Researchers used AI-guided design to partially eliminate isoleucine from proteins in an engineered Escherichia coli system by proposing structurally similar replacements across dozens of bacterial ribosome proteins, with results published in Science. The work targets one of the hardest synthetic-genome constraints—removing a non-replaceable amino-acid requirement from the cell’s protein-making machinery. The study built on prior observations that many substitutions in nature trade one amino acid for a similar partner, frequently swapping isoleucine with valine or leucine. The team then focused on ribosome subunits containing isoleucine, testing the feasibility of a “fewer building blocks” approach. While the engineered bacteria are not presented as ready for immediate medical use, the approach points to new routes for making proteins with bespoke properties and supports the broader use of structure prediction to de-risk synthetic biology experiments. The article also highlights expert commentary from Imperial College London’s Tom Ellis and Columbia University’s Harris Wang on the boldness of the ribosome-centered elimination strategy.