6 ms·
I Built a Private Genomics Study with Stoffel MPC
- thisisauserid 2mo agoMPC for Multi-Party Computation, not for Model Predictive Control.
- cinntaile 2mo agoNowadays people would have guessed it was Model Context Protocol.
- usernametaken29 2mo agoIn genetics the end user is never the consumer. You have to collect blood, prepare it, sequence it. These are all highly specialised steps in which something will and can go wrong, and that’s why typically a hospital carries them out. That’s before even getting the device to do sequencing. The cheapest nanopore comes you at a couple ten thousand euros. Given the premise your work strikes me as oddly theoretical. Who is that mystery home DNA lab that needs decentralised private compute? Also, and funnily enough, for much of the genomic pipelines to run (eg a paternity test or a cancer test), a home computer is sufficient. Again, of course, you would need proper medical training to read, interpret and judge the results, which is why a hospital does it… so what problem exactly are you solving?
- timschmidt 2mo agohttps://www.the-odin.com/ https://www.the-odin.com/ $599 for 30x coverage sequencing of any whole genome: https://www.the-odin.com/whole-genome-sequencing-30x/ https://www.the-odin.com/whole-genome-sequencing-30x/ Or purchase the equipment you need to DIY.
- optionalsquid 2mo agoThat's a kit for sample collection. You would then ship off your sample to the company in question, and they would do the actual sequencing, and then send the DNA data and genetic variants they've inferred back to you. I think that defeats what the article author is trying to accomplish, since you still share your entire genome with a third party. The actual instruments used by that company are a whole lot more expensive, and you have to buy reagents as well: https://www.machinio.com/illumina/dna-sequencers https://www.machinio.com/illumina/dna-sequencers
- vishakh82 2mo agoWe have shown that we can batch user samples and then send them anonymously to the lab. Users can then use an MPC protocol to securely fetch their data without losing anonymity. We plan to roll out a product called Batcher which does this soon.
- usernametaken29 2mo agoSo essentially you’re telling people: “Don’t use the evil labs they can read your DNA, instead send DNA to our lab and with magic maths we won’t be able to read your DNA but diagnose you!” Sounds about right to me
- vishakh82 2mo agoNot our lab. We simply aggregate the anonymous samples and then work with third party accredited labs.
- optionalsquid 2mo agoI'm not sure I understand how that that would make a material difference. The company might not know who submitted each sample, but the company would still have their entire genomes in their possession. That is no different, than if somebody obtained the data from a biobank. The data in biobanks is normally also anonymized, insofar that you can anonymize whole-genome DNA data
- intheitmines 2mo agoHave you used them? Interested myself
- vishakh82 2mo agoThere are many motivations for consumer genomics and they might pick up further. https://vishakh.blog/2025/07/08/using-mpc-for-anonymous-and-private-dna-analysis/ https://vishakh.blog/2025/07/08/using-mpc-for-anonymous-and-... If you look at a previous experiment we did, the costs for getting genotyped are pretty moderate (<$80) and going down further. Blended genome-exome (<$150) and full genome sequencing (<$500) are getting cheaper as well. Through our Explorer product (https://explorer.monadicdna.com/ https://explorer.monadicdna.com/) we let people glean information about their genome using GWAS Catalog data. This PoC addresses the next step, i.e. how to keep genetic data private at the user level while still allowing aggregate studies to happen, with or without financial incentives.
- mishellaneous 2mo agocouldn't this argument could be made against a lot of scientific research? i bet it took a long time for the average consumer to benefit from the discovery of the double helix structure of the DNA. and the government funds that kind of stuff even. so what's wrong with a private person doing it?
- optionalsquid 2mo ago> Most genomic studies begin by asking participants to upload one of the most identifying and irrevocable pieces of data they own, their DNA. I'm curious what the author means by "asking participants to upload ... their DNA", because I am not aware of any study that asked participants to do that.
- vishakh82 2mo agoYou're right, most analyses are done using biobanks, etc right now. The sentence is an oversimplification but people still implicitly contribute their data to insecure and centralized sources. If we can enable people to get sequenced not just when they are sick but for altruistic or financial reasons without leaking their DNA data, we could unlock much bigger datasets than currently available, especially from populations that are underrepresented or ignored.
- doctorpangloss 2mo agoIt's a very interesting project. What are the big picture goals? I am asking from the POV of having spoken to many people in the genomics data ecosystem, such as pharmaceutical "chief commercialization officers," not just engineers.
- vishakh82 2mo agoLet people get their DNA sequenced without giving up anonymity or privacy. Let people benefit from DNA data analysis without the data leaving their device or sphere of control. Let people monetize their genetic data without compromising their security.
- doctorpangloss 2mo agookay, it's interesting that you chose those three goals, i would have picked three very different goals (i support your mission don't misunderstand me). here is an illuminating reason why: let's say i recorded a video of the ENTIRE end user experience of (A) collecting the genomics data with your solution (B) using an existing solution. which scene in the video would be different? see, the problem is, not a single scene would be different.
- vishakh82 2mo agoI think I understand what you’re saying. We’re focusing on an incremental improvement. There are people actively working on private home sequencing, at home analysis devices, etc and they’re best placed to do it. We’re trying to get ahead of a near future where people, implicitly or explicitly, are paid for their genetic data in insecure or exploitative ways.
- doctorpangloss 2mo agoif i recorded the video of "private home sequencing," what scene would be different? i suppose you would put the sample collector into some kind of not-yet-existent affordable device? and then what, you get an sd card? all the value is in the analysis and sharing. define insecure or exploitative. it is a very shallow argument to say, iNsUrAnCe cOmPaNiEs - genomic data harvested by insurance companies will never benefit the consumer, but we already have laws against that scenario, the technology doesn't really matter. or the narrow situation with, something something, medical research derived from tissues and the tissue-donators not given a just compensation? i don't know if that's exploitative. and people are already walking, insecure DNA-donators. you are surrounded by easily deanonymized DNA samples everywhere you walk right now...
- badcryptobitch 2mo agoFounder of stoffelmpc.com here. Thank you vishakh82 for sharing. It's great seeing folks apply PETs to key problems, no matter how theoretical they might seem now as evidenced by the comments. PoCs like this are steps in the right direction to make such work and ideas practical. If any of you are looking to try out Stoffel, you can check out our docs at docs.stoffelmpc.com or go to our github repo: https://github.com/Stoffel-Labs/stoffel https://github.com/Stoffel-Labs/stoffel
- mfld 2mo agoVery interesting project! I can imagine usefulness even beyond the intended use case, namely in joining the information from multiple large genomics studies. The sharing and access of such data becomes increasingly difficult, and by this approach one could increase the effective sample size while keeping the data private.
- vishakh82 2mo agoThanks! And, yes, exactly..MPC can be a boon for such data sharing.
- enbarca 2mo agoVery cool PoC for self-sovereign AI! Federated learning across private data resident on user's phone makes sense, but still begs the question of how the user gets access and controls access to their genomic data in a private way. From my perspective, the solution requires giving the user programmatic control over their data, no matter where it resides. At least that's the premise behind work on LifeCare.ID. Let me know if your open to collaborating. https://lifecare.id https://lifecare.id