A new proof of concept demonstrates a private genomics study model using Stoffel MPC, allowing aggregate calculations without collecting individual DNA data. This approach contrasts with traditional studies that require participants to upload highly identifiable genetic information.

Stoffel MPC aims to make privacy-preserving applications accessible to developers. Its platform enables multiple computers to compute on private data without any single party seeing complete inputs. The proof of concept simulated 100 participants contributing synthetic SNP values to calculate aggregate allele counts and a cohort score. Each simulated user's data was locally shared before computation, ensuring no central server or individual party accessed raw genomes.

The experiment's architecture involved 100 client processes, four Stoffel MPC parties, and a coordinator. The private computation, written in Stoffel Lang, aggregated values securely, revealing only cohort totals. This exploration focuses on the MPC option via Stoffel, comparing it to fully homomorphic encryption (FHE) which uses a different data privacy model. The source code is available on GitHub.