openSNP

» see opensnp.org as it lived, on the wayback machine

~ the goodbye ~

It began as an email on 13 April 2011 from Bastian Greshake to Philipp Bayer. By October there was a half-working prototype, Helge Rausch had come in to help finish it, and the site went public on 27 September 2011. The pitch was one sentence long: if you had paid a company to genotype you, upload the raw file here, put it in the public domain, and let anyone who wants it have it. No accounts to approve and no institutional review board standing at the door.

It was never funded like a startup and never tried to be. The money in the first year came from winning the Mendeley and PLOS "Binary Battle" contest, which paid $10,001 plus a thousand dollars of Amazon credits, and from a 5,000 euro grant from Wikimedia Deutschland that went on buying twenty-odd genotyping kits for people who were underrepresented in the data and could not afford a test. A Patreon opened in 2015. Three people ran the thing for fourteen years on roughly the budget of a used car.

What accumulated there was stranger and more useful than a pile of genomes. Alongside your genotype file you answered surveys about yourself, so the database paired DNA with self-reported traits, and by October 2012 that already meant 215,546,685 individual genotype calls from 633 users, against 4,743 phenotype entries spread over 130 categories, eye color being much the most popular. Everything was licensed CC0 and published as bulk comma-separated dumps with a JSON API beside them. Uploads were accepted from 23andMe and deCODEme at first, with FamilyTreeDNA and later Ancestry added on.

Direct-to-consumer genotyping never became a mass habit in Turkey the way it did in the United States, where mailing a tube of spit to a company was for years an ordinary Christmas present. So I am reading this one as an engineer rather than as a former user: a small open dataset, run by volunteers, that did the thing the large commercial holders of the same data have never once done.

~ the chronicle ~

~ the proximate reason ~

23andMe filed for Chapter 11 bankruptcy on 23 March 2025. Greshake Tzovaras published his closing post eight days later and called the filing "the proximate reason", partly on plain logistics, since most of openSNP's files had come from 23andMe customers and that supply was about to thin out. The larger argument was about what the fourteen years had actually demonstrated. He wrote that the transformative use of consumer genetic data had turned out to be law enforcement genealogy rather than biomedical research, and pointed at the purchase of platforms like GEDmatch by forensics companies.

He put the conclusion plainly: "The risk/benefit calculus of providing free & open access to individual genetic data in 2025 is very different compared to 14 years ago." Speaking to TechCrunch the next day he was considerably blunter, tying the timing to the political situation in the United States since January 2025 and to a worry that genetic data could be conscripted into a revived eugenics. Those are his characterizations and the grave leaves them as his. One fact from that interview is worth keeping separately, because it cuts against the fear rather than with it: in fourteen years of publishing thousands of genomes openly, openSNP never received a single law enforcement request.

~ twenty days early ~

The announced date was 30 April 2025, and that is the date the press reported and the date most references still carry. The farewell page served at opensnp.org today says something else. Its wording is "we have sunset openSNP on April 10", in the past tense, and the page itself was last modified on 18 April, which puts its writing after the tenth and well before the thirtieth. On the evidence of the site's own words, the deletion happened twenty days ahead of schedule, and nobody has ever explained the gap.

~ what deletion means ~

Almost nothing on this site ends this way. Services die and their databases sit on a disk somewhere, or get sold with the rest of the furniture, or turn up years later on a forum. The operators deleted the email addresses, the passwords, the genotypes and the phenotype answers, and said in as many words that destroying the archive was the responsible form of stewardship for it. The site is a GitHub Pages stub now, signed off "Gratefully, Helge, Philipp & Bastian".

The comparison writes itself, which is why it is worth being careful with. The company whose bankruptcy triggered all this did the opposite: 23andMe's assets, genetic database included, went through a bankruptcy auction and were bought on 14 July 2025 for $305 million. Thirteen thousand people gave their genomes to a volunteer project on the understanding that the data was nobody's property, and when the project ended it kept that promise by erasing them. Fourteen million people paid a company for the same test, and their data was an item on a creditor's schedule.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »