Private genome analysis through homomorphic encryption.

Area of Science:

• Genomics
• Cryptography
• Bioinformatics

Background:

• Genome sequencing generates vast datasets, posing privacy risks when processed in the cloud.
• Outsourcing genome data analysis requires robust privacy-preserving solutions.

Purpose of the Study:

• To present a practical solution for secure genome data processing in the cloud using homomorphic encryption.
• To enable computations on encrypted genome data without decryption, ensuring privacy.

Main Methods:

• Developed algorithms for secure computation of minor allele frequencies and chi-squared statistics.
• Implemented private computation of Hamming distance and approximate Edit distance for encrypted DNA sequences.
• Compared the performance of two homomorphic encryption schemes: BGV and YASHE.

Main Results:

• The YASHE scheme processed data from 400 individuals in approximately 2 seconds for variant association analysis.
• The BGV scheme computed approximate Edit distance for 5K DNA sequences in about 65 seconds.
• Demonstrated the feasibility of secure computation for various genomic analyses.

Conclusions:

• BGV scheme is more efficient for deep circuits like Hamming or approximate Edit distance.
• YASHE scheme is more efficient for low-degree computations such as minor allele frequencies or chi-squared tests.
• Homomorphic encryption offers a viable approach for privacy-preserving cloud-based genome data analysis.