**华大基因团队研发SpatialGlue图神经网络模型登Nature子刊**
近日，新加坡科技研究局（A*STAR）、华大基因和上海交通大学医学院附属仁济医院联合宣布，他们共同研发了一种具有双注意力机制的图神经网络模型——SpatialGlue。这一创新技术在空间转录组学领域取得了重要突破，整合多组学数据，于6月21日在《Nature Methods》上发表最新研究成果。
SpatialGlue模型能够以空间感知的方式整合多种数据模态与其空间背景，揭示组织样本的组织学结构。相比传统方法，SpatialGlue展现出更高的解析能力，可捕捉更多的解剖细节。研究展示了大脑皮层等多个空间域的分析结果，证明了该方法的准确性。更令人振奋的是，该模型还识别出了原始数据未标注的三个不同区域的细胞类型，如脾脏巨噬细胞亚群。
专家表示，这一研究不仅展示了SpatialGlue模型的强大能力，也凸显了多模态空间组学在分析生物复杂性方面的巨大潜力。这一进步对生物医学研究及未来临床治疗具有重大意义。
随着研究的深入和技术的不断进步，相信不久将来，我们能够更深入地理解生命的奥秘，为人类的健康福祉作出更大的贡献。
英语如下：
News Title: BGI Team’s SpatialGlue Graph Neural Network Model Unveils the Mystery of Spatial Multi-omics on Nature Sub-journal
Keywords: Gene Research, Spatial Perceptual Integration, Multi-omics Data
News Content: **BGI Team Develops SpatialGlue Graph Neural Network Model and Publishes in Nature Sub-journal**
Recently, the Singapore Agency for Science, Technology and Research (A*STAR), BGI, and Renji Hospital affiliated to Shanghai Jiao Tong University School of Medicine jointly announced the development of a dual attention mechanism graph neural network model called SpatialGlue. This innovative technology has achieved significant breakthroughs in spatial transcriptomics, integrating multi-omics data, and published its latest research findings on June 21st in Nature Methods.
The SpatialGlue model can integrate various data modalities and their spatial backgrounds in a spatially-aware manner, revealing the histological structure of tissue samples. Compared to traditional methods, SpatialGlue demonstrates superior analytical capabilities, capturing more anatomical details. The research showcases the analysis results of multiple spatial domains such as the cerebral cortex, proving the accuracy of the method. More excitingly, the model has identified cell types in three different regions that were not originally annotated in the data, such as spleen macrophage subsets.
Experts indicate that this research not only showcases the powerful capabilities of the SpatialGlue model but also highlights the enormous potential of multi-modal spatial proteomics in analyzing biological complexity. This advancement holds significant importance for biomedical research and future clinical treatments.
With continued research and technological advancements, we believe that we will be able to better understand the mysteries of life and make greater contributions to human health and well-being in the future.
【来源】https://www.jiqizhixin.com/articles/2024-07-03-3
Views: 2