Secure large-scale genome-wide association studies using homomorphic encryption.

Area of Science:

• Genomics
• Computational Biology
• Cryptography

Background:

• Genome-wide association studies (GWASs) are crucial for understanding complex diseases by identifying genetic variants.
• Individual-level data privacy concerns pose a significant challenge for large-scale GWASs.
• Existing privacy-preserving methods often require user interaction and are computationally intensive.

Purpose of the Study:

• To develop and demonstrate a privacy-preserving framework for conducting accurate GWASs.
• To overcome the limitations of traditional GWAS data sharing by encrypting individual-level data.
• To improve the efficiency and scalability of privacy-preserving GWAS analyses.

Main Methods:

• Implementation of a novel framework leveraging advances in homomorphic encryption.
• Performance evaluation on a real-world dataset exceeding 25,000 individuals.
• Extrapolation of performance for larger datasets (100,000 individuals, 500,000 SNPs).

Main Results:

• Accurate GWAS analysis performed on encrypted individual-level data for over 25,000 participants.
• Demonstrated significant speed improvements (over one order of magnitude faster) compared to prior secure multi-party computation methods.
• Achieved similar accuracy to existing methods while eliminating the need for user interaction.

Conclusions:

• The proposed homomorphic encryption framework offers a highly efficient and accurate solution for privacy-preserving GWASs.
• This approach significantly advances the feasibility of large-scale genetic analyses while safeguarding participant privacy.
• The developed techniques have broader applicability in secure, large-scale statistical analyses of encrypted data across various domains.