Researchers at Duke University School of Medicine have developed an artificial intelligence framework that can redesign proteins on a scale previously seen only in natural evolution. The tool can create shorter, longer, and highly modified versions of proteins while preserving their structure and function.
Rather than design new proteins from scratch, the method, called Raygun , builds on protein language models — AI systems trained on millions of natural protein sequences — to make extensive modifications to existing proteins.
The findings, published in Nature , open new possibilities for engineering proteins with customized sizes and properties, to be used in biological research and ultimately to be developed as targeted medicines.
“There is a language that governs how a protein's amino acid sequence gives rise to its shape and function, but scientists don’t fully understand that language,” said Rohit Singh, PhD , co-corresponding author of the study, and an assistant professor of biostatistics and bioinformatics and cell biology. “Protein language models act as a kind of translator, learning patterns from millions of protein sequences and linking those patterns to biological structure and function.”
Raygun builds on those models with a new way of representing proteins. Rather than treating them simply as amino acid chains of different lengths, the system converts them into a standardized mathematical representation that captures key patterns learned by the AI.
Because proteins of any size can be represented in the same format, Raygun can compare, shrink, expand, or otherwise redesign proteins while preserving important structural and functional features.
Researchers can steer Raygun using only two settings: one that determines how much a protein's sequence changes and another that controls whether the protein becomes shorter or longer.
A range of tests showed that the AI-generated proteins maintained predicted structural integrity and preserved important functional sites. Studies in living cells confirmed the approach in fluorescent proteins used for imaging and in enzymes that perform essential cellular tasks. Receptor-binding experiments validated Raygun-designed variants of epidermal growth factor (EGF), a protein involved in wound healing that is also linked to cell-growth signaling pathways often altered in cancer.
The approach could help researchers design more efficient tools for gene therapy, biotechnology, and biomedical research while offering new insight into how proteins evolve in nature.
Other Duke authors : Kapil Devkota, PhD, Scott Soderling, PhD, Daichi Shonai, and Joey Mao.
Funding: The National Institutes of Health, the Whitehead Foundation.
Nature
Experimental study
Cells
Miniaturizing and modifying natural proteins with Raygun
29-Jul-2026