Hong Yingzhou, Song Kai, Zhang Zongbo, Deng Yuxia, Zhang Xue, Zhao Jinqian, Jiang Jun, Zhang Qing, Guo Chunming, Peng Cheng
Center for Life Sciences, School of Life Sciences, Yunnan University, Kunming, 650500, China.
Cell Death Discov. 2023 Jul 27;9(1):264. doi: 10.1038/s41420-023-01569-w.
To understand how brain regions form and work, it is important to explore the spatially variable genes (SVGs) enriched in specific brain regions during development. Spatial transcriptomics techniques provide opportunity to select SVGs in the high-throughput way. However, previous methods neglected the ranking order and combinatorial effect of SVGs, making them difficult to automatically select the high-priority SVGs from spatial transcriptomics data. Here, we proposed a novel computational pipeline, called SVGbit, to rank the individual and combinatorial SVGs for marker selection in various brain regions, which was tested in different kinds of public datasets for both human and mouse brains. We then generated the spatial transcriptomics and immunohistochemistry data from mouse brain at critical embryonic and neonatal stages. The results show that our ranking and clustering scheme captures the key SVGs which coincide with known anatomic regions in the developing mouse brain. More importantly, SVGbit can facilitate the identification of multiple gene combination sets in different brain regions. We identified three dynamical sub-regions which can be segregated by the staining of Sox2 and Calb2 in thalamus, and we also found that Nr4a2 expression gradually segregates the neocortex and hippocampus during the development. In summary, our work not only reveals the spatiotemporal dynamics of individual and combinatorial SVGs in developing mouse brain, but also provides a novel computational pipeline to facilitate the selection of marker genes from spatial transcriptomics data.
为了了解脑区如何形成和运作，探索发育过程中在特定脑区富集的空间可变基因（SVG）非常重要。空间转录组学技术提供了以高通量方式选择SVG的机会。然而，以前的方法忽略了SVG的排名顺序和组合效应，使得它们难以从空间转录组学数据中自动选择高优先级的SVG。在这里，我们提出了一种名为SVGbit的新型计算流程，用于对各种脑区中用于标记选择的单个和组合SVG进行排名，该流程在人类和小鼠大脑的不同类型公共数据集中进行了测试。然后，我们在关键的胚胎和新生儿阶段从小鼠大脑生成了空间转录组学和免疫组织化学数据。结果表明，我们的排名和聚类方案捕获了与发育中小鼠大脑中已知解剖区域一致的关键SVG。更重要的是，SVGbit可以促进不同脑区中多个基因组合集的识别。我们确定了三个动态子区域，它们可以通过丘脑Sox2和Calb2的染色来区分，并且我们还发现Nr4a2表达在发育过程中逐渐将新皮层和海马区分开。总之，我们的工作不仅揭示了发育中小鼠大脑中单个和组合SVG的时空动态，还提供了一种新型计算流程，以促进从空间转录组学数据中选择标记基因。
