Next-Generation Enzyme Engineering at Ginkgo

Ginkgo Bioworks has published a new white paper, Next-Generation Enzyme Engineering at Ginkgo, describing the enzyme engineering platform Ginkgo has built over the past decade. The platform combines AI and physics-based computational design with flexible, integrated laboratory automation, and the paper emphasizes programs delivered for U.S. Government partners.

Enzymes can achieve in one step what often takes several with traditional catalysts, but they are harder to engineer than binding proteins like antibodies: activity depends on tradeoffs among multiple kinetic parameters, and epistatic mutations can be distributed across hundreds to thousands of residues. The white paper, authored by Aarya Venkat, Tim Curran, Maria Fitzpatrick, Nathan Schmidt, and Kunal Mehta, describes how Ginkgo approaches this through enzyme discovery, zero-shot design, and iterative ML-guided optimization, with building and testing carried out on Reconfigurable Automation Carts (RACs), Catalyst software, and Nebula, Ginkgo's 100+ RAC system.

Highlights

• Enzyme discovery from a library of close to six billion genes, roughly 60% of them proprietary to Ginkgo, now delivers 1–10 positive hits from screens of tens to hundreds of candidates rather than thousands.
• Adding physics-based modeling to ML raised the share of designs outperforming the seed enzyme from ~20% to ~55% in one program, and from ~25% to ~73% in another.
• Optimization campaigns delivered enzymes with more than 80x and 10x activity improvements to clients, and a 99.6% improved T7 RNA polymerase became Ginkgo's first commercial enzyme product.
• Enzyme engineering protocols now run on Nebula, Ginkgo's autonomous lab of 100+ Reconfigurable Automation Carts, from DNA synthesis and cell-free or in vivo expression to magnetic bead purification and HiBiT quantification of up to 10,000 samples per batch.
• A 2027–2028 roadmap for further automating library design, protein production, purification, and assays, and a path for transitioning this technology through federal R&D programs including SBIR/STTR.

Download the white paper (PDF)