Google AI subsidiary DeepMind recently published a paper introducing its developed AlphaMissense model. This model, based on AlphaFold for predicting protein structures, undergoes fine-tuning and can automatically predict the 71 million possible missense mutations in the human genome. Studies show that AlphaMissense's predictions closely match experimental MAVE results, with an accuracy rate of 90%. This not only aids medical professionals in more accurately predicting genetic and rare diseases, providing new insights for disease prevention and treatment, but also drives advancements in multiple fields including molecular biology, genomics, and clinical medicine. DeepMind has open-sourced the AlphaMissense model and its predictions, allowing researchers to use them for free. This research achievement, published in the prestigious journal Science, demonstrates DeepMind's strong capability in leveraging AI for cutting-edge fundamental scientific research.