整合组织学图像信息增强空间转录组细胞聚类的分辨率

doi:10.13560/j.cnki.biotech.bull.1985.2024-0334

摘要/Abstract

摘要：

【目的】增加空间转录组基因表达的空间分辨率以提升遗传发育与疾病研究中的细胞谱系和类型变化的精度，提供更精细的分子表型信息。【方法】通过图像分割实现空间转录组点阵的细胞空间分布模拟，使用线性插值方法重构超分辨率基因空间表达，并利用图聚类方法揭示组织中细胞分布的空间偏好性。【结果】将新方法SpaGMM在小鼠后脑10X Visium数据集上进行检验，可以精确识别小鼠脑神经空间结构域。通过与几种空间转录组聚类的常用方法进行比较，结果显示SpaGMM的聚类结果更加符合组织学区域的注释，这些区域具有大量标记基因的空间表达支持。SpaGMM还可以从小鼠小脑区域中区分出浦肯野细胞（Purkinje cell）和伯格曼胶质细胞（Bergmann glial cell）所对应的组织区域，发现不同细胞层中存在互补的基因表达模式。【结论】SpaGMM可以通过提高点阵的空间分辨率揭示组织结构域的精细结构。

关键词: 空间转录组学, 细胞分割, 空间域识别, 细胞聚类

Abstract:

【Objective】Increasing the spatial resolution of transcriptome gene expression can improve the accuracy of cell lineage and type changes in genetic development and disease studies, providing more detailed molecular phenotypic information.【Method】Using image segmentation to simulate the spatial distribution of cells in a spatial transcriptomics grid, the high-resolution spatial gene expression using linear interpolation was reconstructed. Graph clustering methods were used to reveal the spatial preferences of cell distribution within tissues.【Result】Application of SpaGMM on the 10X Visium dataset of the mouse posterior brain helped to accurately identify various spatial domains of the mouse brain. Compared with several commonly used methods for spatial transcriptome clustering, the results showed that the clustering results by SpaGMM were more consistent with the annotated histological regions, which were supported by the spatial expression of many marker genes. SpaGMM also recognized layers corresponding to Purkinje cells and Bergmann glial cells from mouse cerebellum, revealing complementary gene expression patterns in different cell layers.【Conclusion】SpaGMM may reveal the fine structure of tissue domains by improving the spatial resolution of spots.

Key words: spatial transcriptomics, cell segmentation, spatial domain recognition, cell clustering

王睿, 戚继. 整合组织学图像信息增强空间转录组细胞聚类的分辨率[J]. 生物技术通报, 2024, 40(8): 39-46.

WANG Rui, QI Ji. Integrating Histological Image Information to Enhance Cell Clustering Resolution in Spatial Transcriptome[J]. Biotechnology Bulletin, 2024, 40(8): 39-46.

图/表 4

图1 四种算法在10X Visium小鼠后脑数据集性能比较 A：各方法将小鼠后脑数据的分组结果（30组）； B：各方法结果中的离群点数量

Fig. 1 Performance comparison of the dataset of posterior section of mouse brain on 10X Visium platform by four methods A: Illustration of clustering results（30 groups）of posterior section of mouse brain by each method. B: Outlier numbers in clustering results by the each method

图2 SpaGMM在小鼠海马体区域的基因空间表达模式重建及点阵分类结果 A：海马体CA1、CA3和DG标记基因空间表达的对比；B：基于标记基因C1ql2的空间表达信息比较4种方法在齿状回型神经元的分类结果

Fig. 2 Reconstruction of gene spatial expression pattern of SpaGMM in the hippocampus region of mouse brain and classification result A: Comparison of spatial expressions of marker genes in the CA1，CA3 and DG regions of hippocampus. B: Evaluation of the four algorithms on the classification of dentate gyrus type neurons according to spatial expression of marker gene C1ql2

图3 四种方法在小鼠小脑区域分类结果的比较 A：分子层、颗粒层的标记基因空间表达模式和4种方法分类结果的AUC评估；B：SpaGMM将小脑区域划分为连续层级结构；C：展示浦肯野细胞和伯格曼细胞标记基因在内外空间域的表达情况

Fig. 3 Comparison of classification results on the cerebellum of mouse brain by four methods A: Spatial expression pattern of marker genes at molecular layer and particle layer, and evaluation of the classification results of the four methods. B: Illustration of different layers on cerebellum recognized by SpaGMM. C: Illustration of expression of marker genes for Purkinje and Bergman cells in different spatial domains

图4 浦肯野细胞层中Aldoc（Zebrin II阳性）与Cck、H2-D1（Zebrin II阴性）基因的空间表达模式图片由上至下分别代表原始基因表达热图（颜色由蓝到红代表基因表达量从低至高）、SpaGMM重构的超分辨率基因表达热图（颜色由从浅至深代表基因表达量从低至高）、整合多个基因展示浦肯野细胞层中的互补表达模式

Fig. 4 Spatial expression patterns of Aldoc（Zebrin II positive）and Cck, H2-D1（Zebrin II negative）genes in Purkinje cell layer From top to bottom, the images indicate the heat map of original gene expression（the color from blue to red indicates the gene expressions from low to high）, the heat map of SpaGMM reconstructed super-resolution gene expression（the color from light to deep indicates the gene expressions from low to high）, and the integration of multiple genes shows the complementary expression pattern in the Purkinje cell layer

参考文献 35

[1]	Lee J, Hyeon DY, Hwang D. Single-cell multiomics: technologies and data analysis methods[J]. Exp Mol Med, 2020, 52(9): 1428-1442.
[2]	Ahmed R, Zaman T, Chowdhury F, et al. Single-cell RNA sequencing with spatial transcriptomics of cancer tissues[J]. Int J Mol Sci, 2022, 23(6): 3042.
[3]	Lei YL, Tang R, Xu J, et al. Applications of single-cell sequencing in cancer research: progress and perspectives[J]. J Hematol Oncol, 2021, 14(1): 91.
[4]	Moses L, Pachter L. Museum of spatial transcriptomics[J]. Nat Methods, 2022, 19(5): 534-546. doi: 10.1038/s41592-022-01409-2 pmid: 35273392
[5]	Lake BB, Menon R, Winfree S, et al. An atlas of healthy and injured cell states and niches in the human kidney[J]. Nature, 2023, 619(7970): 585-594.
[6]	Galeano Niño JL, Wu HR, LaCourse KD, et al. Effect of the intratumoral microbiota on spatial and cellular heterogeneity in cancer[J]. Nature, 2022, 611(7937): 810-817.
[7]	Li QY, Zhang XY, Ke RQ. Spatial transcriptomics for tumor heterogeneity analysis[J]. Front Genet, 2022, 13: 906158.
[8]	Mazzei M, Vascellari M, Zanardello C, et al. Quantitative real time polymerase chain reaction(RT-qPCR)and RNAscope in situ hybridization(RNA-ISH)as effective tools to diagnose feline herpesvirus-1-associated dermatitis[J]. Vet Dermatol, 2019, 30(6): 491.
[9]	Eng CH L, Lawson M, Zhu Q, et al. Transcriptome-scale super-resolved imaging in tissues by RNA seqFISH[J]. Nature, 2019, 568(7751): 235-239.
[10]	Zhou XY, Karpur A, Luo LJ, et al. StarMap for category-agnostic keypoint and viewpoint estimation[C]// European Conference on Computer Vision. Cham: Springer, 2018: 328-345.
[11]	Fürth D, Hatini V, Lee JH. In situ transcriptome accessibility sequencing(INSTA-seq)[J]. bioRxiv, 2019. DOI: 10.1101/722819.
[12]	Stickels RR, Murray E, Kumar P, et al. Highly sensitive spatial transcriptomics at near-cellular resolution with Slide-seqV2[J]. Nat Biotechnol, 2021, 39(3): 313-319. doi: 10.1038/s41587-020-0739-1 pmid: 33288904
[13]	Kephart M. A 10x visium approach: a spatial RNA-Seq analysis of renal tissue in Peromyscus eremicus[D]. ProQuest Dissertations Publishing, 2023.
[14]	Luoma J, Nastou K, Ohta T, et al. S1000: a better taxonomic Name corpus for biomedical information extraction[J]. Bioinformatics, 2023, 39(6): btad369.
[15]	Xia KK, Sun HX, Li J, et al. The single-cell stereo-seq reveals region-specific cell subtypes and transcriptome profiling in Arabidopsis leaves[J]. Dev Cell, 2022, 57(10): 1299-1310.e4.
[16]	Su G, Qin XY, Enninful A, et al. Spatial multi-omics sequencing for fixed tissue via DBiT-seq[J]. STAR Protoc, 2021, 2(2): 100532.
[17]	Satija R, Farrell JA, Gennert D, et al. Spatial reconstruction of single-cell gene expression data[J]. Nat Biotechnol, 2015, 33(5): 495-502. doi: 10.1038/nbt.3192 pmid: 25867923
[18]	Wolf FA, Angerer P, Theis FJ. SCANPY: large-scale single-cell gene expression data analysis[J]. Genome Biol, 2018, 19(1): 15. doi: 10.1186/s13059-017-1382-0 pmid: 29409532
[19]	Hu J, Li XJ, Coleman K, et al. SpaGCN: integrating gene expression, spatial location and histology to identify spatial domains and spatially variable genes by graph convolutional network[J]. Nat Methods, 2021, 18(11): 1342-1351. doi: 10.1038/s41592-021-01255-8 pmid: 34711970
[20]	Zhao E, Stone MR, Ren X, et al. Spatial transcriptomics at subspot resolution with BayesSpace[J]. Nat Biotechnol, 2021, 39(11): 1375-1384. doi: 10.1038/s41587-021-00935-2 pmid: 34083791
[21]	Scrucca L, Fop M, Murphy TB, et al. Mclust 5: clustering, classification and density estimation using Gaussian finite mixture models[J]. R J, 2016, 8(1): 289-317. pmid: 27818791
[22]	Traag VA, Waltman L, van Eck NJ. From Louvain to Leiden: guaranteeing well-connected communities[J]. Sci Rep, 2019, 9(1): 5233. doi: 10.1038/s41598-019-41695-z pmid: 30914743
[23]	Ahmed M, Seraj R, Islam SMS. The k-means algorithm: a comprehensive survey and performance evaluation[J]. Electronics, 2020, 9(8): 1295.
[24]	Pham D, Tan X, Xu J, et al. stLearn: integrating spatial location, tissue morphology and gene expression to find cell types, cell-cell interactions and spatial trajectories within undissociated tissues[J]. bioRxiv, 2020. DOI: 10.1101/2020.05.31.125658.
[25]	Pan V, Schreiber R. An improved Newton iteration for the generalized inverse of a matrix, with applications[J]. SIAM J Sci Stat Comput, 1991, 12(5): 1109-1130.
[26]	Chung MK. Gaussian kernel smoothing[EB/OL]. 2020: arXiv: 2007.09539. http://arxiv.org/abs/2007.09539.
[27]	Achanta R, Shaji A, Smith K, et al. SLIC superpixels compared to state-of-the-art superpixel methods[J]. IEEE Trans Pattern Anal Mach Intell, 2012, 34(11): 2274-2282.
[28]	Hurtik P, Madrid N. Bilinear interpolation over fuzzified images: enlargement[C]// 2015 IEEE International Conference on Fuzzy Systems(FUZZ-IEEE). Istanbul, Turkey. Piscataway, NJ: IEEE, 2015: 1-8.
[29]	Reynolds D. Gaussian mixture models[M]// Li SZ, Jain A. Encyclopedia of Biometrics. Bostonm MA: Springer, 2009: 659-663.
[30]	De Meo P, Ferrara E, Fiumara G, et al. Generalized louvain method for community detection in large networks[C]// 2011 11th International Conference on Intelligent Systems Design and Applications, 2011: 88-93.
[31]	Xu C, Jin XY, Wei SR, et al. DeepST: identifying spatial domains in spatial transcriptomics by deep learning[J]. Nucleic Acids Res, 2022, 50(22): e131. doi: 10.1093/nar/gkac901 pmid: 36250636
[32]	Chen YM, Zhou SY, Li M, et al. STEEL enables high-resolution delineation of spatiotemporal transcriptomic data[J]. Brief Bioinform, 2023, 24(2): bbad068.
[33]	Franzén O, Gan LM, Björkegren JLM. PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data[J]. Database, 2019, 2019: baz046.
[34]	Perkel DJ, Hestrin S, Sah P, et al. Excitatory synaptic currents in Purkinje cells[J]. Proc R Soc Lond B, 1990, 241(1301): 116-121.
[35]	Lin YC, Hsu CC H, Wang PN, et al. The relationship between zebrin expression and cerebellar functions: insights from neuroimaging studies[J]. Front Neurol, 2020, 11: 315.