Science (2026). DOI: 10.1126/science.aeb5171"> Simplifying the canonical amino acid alphabet with generative AI. Credit: Science (2026). DOI: 10.1126/science.aeb5171

One of life's many mysteries is how it ended up choosing only a set of 20 amino acids to build proteins for its wide catalog of organisms, from single-celled bacteria to behemoth whales. From a chemical standpoint, many of the canonical amino acids share similar chemical structures and properties, which might make them expendable. This raises an intriguing question: could life manage with one less amino acid?

In a recent study in Science , researchers used generative AI and deep-learning models like AlphaFold2, which can predict protein 3D structures, to design Ec19 —a genetically engineered strain of E. coli that functions using just 19 amino acids instead of the usual 20.

In Ec19, the researchers set out to see if they could remove isoleucine (Ile) and still obtain a living, healthy cell.…