Privacy Preserving Genome Analysis

Division Of Basic Sciences - Nci

Investigators

Linked publications & trials

Abstract

Trusted Execution Environments (TEEs) offered by the current generation microprocessors offer a hybrid hardware/software platform to execute programs in a secure and privacy preserving manner. Intel's SGX platform is the quintessential example for available TEEs. Unlike purely algorithmic/cryptographic approaches such as homomorphic encryption or multi-party-computation, TEEs are highly scalable and enable any computer program to be executed within a dedicated, restrictive segment of the memory, not accessible even by the operating system. As a result they have been gaining popularity in large scale applications in computer systems where privacy is of peak importance. Unfortunately the Secure Enclave supported by Intel SGX is currently restricted by 128MB, highly restrictive for human genome scale applications (where a single BAM file could easily exceed 250GB). In order to address this memory limitation we are developing cryptographic protocols utilizing data compression and compressed data processing methods and compressed/sketching data structures to perform basic GWAS, genomic imputation and clinical trial recruitment.