Imaging-based data can be interpreted as a point pattern either via the transcript locations or via cell centroids. Alternatively, the segmented cellular outlines can be interpreted as an irregular lattice. High throughput (HTS)-based approaches are most often recorded on a regular lattice. If resolution is high enough, segmentation into cells can be done. This enables either point pattern or lattice data analysis (Baddeley, Rubak, and Turner 2015; Pebesma and Bivand 2023; Rao et al. 2021). Depending on the data representation and the feature of interest, different approaches of data analysis are available. Shown are exemplary questions divided by feature of interest, type of analysis and their corresponding methods. Data: (left: Shi et al. 2023; right: 10x Genomics 2019). Clustering using Banksy (Singhal et al. 2024); (code).

This website provides an overview of exploratory data analysis techniques tailored to the data modalities arising from the two primary spatial omic methods: image-based methods (e.g., MERFISH, CosMx, Xenium, 4i, IMC) and HTS-based approaches (e.g., Visium, Slide-seq). It accompanies the corresponding publication (Emons et al. 2024) ( available from arXiv).

Imaging-based techniques yield subcellular-resolution images, allowing for the study of cells within their natural context. This data can be analysed as spatial point pattern where the points represent cells. Alternatively, we can rely on the segmentation of each individual cell and interpret the collection of all cells as an irregular lattice.

In contrast, HTS-based methods, produce data represented as a regularly spaced grid corresponding to the sampling locations (spots). Each location has the same area and consists of a collection of different measurements.

First, we consider the technology and categorise into a HTS-based or imaging-based approach. Second, we distinguish between univariate (one type of mark) and multivariate (more than one type of mark) methods and outline different methods in specific vignettes. In addition, we provide overview vignettes summarising the two approaches.

©2024 The pasta authors. Content is published under Creative Commons CC-BY-4.0 License for the text and GPL-3 License for any code.

References

10x Genomics. 2019. “Mouse Brain Section (Coronal), Spatial Gene Expression Dataset Analyzed Using Space Ranger 1.0.0.”

Baddeley, Adrian, Ege Rubak, and Rolf Turner. 2015. Spatial Point Patterns. 1st ed. CRC Interdisciplinary Statistics Series. CRC Press, Taylor & Francis Group.

Emons, Martin, Samuel Gunz, Helena L. Crowell, Izaskun Mallona, Reinhard Furrer, and Mark D. Robinson. 2024. “Pasta: Pattern Analysis for Spatial Omics Data.” arXiv. https://doi.org/10.48550/arXiv.2412.01561.

Pebesma, Edzer, and Roger Bivand. 2023. Spatial Data Science: With Applications in R. 1st ed. New York: Chapman and Hall/CRC. https://doi.org/10.1201/9780429459016.

Rao, Anjali, Dalia Barkley, Gustavo S. França, and Itai Yanai. 2021. “Exploring Tissue Architecture Using Spatial Transcriptomics.” Nature 596 (7871): 211–20. https://doi.org/10.1038/s41586-021-03634-9.

Shi, Hailing, Yichun He, Yiming Zhou, Jiahao Huang, Kamal Maher, Brandon Wang, Zefang Tang, et al. 2023. “Spatial Atlas of the Mouse Central Nervous System at Molecular Resolution.” Nature 622 (7983): 552–61. https://doi.org/10.1038/s41586-023-06569-5.

Singhal, Vipul, Nigel Chou, Joseph Lee, Yifei Yue, Jinyue Liu, Wan Kee Chock, Li Lin, et al. 2024. “BANKSY Unifies Cell Typing and Tissue Domain Segmentation for Scalable Spatial Omics Data Analysis.” Nature Genetics. https://doi.org/10.1038/s41588-024-01664-3.