Computational Molecular Biology 2025, Vol.15, No.4, 183-192 http://bioscipublisher.com/index.php/cmb 187 locations into different clusters based on the similarity of gene expression. Most of these clusters represent a group of cells with similar molecular characteristics. However, merely looking at the expression is not enough. Only by taking into account the spatial proximity relationship can these clusters be more "coherent" in their organizational structure. Sometimes researchers also change their approach, using specific marker genes to identify cell types and then map their distribution. If the data resolution is not high and a single point contains multiple cells, the data from single-cell sequencing can still be used as a reference to infer who is in each point. In this way, the cellular composition and spatial pattern in the tumor microenvironment can be gradually restored (Saqib et al., 2023; Zhang et al., 2024). Figure 2 Overview of STAGATE (Adopted from Dong and Zhang, 2021) Image caption: STAGATE first constructs a spatial neighbor network (SNN) based on a pre-defined radius, and another optional one in the dashed box for 10x Visium data by pruning it according to the pre-clustering of gene expressions to better characterize the spatial similarity at the boundary of spatial domains. STAGATE further learns low-dimensional latent representations with both spatial information and gene expressions via a graph attention auto-encoder. The input of the auto-encoder is the normalized expression matrix, and the graph attention layer is adopted in the middle of the encoder and decoder. The output of STAGATE can be applied for identifying spatial domains, data denoising, and extracting 3D spatial domains (Adopted from Dong and Zhang, 2021) 5.2 Inference of intercellular communication networks and modeling of signal paths In the tumor microenvironment, cells do not fight on their own; they "speak" through the cooperation of receptors and ligands. The data of the spatial transcriptome makes it possible to capture this kind of communication. Usually, researchers observe adjacent cells: if one cell expresses a certain ligand and the adjacent cell happens to highly express the corresponding receptor, then there is a high probability that signal transmission is taking place between the two (Chowdhury et al., 2021). By aggregating hundreds or even thousands of such relationships, a communication network can be drawn - nodes represent different cell types or clusters, and the connections are their communication channels. By comparing these relationships with the known signaling pathways, it can be seen which pathways are more "talkative" in specific tumor environments, that is, the most notable part in regulatory and intervention research (Liu et al., 2024).
RkJQdWJsaXNoZXIy MjQ4ODYzNA==