Tuning Privacy-Utility Tradeoff in Genomic Studies Using Selective SNP Hiding

Area of Science:

• Genomics and Bioinformatics
• Data Privacy and Security

Background:

• Genomic datasets are vital for understanding human genetics but contain sensitive personal information.
• Sharing genomic data is essential for research, yet poses significant privacy risks.
• Existing privacy-preserving mechanisms, like differential privacy, face challenges with dependent data common in genomics.

Purpose of the Study:

• To develop a novel mechanism for secure sharing of genomic datasets with dependent tuples.
• To mitigate inference attack vulnerabilities in differentially private genomic data.
• To balance privacy preservation with data utility for research.

Main Methods:

• Proposed a utility-maximizing, privacy-preserving approach for sharing genomic statistics.
• Implemented a strategy of selectively hiding single nucleotide polymorphisms (SNPs) from family members within the dataset.
• Evaluated the mechanism on a real-world genomic dataset.

Main Results:

• The proposed mechanism significantly enhances privacy protection for genomic datasets with family relationships.
• Achieved up to 40% better privacy compared to state-of-the-art differential privacy solutions.
• Demonstrated near-optimal minimization of data utility loss.

Conclusions:

• The novel mechanism effectively addresses privacy vulnerabilities in differentially private genomic data sharing.
• Offers a robust solution for researchers needing to share sensitive genomic information securely.
• Balances privacy guarantees with the practical utility required for advancing genomic research.