Homomorphic encryption enables privacy preserving polygenic risk scores

Area of Science:

• Genomics
• Cryptography
• Precision Medicine

Background:

• Polygenic risk scores (PRSs) are crucial for personalized risk prediction in precision medicine.
• The use of PRSs raises significant privacy concerns due to the sensitive nature of genomic data.

Purpose of the Study:

• To develop and evaluate an open-source implementation of fully homomorphic encryption (FHE) for privacy-preserving polygenic risk score (PRS) models.
• To assess the feasibility and accuracy of using FHE for secure genomic data computation.

Main Methods:

• Developed HEPRS, an open-source system utilizing a three-party model (clients, modelers, evaluators) for FHE-based PRS.
• Applied HEPRS to synthetic datasets and a 110,000 SNP schizophrenia risk model.
• Investigated encryption parameters affecting accuracy, memory, and computation time.

Main Results:

• Encrypted and plaintext PRSs showed close agreement, indicating high accuracy.
• HEPRS demonstrated practical usability on a single CPU.
• The study confirmed that FHE enables realistic PRS with negligible accuracy loss.

Conclusions:

• Fully homomorphic encryption offers a viable solution for privacy-preserving PRS.
• HEPRS supports secure and scalable genomic analytics, enhancing precision medicine applications.
• The developed FHE implementation facilitates confidential risk prediction from sensitive genomic data.