While biobanking is now considered an integral part of translational research by many of the world’s foremost scientists, the industry still faces a number of challenges, especially when it comes to biorepository management. For example, as biobanks begin moving towards personalized medicine, many researchers are trying to find new ways to keep the genetic data and medical records in their freezer inventory secure while still enabling the cloud-based systems needed to accurately analyze this information. Currently, the most efficient solutions seem to be a long way off. However, cryptographers are currently testing a new form of data encryption which could solve this problem sooner rather than later.
On March 16, the University of California, San Diego (UCSD) hosted a workshop for cryptographers analyzing genetic data. During this event, one group was able to use a method called homomorphic encryption to find disease-associated gene variants within ten minutes. The process encountered difficulties when the amount of information was increased, and the scientists admitted that biobanking software is often consumed by other, more realistic tasks. However, many still believe that this process could be part of the future of biorepository management.
As American biobanks anticipate collecting millions of genetic samples from patients as part of the push for precision medicine, many facilities have predicted that they will need to use networked cloud computers to store the sheer quantity of data they will be managing. Unfortunately, several online security breaches in the past few years have frightened industry members so much that administrators at the US National Institutes of Health currently forbid researchers from storing data on Internet-connected computers. Homomorphic encryption could be the perfect answer to these fears: the process allows researchers to encrypt their data on a local data and then deposit this mathematically scrambled form in the cloud. Computations on the data would then be performed in the cloud and an encrypted result would be sent back to the local computer, which translates the information back into a readable form. This means that the underlying data would be protected even if the encrypted form was stolen at any point of the process.
This potential facet of biorepository management was first proposed in 1978 but remained largely theoretical until 2009. Today, proponents say that homomorphic encryption is one of the safest possibilities for modern biobanks: unlike other encryption schemes, this method never allows the cloud to actually “see” the numbers it is working with while still giving accurate calculations. However, others are more skeptical, pointing out that some amounts of data are too large for current methods to handle, and the process doesn’t protect information on scientists’ local computers. Some also doubt that the process provides enough analytical flexibility. But despite these concerns, a few cryptographers have already deployed limited homomorphic encryption systems in some of the world’s biobanks. For example, an HIV study at a hospital biobank in Switzerland is using the process because the project requires only a small set of relatively simple operations on the data, meaning it doesn’t require a great deal of speed. However, due to the nature of the information, the research team is worried about a potential release of their information, making homomorphic encryption an attractive option. Will this form of genetic safety become part of standard biorepository management measures in the future? Only time will tell.