In this talk, I will highlight some existing privacy issues about genomic data sharing and discuss potential countermeasures. First, I will talk about the privacy risks of genomic beacons (standardized interface for data sharing by only allowing yes/no queries on the presence of specific alleles in a dataset) and present a novel re-identification attack which shows that the privacy risk is more serious than previously thought.
It is well-known that sharing one’s genome (e.g., for research) may have serious consequences in terms of privacy. On the other hand, by sharing, one wants to provide useful resources for researchers or benefit from direct-to-consumer services. This creates a trade-off between privacy and utility of genomic data. I will discuss how this trade-off can be studied by using the differential privacy concept. In the remaining of the talk, I will discuss the liability issues that may arise in the case of unauthorized sharing of genomic data and I will introduce a novel optimization-based watermarking scheme for sharing of genomic data.