Improving Biosecurity through Metadata Capture

What happens when a DNA synthesis company receives an order for a sequence that doesn’t resemble anything found in nature? Right now, the answer is a patchwork of case-by-case decisions. And this creates real biosecurity risks.

NTI | bio and Lattice Automation seek to provide a more complete answer to this complicated question in a recent proposal for a biodesign metadata standard.

Biosecurity Gaps Leave DNA Synthesis Vulnerable

Moving from computer-based modeling to working with real DNA carries much higher risks. DNA synthesis companies understand this, and many of them, especially the largest companies, screen both their customers and the orders they receive to help prevent the creation of dangerous biological materials—whether by accident or on purpose. Current screening methods, such as those used by members of the International Gene Synthesis Consortium (IGSC), function by comparing requested DNA sequences to known pathogenic sequences.

However, these methods struggle to detect potential threats when the requested sequence deviates substantially from sequences found in nature—unknown sequences. These novel biological designs are becoming increasingly common as biological AI capabilities continue to advance rapidly. Encountering these design requests can result in an inconclusive screening process, leaving the DNA synthesis provider with a difficult choice: reach out to the customer to initiate a potentially time-consuming and costly follow-up process, or proceed to fill the order without full context.

New Metadata Standard Closes the Gap

We propose a biodesign metadata standard that would capture and transmit details about the biodesign process—not just the DNA sequence itself. This metadata includes the design’s origin, editing history, and the tools and users involved in its creation. By offering this additional context, the standard can enable more accurate risk assessments, particularly when the sequences do not resemble those found in nature. This saves time, reduces false positives and false negatives, and will enable innovative biodesign activities. For example, the captured metadata can form the basis of future workflows or provide a backstop of institutional knowledge in high-turnover environments such as an academic research lab. It can also allow researchers to trace their operations to identify changes to a design that were ineffective or irrelevant to their research.

In addition to aiding in DNA screening, the metadata can create a “digital fingerprint” that tracks who performed each operation, when they performed it, and which tools they used. This digital fingerprint can then be used as part of the customer screening process to identify potentially suspicious changes and minimize the threat posed by malicious actors.

To ensure the benefits are felt across the biodesign ecosystem, the proposed standard can be integrated into existing screening processes. The standard is designed to operate smoothly—it doesn’t make files sizes much larger or add significant time to the process. It can be made “backward compatible” with existing data transmission and adapted to tools coded in different programming languages.

As the capabilities of biological AI tools continue to advance at an accelerating pace, it is vital that DNA synthesis providers and others in the biosecurity space embrace new guardrails to prevent their misuse.

To learn more about the proposed biodesign metadata standard, including the technical specifications, download the white paper here.