Efficient Differentially Private Methods for a Transmission Disequilibrium Test in Genome Wide Association Studies

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Investigating the relationship between diseases and human genomes is vital for personalized medicine.
• Large-scale genetic studies, like genome-wide association studies, risk individual identification through released statistics.
• Existing differentially private methods for family-based association tests are computationally intensive and lack guaranteed sensitivity.

Purpose of the Study:

• To develop novel, efficient, and differentially private methods for transmission disequilibrium tests (TDT).
• To address the computational inefficiency and sensitivity concerns of current privacy-preserving TDT approaches.
• To enable large-scale genetic association studies while safeguarding individual privacy.

Main Methods:

• Developed an exact algorithm for TDT with a time complexity of O(nm) for n families and m SNPs.
• Proposed a faster approximation algorithm for TDT with proven sensitivity of 1.
• Evaluated algorithm performance on datasets with varying numbers of families and SNPs.

Main Results:

• The exact algorithm demonstrated a 10,000-fold speed improvement over existing methods for small cohorts (5,000 SNPs).
• The proposed methods are the first capable of handling large cohorts, including those with 10^6 SNPs.
• The approximation algorithm's suitability for specific datasets was examined.

Conclusions:

• The developed differentially private TDT methods offer significant computational advantages and privacy guarantees.
• These advancements facilitate large-scale genomic research essential for personalized medicine.
• The study provides practical tools for privacy-preserving genetic association analysis.