Simplifying in silico protein evolution with minimal screening by unZipro.

unZipro achieves a 61% success rate in identifying high-activity protein variants from minimal libraries across nine diverse proteins, boosting gene-editing activity up to 28-fold.

• Why it matters: Current AI-driven protein engineering methods face challenges like high computational costs, limited success, and are underexplored for plant proteins, hindering progress in biotechnology and agriculture.
• What they did: The approach combines a pre-trained inverse folding model with meta-learning to create family-specific fitness landscapes, enabling zero-shot in silico evolution with about 10 candidates per experiment.
• The result: This framework enables cost-effective, scalable protein design, leading to improved plant gene-editing tools, potent luciferase variants, and antiviral proteins, transforming biological engineering.