Quantifying Genomic Privacy via Inference Attack with High-Order SNV Correlations

Sahel Shariati Samani, Zhicong Huang, E. Ayday, Mark Elliot, J. Fellay, J. P. Hubaux, Z. Kutalik

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

170 Downloads (Pure)

Abstract

As genomic data becomes widely used, the problem
of genomic data privacy becomes a hot interdisciplinary research
topic among geneticists, bioinformaticians and security and pri-
vacy experts. Practical attacks have been identified on genomic
data, and thus break the privacy expectations of individuals who
contribute their genomic data to medical research, or simply
share their data online. Frustrating as it is, the problem could
become even worse. Existing genomic privacy breaches rely on
low-order SNV (Single Nucleotide Variant) correlations. Our
work shows that far more powerful attacks can be designed if
high-order correlations are utilized. We corroborate this concern
by making use of different SNV correlations based on various
genomic data models and applying them to an inference attack on
individuals’ genotype data with hidden SNVs. We also show that
low-order models behave very differently from real genomic data
and therefore should not be relied upon for privacy-preserving
solutions.
Original languageEnglish
Title of host publicationProceedings of the 2015 IEEE Security and Privacy Workshops
Subtitle of host publicationSan Jose, CA, 21-22 May 2015
PublisherIEEE Computer Society
Pages32-40
Number of pages9
DOIs
Publication statusPublished - 2015

Publication series

Name2015 IEEE CS Security and Privacy Workshops

Fingerprint

Dive into the research topics of 'Quantifying Genomic Privacy via Inference Attack with High-Order SNV Correlations'. Together they form a unique fingerprint.

Cite this