Core Insights - The Chinese Academy of Agricultural Sciences has developed the world's largest experimental validation dataset for agricultural gene editing technology and created an AI model named AlphaCD, which can efficiently predict the enzymatic characteristics of over 20,000 cytosine deaminase variants and design new high-performance base editing tools [1] Group 1 - The research team spent three years testing 1,100 variants of cytosine deaminase for catalytic efficiency and off-target effects, establishing a large-scale experimental database covering multidimensional functional characteristics [1] - AlphaCD significantly outperforms traditional methods in predicting key indicators such as catalytic efficiency, off-target activity, targeting window, and motif preference [1] - The development of AlphaCD addresses the limitations of traditional cytosine base editors, which rely on extensive trial-and-error experiments that are costly and time-consuming [1] Group 2 - AlphaCD overcomes the limitations of current AI models that depend on supercomputing resources, allowing model training to be completed on personal computers using a lightweight design [2] - This lightweight design enables academic institutions to conduct high-throughput functional predictions at a low cost [2]