Privacy-Preserving Whole-Genome Variant Queries

Abstract

Medical research and treatments rely increasingly on genomic data. Queries on so-called variants are of high importance in, e.g., biomarker identification and general disease association studies. However, the human genome is a very sensitive piece of information that is worth protecting. By observing queries and responses to classical genomic databases, medical conditions can be inferred. The Beacon project is an example of a public genomic querying service, which undermines the privacy of the querier as well as individuals in the database. By secure outsourcing via secure multi-party computation (SMPC), we enable privacy-preserving genomic database queries that protect sensitive data contained in the queries and their respective responses. At the same time, we allow for multiple genomic databases to combine their datasets to achieve a much larger search space, without revealing the actual databases' contents to third parties. SMPC is generic and allows to apply further processing like aggregation to query results. We measure the performance of our approach for realistic parameters and achieve convincingly fast runtimes that render our protocol applicable to real-world medical data integration settings. Our prototype implementation can process a private query with 5 genetic variant conditions against a person's exome with 100,000 genomic variants in less than 180 ms online runtime, including additional range and equality checks for auxiliary data.

PILOT: Practical Privacy-Preserving Indoor Localization Using OuTsourcing

“…It is important to note that by using the outsourcing setting we achieve practical PPIL with security against malicious PPIL clients and malicious PPIL servers, whereas in the client/server setting even solutions with substantially weaker security against a semihonest client currently have impractical runtimes and communication. Furthermore, the outsourcing model with multiple STTPs is widely adopted not only in recent academic papers, e.g., for private machine learning [70,71], and genomic privacy [72,73,74,75,76,77], to name just a few, but also deployed in industrial products, see [78] for a few examples.…”

EPISODE: Efficient Privacy-PreservIng Similar Sequence Queries on Outsourced Genomic DatabasEs

“…The privacy of other applications of genomics were also addressed in the literature, e.g., outsourcing of genome data storing [41], pattern matching [43], genome sequence queries [17], and Genome-Wide Association Studies (GWAS) [10,11,15,42], see [33] for a good survey on genomic privacy.…”

Secure Similar Sequence Query on Outsourced Genomic Data

“…Note that we use the two-server architecture which has been widely used in recent works [11,13,26,28]. We don't use the single-server model due to its difficulty in fulfilling various computation operations in our system while achieving good efficiency and security performances simultaneously.…”

“…Privacy-preserving query and analysis over genetic data have received much attention recently. Existing works [7,11,16,33,34,37] deal with genome variant query, pattern matching, range query, count query and statistic information computation over genomic data in a privacy-preserving manner. Secure similar sequence query on genomic data, which is the focus of our research, is a special case of secure query processing on genomic data.…”