- Spatial transcriptomics and cellular mapping are important areas of bioinformatics that investigate gene expression while preserving information about the physical location of cells within tissues. Traditional bulk RNA sequencing measures gene expression across a mixture of cells, while conventional single-cell RNA sequencing (scRNA-seq) provides gene-expression profiles for individual cells but generally loses their original spatial positions during tissue dissociation. Spatial transcriptomics addresses this limitation by measuring RNA molecules in intact tissue sections or spatially organized biological samples. It enables researchers to investigate not only which genes are expressed but also where their expression occurs, which cells are located near one another, and how molecular activities are organized within tissues and organs.
- Spatial transcriptomics refers to experimental and computational approaches that combine gene-expression measurements with spatial coordinates or tissue-location information. These methods allow researchers to construct spatial maps of gene activity and investigate the organization of different cell populations. A spatial transcriptomic dataset typically contains gene-expression measurements associated with defined tissue locations, spots, cells, or individual RNA molecules, depending on the technology. The resulting information can be integrated with histological images, cell-type annotations, single-cell RNA sequencing, and other molecular datasets to provide a detailed understanding of tissue structure and function.
- The central concept of spatial transcriptomics is the preservation of the relationship between molecular information and tissue architecture. Biological tissues contain multiple cell types arranged in organized structures, such as epithelial layers, blood vessels, glands, muscle fibres, immune-cell regions, and connective tissue. Cells located in different tissue regions may express different genes even when they belong to the same broad cell type. Spatial information therefore helps reveal regional gene-expression patterns, tissue-specific cell states, and interactions between neighbouring populations. For example, immune cells located near a tumour may have different expression profiles from immune cells in surrounding healthy tissue. Similarly, cells within different regions of the brain, liver, mammary gland, or developing embryo may perform distinct biological functions.
- Several experimental technologies have been developed for spatial transcriptomics. Spot-based spatial transcriptomics measures RNA expression at predefined locations, with each spot potentially containing transcripts from multiple neighbouring cells. The 10x Genomics Visium platform is a widely used example. Tissue sections are placed on slides containing spatially indexed capture probes, allowing polyadenylated RNA molecules to be captured and associated with known tissue coordinates. The resulting data provide spatial gene-expression profiles that can be combined with tissue images. Because a spot may contain more than one cell, computational methods are often required to estimate the cellular composition of each location.
- High-resolution spatial transcriptomics aims to measure gene expression at cellular or subcellular resolution. Technologies such as 10x Genomics Xenium, NanoString CosMx Spatial Molecular Imager, MERFISH, and seqFISH use imaging-based approaches to detect selected RNA molecules directly within tissue sections. These methods can provide information about individual cells and the spatial distribution of specific transcripts. Some platforms use targeted gene panels, allowing detailed measurement of selected genes, while other approaches support broader transcriptomic coverage. Their advantages and limitations differ in terms of gene coverage, spatial resolution, sensitivity, throughput, and experimental cost.
- Another important approach is in situ sequencing, in which RNA molecules are detected and, in some technologies, sequenced directly within their tissue context. Methods such as STARmap and related spatial molecular approaches can provide high-resolution spatial information about gene expression. These technologies help investigate the organization of complex tissues and the distribution of molecular signals within individual cells. The choice of spatial platform depends on the research objective, tissue type, required resolution, available sample material, and whether broad transcriptome coverage or targeted gene detection is more important.
- Spatial transcriptomics experiments generally begin with tissue collection and preparation. Tissue samples are collected, preserved, sectioned, and mounted on specialized slides or prepared for imaging-based analysis. Histological staining, such as hematoxylin and eosin (H&E), can provide morphological information that helps identify tissue structures. Depending on the platform, RNA molecules are captured using spatially barcoded probes, detected through fluorescence-based imaging, or processed through in situ molecular techniques. Sequencing or imaging then generates molecular measurements associated with spatial positions. Quality control is performed to evaluate tissue integrity, RNA quality, signal detection, sequencing depth, and the reliability of spatial measurements.
- The computational analysis of spatial transcriptomics begins with processing raw sequencing or imaging data. Platform-specific software converts raw measurements into gene-expression matrices and spatial coordinates. For example, Space Ranger processes data from 10x Genomics Visium experiments, while Xenium Ranger supports processing of Xenium datasets. Imaging-based technologies have their own processing pipelines for transcript detection, cell segmentation, and spatial assignment. The resulting data may include a gene-by-spot or gene-by-cell expression matrix, tissue images, spatial coordinates, quality-control measurements, and cell-boundary information. These datasets form the foundation for downstream spatial analysis.
- Spatial quality control evaluates the reliability of molecular measurements and their relationship to tissue structure. Important metrics may include the number of detected genes, total transcript counts, sequencing depth, percentage of mitochondrial transcripts, tissue coverage, background signal, and the number of detected molecules per cell or spot. For imaging-based methods, segmentation quality and transcript-assignment accuracy are particularly important. Poor-quality tissue regions, damaged cells, background fluorescence, and incorrectly assigned transcripts can affect downstream results. Appropriate filtering and visualization help identify unreliable measurements before further analysis.
- Spatial clustering groups tissue locations or cells according to similarities in their gene-expression profiles and spatial characteristics. It can identify molecularly distinct tissue regions, cellular neighbourhoods, and functional compartments. Methods may use gene-expression similarity, spatial proximity, graph-based clustering, or combinations of these features. Common approaches include Leiden clustering, Louvain clustering, and graph-based methods implemented in tools such as Seurat, Scanpy, and Squidpy. Spatial clustering differs from conventional single-cell clustering because the physical relationships between cells or spots can contribute to the analysis. It can reveal tissue organization that may not be apparent from gene expression alone.
- Cellular mapping involves identifying and locating different cell types within their original tissue environments. Cell-type mapping may be performed using marker genes, reference-based annotation, deconvolution, or integration with single-cell RNA-seq datasets. For example, markers such as EPCAM may help identify epithelial populations, PECAM1 endothelial cells, COL1A1 fibroblasts, CD3D T cells, and MS4A1 B cells. However, the expression of individual marker genes is not always exclusive to one population, so multiple markers and reference-based evidence are generally more reliable. Cellular mapping can produce spatial maps showing where specific cell types are located and how their distributions vary across tissue regions.
- Spatial deconvolution is especially important for spot-based technologies in which individual measurements contain RNA from multiple cells. Deconvolution attempts to estimate the proportions or abundance of different cell types contributing to each spatial spot. It commonly uses reference expression profiles obtained from single-cell or single-nucleus RNA sequencing. Computational methods compare the spatial expression profile with reference cell-type signatures to estimate the cellular composition of each location. Important tools include RCTD, cell2location, SPOTlight, stereoscope, DestVI, and Tangram. These methods differ in their statistical assumptions, ability to model cell states, reference requirements, and computational performance. Their results are estimates rather than direct observations of individual cells and should be interpreted accordingly.
- Spatial gene-expression analysis can also identify spatially variable genes. These are genes whose expression patterns are associated with tissue location rather than being distributed randomly across the sample. Spatially variable genes may mark tissue boundaries, anatomical regions, developmental zones, disease-associated areas, or cellular microenvironments. Methods such as SpatialDE, SPARK, SPARK-X, and spatial analysis functions in Seurat and Squidpy can help identify these genes. Their expression patterns can be visualized using spatial feature plots, heatmaps, and tissue overlays. Functional enrichment using Gene Ontology (GO), KEGG, and Reactome can then help identify biological processes associated with particular tissue regions.
- Spatial neighbourhood analysis investigates how cells are arranged relative to one another. It can identify cellular neighbourhoods containing particular combinations of cell types, such as immune cells surrounding tumour cells or endothelial cells located near perivascular populations. Neighbourhood analysis can also reveal spatial relationships between stromal, epithelial, immune, and vascular compartments. Tools such as Squidpy, Giotto, and Seurat support different forms of spatial graph construction and neighbourhood analysis. These methods are useful for studying tissue organization, cell niches, immune infiltration, and the spatial structure of disease environments.
- An important application of spatial transcriptomics is the investigation of cell–cell communication. Cells communicate through secreted ligands, receptors, extracellular matrix molecules, direct contact, and other signalling mechanisms. Spatial information helps determine whether potential sender and receiver populations are located near one another, making certain local signalling interactions more biologically plausible. Tools such as CellChat, COMMOT, Giotto, and Squidpy can support spatial communication analysis, depending on the data type and method. Integrating ligand–receptor expression with cell locations can help investigate signalling in tumour microenvironments, immune tissues, developing organs, and other complex biological systems. However, spatial proximity and gene co-expression alone do not prove that a functional interaction occurs.
- Spatial transcriptomics can be integrated with conventional single-cell RNA sequencing to combine detailed cell-type identification with preserved tissue architecture. Single-cell data provide reference expression profiles that can help annotate spatial locations, identify cell populations, and estimate cell-type proportions. Spatial data then reveal where these populations are located within the tissue. Integration methods include Seurat, Scanpy, Squidpy, Tangram, cell2location, and Giotto. This combined approach is particularly valuable when spatial platforms have limited gene panels or spot-level measurements that contain multiple cells.
- Spatial analysis can also be combined with other molecular technologies. Spatial proteomics measures protein distribution within tissue sections, providing information about molecular activity that may not be captured by RNA expression alone. Spatial metabolomics investigates the distribution of metabolites and other small molecules. Spatial epigenomics can reveal how chromatin accessibility and regulatory mechanisms vary across tissue regions. Technologies such as CITE-seq and other multi-omics approaches can provide complementary molecular information, while spatially resolved proteomic platforms can help connect gene-expression patterns with protein abundance. Integrating these datasets can produce a more complete picture of tissue organization and biological function.
- Spatial transcriptomics has major applications in cancer research. Tumours contain malignant cells, immune cells, fibroblasts, endothelial cells, and other populations arranged in complex microenvironments. Spatial mapping can reveal tumour boundaries, immune-cell infiltration, stromal regions, vascular structures, and areas associated with invasion or metastasis. It can also help identify spatially restricted gene-expression signatures and investigate how tumour cells interact with their surrounding microenvironment. In developmental biology, spatial transcriptomics can reveal gene-expression gradients, tissue boundaries, and the organization of cells during organ formation. In neuroscience, it can identify molecular differences between brain regions and investigate the spatial organization of neurons and glial cells. In immunology, it helps examine immune-cell distribution and local signalling within lymphoid organs and inflamed tissues.
- Spatial transcriptomics is also valuable in animal genomics, veterinary research, and animal breeding. It can be used to investigate tissue organization and cellular mechanisms associated with growth, reproduction, immunity, metabolism, and production traits in cattle, sheep, goats, pigs, poultry, and other livestock species. In skeletal muscle, spatial mapping can reveal the distribution of muscle fibres, satellite cells, fibroblasts, endothelial cells, and immune populations involved in muscle development and regeneration. In the mammary gland, it can help characterize epithelial, stromal, immune, and vascular populations associated with mammary development and lactation. In reproductive tissues, it can identify spatially organized cell populations involved in follicular development, spermatogenesis, and embryo development. Comparisons between breeds or environmental conditions may reveal tissue-specific molecular mechanisms relevant to production efficiency, disease resistance, and adaptation.
- Despite its advantages, spatial transcriptomics has several technical and computational limitations. Spatial resolution varies considerably between platforms, and spot-based technologies may combine transcripts from multiple cells. Imaging-based methods can provide higher spatial resolution but may rely on targeted gene panels and have limitations in transcript detection efficiency. Tissue preparation can affect RNA integrity and spatial morphology, while background signals, segmentation errors, and transcript-assignment errors may influence results. Different platforms also generate different data structures, making direct comparisons difficult. In addition, spatial transcriptomics datasets can be large and computationally demanding, requiring suitable storage, processing resources, and analysis workflows.
- Another important limitation is that spatial transcriptomics generally measures RNA abundance rather than direct protein activity or functional signalling. A gene expressed in a tissue region does not necessarily produce an active protein, and the presence of two interacting cell types near one another does not establish that they communicate functionally. Spatial measurements may also be affected by tissue sectioning, limited sampling of anatomical regions, biological variation, and differences in experimental protocols. Validation using histology, immunohistochemistry, immunofluorescence, in situ hybridization, protein measurements, and functional experiments is therefore important. Independent biological samples are also required to distinguish reproducible spatial patterns from sample-specific observations.
- A typical spatial transcriptomics workflow includes tissue collection and preparation, spatial RNA measurement, sequencing or imaging, quality control, spatial coordinate processing, gene-expression analysis, cell-type annotation, spatial clustering, deconvolution, identification of spatially variable genes, neighbourhood analysis, and biological interpretation. Common tools include Space Ranger, Xenium Ranger, Seurat, Scanpy, Squidpy, Giotto, SpatialDE, SPARK, RCTD, cell2location, and Tangram. Results are commonly presented using tissue overlays, spatial feature plots, cell-type maps, spatial heatmaps, neighbourhood graphs, and spatial interaction networks. Reproducible analysis requires recording platform details, tissue-processing protocols, reference datasets, software versions, parameters, and sample-level metadata.
- Overall, spatial transcriptomics and cellular mapping have expanded bioinformatics from the analysis of isolated cells to the investigation of molecular activity within intact biological tissues. By combining gene-expression profiles with spatial coordinates, these approaches reveal how cells are organized, where specific biological processes occur, and how cellular environments influence tissue function. Their integration with single-cell sequencing, spatial proteomics, and computational network analysis is creating new opportunities to study development, disease, immunity, and complex traits. In animal genomics, spatial transcriptomics can help connect molecular mechanisms with tissue organization and economically important traits, supporting a more detailed understanding of biological processes relevant to animal health and genetic improvement.