-BY Scriberia and The Turing Way 3.4
RESEARCH SOFTWARE IS A SPECTRUM Research software = Software written by scientists, for science 4
RESEARCH SOFTWARE IS A SPECTRUM Research software = Software written by scientists, for science Individual scripts 4.1
MY PATH INTO RESEARCH SOFTWARE 5.4
MY PATH INTO RESEARCH SOFTWARE 5.5
MY PATH INTO RESEARCH SOFTWARE 5.6
MY PATH INTO RESEARCH SOFTWARE 5.7
MY PATH INTO RESEARCH SOFTWARE 5.8
RESEARCH SOFTWARE & OPEN, REPRODUCIBLE SCIENCE GO HAND IN HAND Image credit: CC-BY Scriberia and The Turing Way Reproducibility Management in Neuroscience: Specific issues and solutions (https://doi.org/10.5281/zenodo.4285927) 6
RESEARCH SOFTWARE & OPEN, REPRODUCIBLE SCIENCE GO HAND IN HAND Image credit: CC-BY Scriberia and The Turing Way Reproducibility Management in Neuroscience: Specific issues and solutions (https://doi.org/10.5281/zenodo.4285927) 6.1
RESEARCH SOFTWARE & OPEN, REPRODUCIBLE SCIENCE GO HAND IN HAND Image credit: CC-BY Scriberia and The Turing Way Reproducibility Management in Neuroscience: Specific issues and solutions (https://doi.org/10.5281/zenodo.4285927) 6.2
JUST PUT THE CODE ON GITHUB ... : Open data of 38% of N=174 studies were not "in principle reusable" 24 out of 35 studies with reusable data: irreproducible main results without assistance of the original authors, 13 out of 24: not exactly reproducible even with assistance : N=62 Registered Reports 36 articles shared both data and code to reproduce results Out of those, main findings of 15 articles could not be reproduced Data Management multiple versions of code /data suboptimal data curation unclear execution order/entry points non-portable code Technical problems wrong software environment proprietary software link rot/accessibility issues Human errors reporting (copy-paste) errors ambiguous analysis specification Hardwicke et al., 2018 Obels et al., 2020 How to write a reproducible research article (https://zenodo.org/records/5508797) 7
"Works on my machine" Image credit: derickbailey.com 8
"This used to work on my machine..." Image credit: Based on derickbailey.com 10.1
"This used to work on my machine..." Image credit: Based on derickbailey.com 10.2
"This used to work on my machine..." Image credit: Based on derickbailey.com 10.3
"This used to work on my machine..." Image credit: Based on derickbailey.com 10.4
WHEN YOU WRITE RESEARCH SOFTWARE YOU INCREASINGLY GET... 11
WHEN YOU WRITE RESEARCH SOFTWARE YOU INCREASINGLY GET... ... funding : Open source as a requirement for grants dedicated funding for research software e.g., DFG, Klaus-Tschira-Stiftung, Volkswagenstifung, Helmholtz, ... 11.1
WHEN YOU WRITE RESEARCH SOFTWARE YOU INCREASINGLY GET... ... funding : Open source as a requirement for grants dedicated funding for research software e.g., DFG, Klaus-Tschira-Stiftung, Volkswagenstifung, Helmholtz, ... ... recognition : software is academic output: The San Francisco Declaration on Research Assessment ( ), the Agreement on Reforming Research Assessment ( ), the DFG, and Helmholtz (soon). DORA CoARA 11.2
WHEN YOU WRITE RESEARCH SOFTWARE YOU INCREASINGLY GET... ... funding : Open source as a requirement for grants dedicated funding for research software e.g., DFG, Klaus-Tschira-Stiftung, Volkswagenstifung, Helmholtz, ... ... recognition : software is academic output: The San Francisco Declaration on Research Assessment ( ), the Agreement on Reforming Research Assessment ( ), the DFG, and Helmholtz (soon). DORA CoARA ... citations : Dedicated Software Journals to make even individual scripts citable CITATION.cff standard 11.3
WHEN YOU WRITE RESEARCH SOFTWARE YOU INCREASINGLY GET... ... funding : Open source as a requirement for grants dedicated funding for research software e.g., DFG, Klaus-Tschira-Stiftung, Volkswagenstifung, Helmholtz, ... ... recognition : software is academic output: The San Francisco Declaration on Research Assessment ( ), the Agreement on Reforming Research Assessment ( ), the DFG, and Helmholtz (soon). DORA CoARA ... citations : Dedicated Software Journals to make even individual scripts citable CITATION.cff standard ... careers : Technical skills are valued Research Software Engineer (RSE) positions 11.4
12
CAPTURE COMPUTATIONAL PROVENANCE Which data was needed at which version, as input into which code, running with what parameterization in which computional environment, to generate an outcome? # execute any command and capture its output # while recording all input versions too % datalad run --input ... --output ... <command> 17
EXHAUSTIVE CAPTURE ENABLES PORTABILITY Precise identification of data and computational environments combined with provenance records form a comprehensive and portable data structure, capturing all aspects of an investigation. # transfer data and metadata to other sites and services # with fine-grained access control for dataset components % datalad push --to <site-or-service> 18
REPRODUCIBILITY STRENGTHENS TRUST Outcomes of computational transformations can be validated by authorized 3rdparties. This enables audits, promotes accountability, and streamlines automated "upgrades" of outputs # obtain dataset (initially only identity, # availability, and provenance metadata) % datalad clone <url> # immediately actionable provenance records # full abstraction of input data retrieval % datalad rerun <commit|tag|range> 19
ULTIMATE GOAL: (RE-)USABILITY Verifiable, portable, self-contained data structures that track all aspects of an investigation exhaustively can be (re-)used as modular components in larger contexts — propagating their traits # declare a dependency on another dataset and # re-use it a particular state in a new context % datalad clone -d <superdataset> <url> <path-in-dataset> 20
WHERE DO I PUT MY STUFF? Services make *the* difference for advertisement, discovery, convenience, openess often chosen based on (free) availability, customs in the lab or field, institutional requirement or offer 21
WHERE DO I PUT MY STUFF? Services make *the* difference for advertisement, discovery, convenience, openess often chosen based on (free) availability, customs in the lab or field, institutional requirement or offer But: Imply dependencies 21.1
WHERE DO I PUT MY STUFF? Services make *the* difference for advertisement, discovery, convenience, openess often chosen based on (free) availability, customs in the lab or field, institutional requirement or offer But: Imply dependencies Make sure data/metadata are self-contained and portable to facilitate/enable transition to another service! 21.2
PORTABILITY "FAIRly big": A large-scale data processing framework, built on the foundation of fully portable data analyses. Process 50k subjects on a super-computer in a day, or on a regular cluster in 6 weeks - or take any subject and recompute on your laptop FAIRly big: A framework for computationally reproducible processing of large-scale data. Wagner et al., 2022 22
PORTABILITY "FAIRly big": A large-scale data processing framework, built on the foundation of fully portable data analyses. Process 50k subjects on a super-computer in a day, or on a regular cluster in 6 weeks - or take any subject and recompute on your laptop FAIRly big: A framework for computationally reproducible processing of large-scale data. Wagner et al., 2022 22.1
THE INFRASTRUCTURE WE RELY ON... Mar 2025 www.404media.co/nih-archives-repositories-marked-for-review-for-potential-modification 23
OPEN SCIENCE PRACTICES FIT DIGITAL SOVEREIGNTY Open science principles for research tools Accessible Open formats and standards Customizable Interoperable Open source , www.sovereign.tech/ eu-stf.openforumeurope.org 27
OPEN SCIENCE PRACTICES FIT DIGITAL SOVEREIGNTY Open science principles for research tools Accessible Open formats and standards Customizable Interoperable Open source , www.sovereign.tech/ eu-stf.openforumeurope.org ... but isn't this often also more complex? 27.1
COMPLEX OR UNFAMILIAR, CONVENIENCE OR LOCK-IN? 28
COMPLEX OR UNFAMILIAR, CONVENIENCE OR LOCK-IN? 28.1
CHOOSE YOUR TOOLS WISELY Open source: You can have the code, customize, fork, self-host Interoperability instead of lock-in: Favour open formats, extensibility Unfamiliar does not necessarily mean unintuitive/complicated Centralized things have a single point of failure Community standards (e.g. BIDS) > idiosynchratic solutions Where you can't choose, have a Plan B in place (Lab requires MatLab? Make sure coded also runs in octave!) 29
EXAMPLE: OPEN TOOLS IN MEDICINE : Collaboration between university clinics in NRW and researchers in INM-7. Provides open source research software stack for data acquisition/patient monitoring, RDM, and analysis D eci s i on C o llec t i on P ro ce ss i n g A n a l ys i s AB CD abcd-j.de 30
NEUROSCIENCE IS AT THE FOREFRONT OF OPEN SOURCE DEVELOPMENT ... and many more 31
INCLUSIVE, COLLABORATIVE, EDUCATIONAL SPACES Brainhack: Interdisciplinary conference format with educational sessions & components from hackathons & unconferences First Brainhack ever: Here in Leipzig! , , , , Top left: , Top right: , Bottom left: , Bottom right: (Craddock et al., 2016) OHBM Brainhack Brainhack Global Neurohackademy SIPS hackathons Turing Way bookdashes OHBM Brainhack Neurohackademy DataLad Handbook contributor count Turing Way bookdash 32
TAKE HOME MESSAGES Software is a key ingredient to a scientific result. Sharable and shared code improves efficiency, accessibility, reproducibility of science - but more than anything else, the science in your next project. Working on and with software is technical and complex. The skills you'll get from that help to make you independent in science and beyond. Open science and open source go hand-in-hand: For transparency and reproducibility, for science-specific requirements, for open formats, for re-use, and to enable interoperability across tools. And for digital sovereignty. 33
-25. Oct 2025 -25. Oct 2025 Düsseld Düsseld 4 4 Distribits 2025 23.-25. October International conference on distributed data management technologies 2 day conference, one day hackathon @ Haus der Universität Düsseldorf In-person and remote via livestream free! DISTRIBITS.LIVE 34