Timeline
The series started in Tokyo in 2008 as a meeting about bioinformatics web services. Until 2015, each event was written up as one large consortium paper, published two to five years later. Since 2019, each project publishes its own short report on BioHackrXiv within weeks.
What the community worked on
Each cell is the share of that year's projects that touch a topic. For 2008 to 2015 the projects are the working groups inside the consortium reports (the 2011 column covers 2011 and 2012, the 2013 column 2013 and 2014); from 2019 they are BioHackrXiv papers. 2019 has only three papers, so its shares are noisy. Web services dominate the first years. Linked data dominates 2010 to 2015. FAIR metadata, workflows and knowledge graphs dominate since 2019, and LLM work appears from 2023.
Show as table
Three shifts
Every report
All BioHackrXiv works (latest version of each), with an automatically generated one-line summary. Titles link to the preprint on BioHackrXiv. Citations are OpenAlex counts.
Before BioHackrXiv: the consortium reports, 2008 to 2015
One paper per event, 57 to 85 authors each. Reports published two to three years after the event are cited 22 to 55 times; the two published four to five years late are cited once and six times.
Show the working groups extracted from these reports
People
Most authors appear once: each year, 68 to 82% of BioHackrXiv authors are first-time authors. A small core carries the continuity, and several people who wrote the 2008 reports still publish today.
Authors per event year
Show as table
Most frequent BioHackrXiv authors
Who came: the DBCLS BioHackathons, 2008 to 2025
The participant lists on biohackathon.org cover 15 Japanese BioHackathons. Claude Sonnet extracted each participant and classified their affiliation. The share of company participants rose from in the first three events to in the last three. Most of these companies are small Japanese software firms whose staff list DBCLS and NBDC services as their projects: TogoGenome, PubCaseFinder, jPOST, GlyCosmos, MGeND.
Participants per event, by sector
No international DBCLS BioHackathon took place in 2020 to 2022 (COVID-19). BioHackathon Europe is not included: we did not harvest participant lists for it.
Show as table
Companies that came most often
The core: most events attended
✓ = also a BioHackrXiv author.
Where the work went
Few hackathon reports are cited much: The larger effect shows in other groups' papers. Europe PMC finds peer-reviewed articles that name a BioHackathon in their full text. Some of them are among the most cited papers in data stewardship and bioinformatics. The sentences below are quoted from those articles.
From hackathon report to journal paper
Where it spread
- Biodiversity science cites BioHackrXiv reports most: specimens and sequences, barcoding, biodiversity digital twins.
- Research software policy: the ELIXIR Software Management Plan report feeds the FAIR principles for research software.
- Toxicology and chemistry: INTOXICOM workshops, the NORMAN suspect list exchange, FAIR implementation profiles.
- Spatial omics: the de.NBI SpaceHack contributes to SpatialData and Xenium analysis workflows.
- Industry, Japan: 70 companies sent participants to the DBCLS BioHackathons. Their share rose from about 4% to about 19% of participants (see Who came). Most are small Japanese software firms whose staff list DBCLS and NBDC services as their projects.
- Industry, elsewhere: staff from NVIDIA, DNAnexus, Google, AWS, Roche, Pfizer, BASF and others take part; DNAnexus, NVIDIA and AWS host hackathons that publish on BioHackrXiv; Nebius demonstrated confidential GA4GH task execution at BioHackathon Europe 2024. We found no company product documented as built on a specific hackathon output: industry uptake goes through shared standards.
Observed interoperability
Papers and acknowledgements say where work was done. They do not show whether resources actually work together. This section measures that directly. We matched every database, ontology and identifier scheme named in the reports and acknowledgements ( resources) to the Bioregistry. Then we compared them with the other registered resources, and checked their live services on .
Wired into shared infrastructure
Share of resources with each cross-registry link. Hackathon-touched resources compared with all other Bioregistry resources.
Show as table
TogoID: ID conversion pairs over time
Wikidata identifier properties by creation year
Live checks: do the connections work today?
SPARQL endpoints that BioHackathon groups built on or published to, plus federated queries that only work when identifiers and vocabularies line up across resources.
How to read this
- The comparison shows association, not cause. The hackathons choose large, central resources, and those resources would be well connected anyway. The time data is more informative: for
- What the data does show: the resources this community works on make up the connected core of life-science data. That core has resolvable identifiers, Wikidata properties, UniProt and TogoID links, and live SPARQL endpoints. LLM tools now query the same core through MCP.
Method and data
- All BioHackrXiv records are harvested from the OSF API. Their number matches Crossref and Europe PMC. Versions are merged into one work each.
- The 2008 to 2015 reports come from Europe PMC and F1000Research full text. Claude Sonnet splits them into working groups.
- Every paper is summarised and classified by Claude Sonnet against a fixed list of 32 topics. Single labels can be wrong; the trends across years are robust.
- Citations: OpenAlex. Acknowledgements: Europe PMC full-text search. Interoperability: Bioregistry, Wikidata (QLever), TogoID git history, live HTTP and SPARQL checks.
- Author identity is by name string. 2026 is partial.
Downloads