根据细胞的“邻里结构”预测分子特性 AI模型助力绘制最精细小鼠脑图

Core Insights - The collaboration between the University of California, San Francisco, and the Allen Institute has led to the development of an AI model named CellTransformer, which has created the most detailed mouse brain map to date, encompassing 1,300 brain regions and subregions [1] - This new model utilizes a Transformer architecture similar to that used in large models like ChatGPT, allowing for the analysis of spatial relationships between adjacent cells to predict molecular characteristics and construct a detailed brain organization atlas [1] - Unlike previous brain maps that were primarily based on cell types, this new approach focuses on the brain's structural organization, automatically defining boundaries based on cellular and molecular features rather than relying on human judgment [1] Summary by Sections AI Model Development - CellTransformer accurately reproduces known brain regions such as the hippocampus and identifies new, more subdivided subregions in less understood areas like the midbrain reticular formation [2] - The model's data-driven approach reveals numerous unknown regions, which are likely associated with unexplored brain functions, akin to transforming a map that only shows continents and countries into one that includes cities [2] Broader Applications - The potential of the CellTransformer model extends beyond neuroscience, with its algorithms being applicable to other organ systems and even cancer tissues, utilizing spatial transcriptomics data to uncover biological mechanisms in health and disease [2]