Science Updates

Key Takeaways: dEVA (design by EVolutionary Algorithm) is a framework that achieves the zero-shot design of a highly efficient metalloenzyme without any reliance on natural templates. Design objectives were tailored to the specific chemistry of metalloenzymes, in particular catalytic zinc sites, and from only three experimentally-tested designs found desB, the most efficient de novo designed…

Key Takeaways: With advancements in AI, the field of protein engineering has shifted from template-based approaches to de novo designs with approaches including seed-based design, deep generative models, and binder hallucination. Design towards specific protein functions is the current frontier factoring in parameters including neosurfaces, conditional binding with biological stimuli, and dynamic modeling, linking deep…

Key Takeaways: NVIDIA’s new Proteina-Complexa model combines generative AI with inference-time search algorithms to operate 30-60x faster than RFdiffusion when designing custom proteins that target specific diseases. In a test to find binders for 127 different targets from over 1 million protein designs, the model successfully found binders for 68% of the targets. The model…

Why this work matters Cyclic peptides are attractive scaffolds because their closed backbone can promote structural rigidity and proteolytic stability. Yet assigning sequences that reliably fold into a given cyclic backbone remains a key challenge. In this study, the authors introduce CyclicMPNN, a fine-tuned version of ProteinMPNN designed specifically to improve sequence design for short…

Why this research matters Pharmaceuticals often end up in wastewater after use, and monitoring them can be technically demanding. Traditional analytical methods are powerful but typically require specialized equipment and off-site processing, which can limit how frequently water is tested. This study describes a protein-based biosensor designed to detect two widely used “profen” non-steroidal anti-inflammatory…

Broader significance: a screenable path from computation to in vivo Virtual screening can now evaluate billions of molecules. But that scale creates a practical problem. Docking pipelines often return thousands of candidates that pass computational filters, and predicted binding energies do not reliably predict what will work in living systems. This study introduces a workflow…

Why this matters When a protein is designed on a computer, tiny “impossible overlaps” between atoms can slip into the model. Those clashes may look small on-screen, but they can push a real protein to adopt a different shape than intended. In a December 2025 bioRxiv preprint, the authors show that steric clashing is a…