Privacy-Preserving Data Exploration in Genome-Wide Association Studies

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Genome-wide association studies (GWAS) are crucial for identifying genetic disease bases.
• Published GWAS data is often removed from public databases due to privacy concerns, hindering research.
• Current privacy-preserving methods for GWAS are limited to specific queries, not exploratory analysis.

Purpose of the Study:

• To develop practical, privacy-preserving data mining algorithms for exploratory GWAS.
• To enable analysts to explore GWAS datasets without prior knowledge of specific SNPs or statistical tests.
• To support privacy-preserving computation of significant SNPs, statistical test significance, and correlation structures.

Main Methods:

• Development of novel privacy-preserving algorithms for GWAS data mining.
• Implementation of algorithms supporting exploratory data analysis, including unknown numbers and locations of significant SNPs.
• Algorithms designed to compute SNP-disease associations, SNP-SNP correlations, and block structures while ensuring differential privacy.

Main Results:

• The proposed framework supports exploratory analysis in GWAS.
• Algorithms accurately compute the number and location of significant SNPs, statistical test significance, and correlation patterns.
• Demonstrated significantly more accurate results compared to existing techniques on real-world datasets, with guaranteed differential privacy.

Conclusions:

• The developed algorithms offer a practical solution for privacy-preserving exploratory GWAS.
• This approach enhances data accessibility and facilitates collaborative research in genomics.
• Guaranteed differential privacy ensures the security of individual genetic information while advancing disease gene discovery.