Protein editing has long been viewed as a potential solution for treating a variety of diseases and addressing other biological challenges. However, previous approaches have often fallen short because altering a protein can unintentionally change its structure or function.
Despite the name, the AI “raygun” does not literally zap and shrink a protein. Instead, Rohit Singh and his colleagues at Duke University in Durham, North Carolina, developed a method based on the existing protein language model ESM2. The AI uses mathematical representations of protein sequences to learn patterns in how amino acids are arranged and how proteins have evolved. Researchers can then use these patterns to design smaller versions of proteins that may retain their original function.
The method has been tested on a variety of proteins, including haemoglobin. However, it is still unclear whether the approach will work effectively with much larger or more complex proteins. Amazingly, the researchers were also able to shrink fluorescent proteins such as EGFP while retaining their useful properties.
This could open the door to new experimental research. Some proteins currently used as the “gold standard” in laboratories are relatively large, which can make them difficult to use in certain experiments. Designing smaller versions that retain their function could make these proteins more practical and potentially reduce the effect they have on the other proteins or biological systems they are being used to study.
While the technology is still at an early stage, the ability to computationally design proteins of different sizes could provide researchers with a powerful new way of studying and engineering proteins.
Leave a comment