Analyzing the effect of kinship for re-identification attacks in genomic data sharing beacons
Author(s)
Advisor
Ayday, ErmanDate
2019-08Publisher
Bilkent University
Language
English
Type
ThesisItem Usage Stats
272
views
views
127
downloads
downloads
Abstract
Genomic data contains sensitive information about an individual. Family members'
genome sequence can be re-constructed with high confidence or individuals'
may face discrimination because of predisposition of a disease if genome
sequence of a person is obtained. To protect the genomic information and provide
a standardize and secure way for using this data the \Beacon project" initiated.
Studies show that the genomic data sharing beacons are vulnerable to
re-identification attacks. Since beacons generally constructed based on types
of diseases, re-identification creates a significant risk for individuals. On the
other hand, genomic data enables researchers to find the cause of diseases and
improves personalized medicine. Previously proposed counter measures against
re-identification attacks proved to be not effective as earlier researches show. In
this thesis, we analyze the kin relationships' effect on the genomic data sharing
beacons. Our study is based on the fact that kinship may be misleading for
re-identification attacks since same SNPs can be appear in multiple family members.
We showed that adding at least one of the parents to the beacon (i) cause
significant decrease in the power of attacks and (ii) increase in the number of
queries needed to confirm an individual's beacon membership. To investigate the
suitability of using kinship as a counter measure for beacons we also calculate the
utility decrease. We further show the effects of adding more distant relatives to
the beacon such as grandparents.