Scalable Privacy-Preserving Data Sharing Methodology for Genome-Wide Association Studies

• Activity Number: 106
• Type: Invited
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Committee on Privacy and Confidentiality
• Author(s): Fei Yu*+ and Stephen Fienberg and Aleksandra Slavkovic and Caroline Uhler
• Companies: Carnegie Mellon and Carnegie Mellon and Penn State and Institute of Science and Technology
• Keywords: genome-side association study ; single-nucleotide polymorphism ; differential privacy ; genetics ; chi-square statistic ; penalized logistic regression

Abstract

The protection of privacy of individual-level information in GWAS) databases has been a major concern of researchers' following the publication of "an attack" on GWAS data in Homer et al. (PLoS Genetics 2008). Traditional statistical methods for confidentiality and privacy protection of statistical databases do not scale well to deal with GWAS databases especially in terms of guarantees regarding protection from linkage to external information. The more recent concept of differential privacy is an approach that provides a rigorous definition of privacy with meaningful privacy guarantees in the presence of arbitrary external information. Building on such notions, we will explore methods for differentially private database release mechanisms pertaining to the Pearson's $\chi^$ test, the trend test, and penalized logistic regression. We also provide interpretation of the methods by assuming the controls' data are known, which is a realistic assumption because some GWAS use publicly available data as controls. We assess the methods' performance through the risk-utility analysis on a real dataset consisting of DNA samples collected by the Wellcome Trust Case Control Consortium.