AI alone can’t necessarily identify the next outbreak, but perhaps by monitoring all the data that’s available out there, we could pick up the subtle clues that something is going wrong and refocus attention. It’s a warning system. Nita Madhav, Senior Director, Ginkgo Bioworks Context The modern tools of biotechnology are paving the way for a thriving bioeconomy that will bolster human and environmental health, food and energy security, and more. At the same time, biology can be a source of risk, whether natural or human-made. Ginkgo Bioworks is building the world’s leading platform for cell programming and biosecurity to make biology easier to engineer while enhancing infrastructure to prevent, detect and respond to biological threats. The AI solution AI is critical to Ginkgo’s work. Beginning in 2018, Ginkgo worked with the US Intelligence Advanced Research Projects Activity (IARPA) on a cutting-edge AI-based computational platform, Engineered Nucleotide Detection and Ranking (ENDAR), which can detect and characterize genetic engineering in biological samples. The platform, which incorporates a human-in-the-loop process for expert review, analyses the genetic content of a sample and identifies whether and how it was engineered. ENDAR heralds a new generation of AI-based detection tools. Historically, methods for detecting signs of biological engineering have been labour-intensive and limited in terms of the types of samples that could be analysed and the types of genetic modification that could be detected. ENDAR can provide early alerts to anomalous samples, detect a huge breadth of potential genetic modifications, and seamlessly integrate with existing biosecurity infrastructure to collect and analyse complex, real-world biological samples. Impact Rapid, reliable and scalable biological engineering detection has long represented a moonshot goal. This capability is critical to appropriately respond to engineered threats and deter biotechnology misuse in the first place. It can also help protect the valuable intellectual property that underlies the burgeoning bioeconomy. ENDAR, when layered with a widespread, persistent biodetection network, represents a critical piece of infrastructure that illustrates the transformative potential of AI for biosecurity. Ginkgo is continuing to evolve ENDAR as the fields of biotechnology and AI develop rapidly. New bioengineering and synthesis capabilities continue to emerge, such as making more nuanced genetic edits and synthesizing DNA with benchtop devices. These advancements necessitate new biodetection methods to allow biosecurity measures to keep pace. Simultaneously, new AI-based large language models (LLMs) have advanced tremendously, creating vast opportunities for analysing complex genetic datasets. To help unlock this potential, Ginkgo continues to curate large and privacy-minded biological datasets to train such models through its bioengineering Codebase and expansion of its global pathogen monitoring network. CA S E S T UDY 3 Detection and deterrence: how Ginkgo Bioworks uses AI to power next-generation biosecurity Scaling Smart Solutions with AI in Health: Unlocking Impact on High-Potential Use Cases 16
RkJQdWJsaXNoZXIy NzQwMjQ=