Building upon mechanistic insights into host-adaptation principles of genome editing tools, we will develop next-generation Red/ET systems enabling cross-kingdom, ultra-multiplex editing in microbes, plants, and animals. Through deep learning-driven mapping of programmable nucleases, we will engineer proprietary Cas proteins and CRISPR-Cas analogs. Marine microbial resources will serve as the foundation for creating disruptive gene editing technologies via deep learning integration. This initiative aims to mitigate technological choke points and accelerate China's foundational bioengineering capabilities.