July 22, 2026
Your genes, but make it messy
I Built a Private Genomics Study with Stoffel MPC
A privacy-first DNA experiment wowed some readers while others asked if the whole premise even makes sense
TLDR: A developer showed that a DNA study could calculate group results without collecting everyone’s full genetic data in one place. Commenters split fast: fans praised the privacy idea, while skeptics argued the real-world genetics pipeline makes the whole app-based premise shaky.
A developer built a proof-of-concept DNA study that tries to answer a very big, very creepy modern question: can researchers learn useful things from people’s genetic info without ever collecting everyone’s raw DNA in one giant honeypot? Using Stoffel MPC, the project simulated 100 participants and showed that a study could produce group totals without any one party seeing the full dataset. In plain English: the math gets done, but nobody gets the whole genetic file.
And oh, the comments immediately turned into a mini courtroom drama. One camp was intrigued, calling it a smart privacy-first direction and exactly the kind of proof-of-concept that could make sensitive medical research less scary. Even Stoffel’s founder popped in with a polite-but-pointed “yes, it’s theoretical, that’s what prototypes are for” vibe. The other camp? Not buying it so easily. Critics pounced on the setup, arguing that real genetics is messy, expensive, and happens in hospitals and labs, not in a neat app fantasy. One commenter basically said, hold on, who is actually “uploading their DNA” here? Another jumped in with peak internet energy just to clarify that “MPC” means multi-party computation, not model predictive control — the kind of nerdy correction that somehow becomes comic relief.
So the real story isn’t just the code. It’s the clash between privacy dreamers, practical skeptics, and drive-by acronym police — all arguing over whether this is the future of health data or just a very elegant science fair project.
Key Points
- •The article presents a proof of concept showing that 100 simulated participants could compute aggregate allele counts and a weighted cohort score using MPC without any party seeing the full dataset.
- •The implementation uses Stoffel MPC version 0.1.0, with applications written in StoffelLang and executed by the Stoffel VM across multiple MPC parties.
- •The experiment is framed around a Monadic DNA mobile-app model in which users keep genotype data on their phones and participate in studies without uploading complete genomes to a central service.
- •Each simulated user had a separate client identity and local process, created shares locally, and sent different shares directly to MPC parties so that only aggregate outputs were revealed.
- •The article argues that genomics is a strong test case for privacy systems because genetic data is highly identifying, may reveal information about relatives, and cannot be changed after a breach.