Researchers have developed a new AI tool called neural iterative selection-expansion (NISE) that designs proteins to bind specific drugs. Unlike traditional methods that rely on lengthy laboratory testing, NISE uses an AI-guided design process to quickly predict and improve protein-drug interactions. In early experiments, nearly all of the AI-designed proteins successfully bound to their target drugs.
The researchers created custom proteins for two medicines. One was exatecan, a powerful cancer drug, where the protein protected a fragile part of the molecule. This approach could eventually improve drug delivery, keep cancer drugs active longer, and reduce side effects. The other was apixaban, a blood thinner, where the designed protein could serve as the basis for a future antidote to reverse the drug’s effects during emergencies.
Although the technology has not yet been tested in animals or humans, it could provide a faster way to develop custom drug carriers, sensors, and protective proteins, potentially making medicines safer and more effective.