Reporters learned from the Institute of Genetics and Developmental Biology of the Chinese Academy of Sciences on August 11 that its researchers, together with researchers from other units, have successfully developed a new protein modification method based on artificial intelligence (AI). This method cleverly utilizes existing universal protein reverse folding AI models, enabling efficient protein evolution simulation and functional design without the need to train dedicated AI models. Protein modification is like "modifying" biomolecules by adjusting the sequence of amino acids to alter protein properties. Compared to genetic modification, protein modification is more direct and efficient, and can quickly obtain characteristics that may have evolved in nature over millions of years. However, existing protein modification technologies have obvious shortcomings: traditional methods rely heavily on expert experience, are time-consuming, and costly; The emerging AI prediction technology requires training specialized models for each protein separately, which not only has poor universality but also consumes computing resources. Faced with these challenges, we urgently need to find a smarter way to develop new AI solutions that are highly versatile, efficient, and do not rely on expensive computing power. ”Gao Caixia, the corresponding author of the paper and a researcher at the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, said. The existing universal protein reverse folding AI model can predict the possible amino acid arrangement of proteins based on a given three-dimensional structure. Based on the existing universal protein reverse folding model, researchers have developed the AiCEsingle module and introduced a novel protein modification method. The test results showed that the prediction accuracy of the new method reached 16%, and its performance improved by 36% -90% compared to other common AI models. At the same time, in the experimental verification stage, researchers successfully modified 8 proteins with different functions using new methods, including key gene editing tools such as deaminase. Gao Caixia stated that this new method of intelligent protein modification is more efficient, applicable, and scalable compared to traditional methods. It represents an important trend in the field of life sciences, which is to partially replace laboratory operations with computational simulations. It is worth mentioning that the new method significantly reduces the threshold for the use of AI technology, allowing ordinary laboratories to enjoy the convenience of intelligent prediction without the need for expensive computing power, benefiting more scientists. (New Society)