MIT researchers have developed a novel AI-driven method to design nanoparticles for enhanced RNA delivery in vaccines and therapies. The team created an AI model, COMET, inspired by transformer architectures, and trained it on a library of approximately 3,000 lipid nanoparticle (LNP) formulations. COMET predicts how multiple interacting components in LNPs affect their performance, unlike traditional AI drug discovery models that optimise single compounds.
After training, COMET proposed LNP formulations that, when tested, outperformed existing and even some commercially available LNPs in delivering mRNA to mouse skin cells. The AI was also used to identify LNPs that can better withstand lyophilisation, a freeze-drying step crucial for medicine storage. The team extended their approach to incorporate a fifth component, branched poly beta amino esters (PBAEs), polymers known for nucleic acid delivery.
This AI approach offers flexibility and accelerates the development of RNA therapies. Researchers are applying these nanoparticles to RNA therapies targeting diabetes and obesity, including GLP-1 mimics similar to existing treatments. The AI model can also predict LNP efficiency in delivering mRNA to distinct cell types.
