MIT chemical engineers developed an artificial intelligence model leveraging an LLM to analyze the codon patterns of the industrial yeast Komagataella phaffii. By learning and predicting optimal codon sequences, the model enhanced the efficiency of yeast-based synthesis of therapeutic proteins such as human growth hormone and cancer-treatment monoclonal antibodies. This approach aims to reduce time and costs associated with developing protein drugs by improving manufacturing yield predictability.
The study, published in the Proceedings of the National Academy of Sciences, demonstrates the model's application across multiple proteins but does not present independent empirical validation beyond yeast strains examined. The results suggest potential for accelerating biopharmaceutical manufacturing pipeline optimization with AI-informed genetic code design.
Source & further reading
New AI model could cut the costs of developing protein drugs ↗Source published 2026-02-16. This report was prepared with AI assistance from the linked primary-source material. Claims and limitations should be checked against the original announcement or research.