Rare Variants Analysis in Genetic Association Studies with Privacy Protection via Hybrid System

Genomic data are becoming widely used for diagnosis and treatment. Researchers require to study a significant number of genomes in order to analyze patient genomes to figure out some disease-gene associations. Cost-efficient commercial cloud computing services can host genomic data and conduct genome-wide association studies on-demand at considerably lower cost and high availability and scalability. Due to privacy and security issues, genomic data providers are unwilling to send their sensitive data to the cloud service providers without applying any privacy-preserving measures. This paper proposes a novel hybrid privacy-preserving framework that utilizes twofold cryptographic protocols and minimal perfect hash functions to reach maximum efficiency and security for outsourcing genomic data with the least time overhead. Our framework leverages a trusted execution environment (i.e. software guard extensions) to perform rare variant case-control association tests securely. Also, the genomic data outsourced from multiple data providers are encrypted through additive homomorphic encryption. We perform the Weighted-Sum Statistic (WSS) test, a rare variant association test, and we show that our scheme achieves high accuracy results without an increase of computational complexity or time overhead.