Department of Computer Engineering
CS 690 SEMINAR
A New Inference Attack against Kin Genomic Privacy
Computer Engineering Department
With the recent advances in genome sequencing technology, it is possible to obtain genomic data in a very cost effective way. Genomic data gives many opportunities to individuals such as learning their predisposition genes to diseases, creating their family trees, etc. In spite of the fact that it provides many useful information, it also brings some concerns about privacy because of its ability to reveal critical genetic information. As one's genome includes sensitive information not only about him but also about his family members, it is crucial to make sure it is stored and processed in a privacy-preserving way. Preserving privacy is another part of researches on kin genomic privacy. In this work, we introduce a new inference attack to reveal one of the parent's SNPs by using the genomes of other family members'. While doing this, we use haplotype blocks that are the regions inherited together and Mendel's Laws. We evaluated our approach on CEPH / Utah Pedigree 1463 dataset, a Caucasian family that is comprised of 4 grandparents, 2 parents, and 11 children containing partial DNA sequences of all family members. We show that accurate inference can be done by using less data, genomic data of fewer family members.
DATE: 07 Kasım, 2016, Monday @ 16:40