Cryptographic approaches to privacy in forensic dna databases

Abstract

Advances in DNA sequencing technology and human genetics are leading to the availability of inexpensive genetic tests, notably tests for individual predisposition to certain diseases. While such information is often valuable, its availability has raised serious concerns over the privacy of genetic information. These concerns are further heightened when genetic information is gathered into databases. We study access control for one class of such databases, forensic DNA databases, used to match unknown perpetrators against groups of potential suspects -usually convicted criminals. Our key observation is that for legitimate forensic queries, the sensitive information belonging to the target individual is already available to the querying agent in the form of a blood or tissue sample from a crime scene. We show how forensic DNA databases may be implemented so that only legitimate queries are feasible. In particular, a person with unlimited access to the database will be unable to extract information about any individual unless the necessary genetic information for that individual is already known. We develop a general solution framework, and show how to implement databases which handle certain cases of missing or incorrect DNA tests. Our framework and techniques are applicable to the general problem of encrypting information based on partially known or partially correct keys, and its security is based on standard cryptographic assumptions.