Ars Technica AI | April 30, 2026

Researchers from Columbia and Harvard have used AI tools to engineer a portion of the ribosome that works without using isoleucine, one of the 20 standard amino acids in the genetic code. The team's work tests hypotheses about how the genetic code initially evolved — most suggest that earlier forms of life had partial genetic codes and used fewer than 20 amino acids. Isoleucine was chosen because it is one of three highly similar amino acids (along with leucine and valine) and is the most frequently substituted amino acid in the E. coli genome. The researchers first replaced isoleucine with valine in 36 essential genes; 17 of them worked without isoleucine. They then focused on redesigning the ribosome, testing 50 different individual genes that contribute proteins to the ribosome. Using deep-learning protein-design software, they produced alternative protein sequences for 25 of 32 genes with reduced fitness. For the remaining five, they forced changes at the isoleucine positions and let the software design compensatory changes in nearby amino acids, leading to successful redesigns for four of the five problem proteins. The work demonstrates that AI-based tools have matured enough that redesigning proteins to use fewer amino acids is far more realistic than it was just a few years ago.

Read more