Personalized medicine software vulnerability uncovered

or toxic to a patient.

Forensic labs and genome sequencing companies that also use this mapping software were temporarily vulnerable to having results maliciously altered in the same way. Information from direct-to-consumer genetic tests was not affected by this vulnerability because these tests use a different sequencing method than whole genome sequencing, Hudson said.

Security cybersleuths
To find this vulnerability, Hudson and his cybersecurity colleagues at the University of Illinois at Urbana-Champaign used a platform developed by Sandia called Emulytics to simulate the process of genome mapping. First, they imported genetic information simulated to resemble that from a sequencer. Then they had two servers send information to Emulytics. One provided a standard genome sequence and the other acted as the “man-in-the-middle” interceptor. The researchers mapped the sequencing results and compared results with and without an attack to see how the attack changed the final sequence.

“Once we discovered that this attack could change a patient’s genetic information, we followed responsible disclosure,” Hudson said. The researchers contacted the open source developers, who then issued a patch to fix the problem. They also contacted public agencies, including cybersecurity experts at the U.S. Computer Emergency Readiness Team, so they could more widely distribute information about this issue.

The research, funded by Sandia’s Laboratory Directed Research and Development program, continues testing other genome mapping software for security weaknesses. Differences between each computer program mean the researchers might find a similar, but not identical, issue, Hudson said. The LDRD funding also supports membership in the National Science Foundation’s Center for Computational Biology and Genomic Medicine.

Along with installing the latest version of BWA, Hudson and his colleagues recommend other “cyberhygiene” strategies to secure genomic information, including transmitting data over encrypted channels and using software that protects sequencing data from being changed. They also encourage security researchers who routinely analyze open source software for weaknesses to look at genomics programs. This practice is common in industrial control systems in the energy grid and software used in critical infrastructure, Hudson said, but would be a new area for genomics security.

“Our goal is to make systems safer for people who use them by helping to develop best practices,” he said.