Spatial Transcriptomics: Principles, Technologies, Workflow, and Applications

Gene expression is organized within tissue architecture: cells occupy defined compartments, respond to local gradients, and interact with neighboring populations. Bulk RNA sequencing measures averaged expression across a specimen, while single-cell RNA sequencing resolves cellular heterogeneity after dissociation but loses the original tissue coordinates. Spatial transcriptomics connects RNA measurements to location, allowing molecular states to be interpreted together with morphology and cellular organization. This guide outlines the core principles, technologies, workflow, applications, and key considerations of spatial transcriptomics, providing researchers with a practical foundation for understanding the field and planning spatially resolved transcriptomic studies.

1. What Is Spatial Transcriptomics?

Spatial transcriptomics is a family of methods that measures gene expression while retaining information about where transcripts, cells, or spatial capture units are located within intact tissue. These approaches link transcriptomic measurements to defined tissue coordinates, allowing gene-expression patterns to be interpreted in relation to their anatomical and morphological context. The resulting spatial maps provide a molecular view of tissue organization and form the basis for investigating how transcriptional programs are distributed across complex tissues (Williams et al., 2022).

This spatial dimension adds information that is lost in conventional transcriptomic measurements. Bulk RNA-seq describes average expression across a sample, while single-cell RNA sequencing resolves transcriptional heterogeneity at the cellular level after cells have been separated from their original tissue architecture. Spatial transcriptomics retains this architecture and connects molecular variation with histological structure. Regional differences in gene expression can then be examined together with local cell composition and tissue morphology, helping researchers identify spatial domains, cellular niches, disease-associated regions, developmental patterns, and other forms of tissue organization that depend on location. Spatial transcriptomics is therefore particularly valuable for biological questions in which tissue structure and local cellular context contribute directly to the observed molecular phenotype.

Comparison of bulk RNA-seq, single-cell RNA-seq, and high-throughput spatial transcriptomics technologies in terms of the profiling resolution (level), data structure, and target discoveries. Reproduced from Jeon et al. (2023), under the Creative Commons Attribution 4.0 International License (CC BY 4.0).

2. Major Spatial Transcriptomics Technologies

A central technical requirement of spatial transcriptomics is to measure gene expression while preserving the spatial origin of RNA signals within tissue. Current spatial transcriptomics technologies achieve this mainly through two approaches: sequencing-based spatial barcoding and imaging-based in situ detection. These two technology families use different strategies to connect RNA measurements with tissue location and form the basis of most current spatial transcriptomics platforms.

2.1 Sequencing-Based Spatial Transcriptomics

Sequencing-based spatial transcriptomics assigns tissue-derived RNA to known coordinates before pooled sequencing. In array-based systems, a tissue section is placed on or transferred to a surface containing spatially indexed capture features. RNA is captured directly or represented through platform-specific probe chemistry, and the resulting cDNA or probe-derived molecules carry both molecular information and a positional barcode. After library construction and sequencing, reads can be mapped back to their original locations to generate a spatial gene-expression matrix. The chemistry and sample workflow differ among platforms, so section preparation, permeabilization, hybridization, reverse transcription, and library construction cannot be treated as one universal protocol.

Representative platforms illustrate how the same principle can be implemented at different scales. Stereo-seq uses DNA nanoball arrays containing coordinate identifiers, molecular identifiers, and poly(T) capture sequences; neighboring array features are spaced at 500 nm, enabling very dense spatial sampling over large tissue areas (Chen et al., 2022). Visium HD uses continuous 2-µm spatial features and provides outputs at multiple bin sizes; larger bins are often used during analysis to increase transcript counts per spatial unit while retaining single-cell-scale context (10x Genomics, 2026). Slide-seqV2 uses barcoded beads at approximately 10-µm spatial resolution and substantially improved RNA capture compared with the original Slide-seq method (Stickels et al., 2021). These examples also show why nominal feature size and biological resolution are not identical: transcript density, capture efficiency, binning, and cell segmentation all influence how finely a tissue can be interpreted.

Representative Sequencing-Based Spatial Transcriptomics Technologies

Technology	Spatial encoding strategy	Transcriptome coverage	Nominal spatial scale	Sample compatibility	Key strengths	Main considerations
10x Genomics Visium HD	RNA is captured on a continuous array of spatially barcoded 2 × 2 μm squares and read by NGS	Whole-transcriptome-scale gene expression	2 μm barcoded squares; data are commonly aggregated or assigned to cells for biological interpretation	FFPE, fresh frozen, and fixed frozen tissue	Standardized workflow; broad transcriptome coverage; direct integration with histology	Transcript counts at the smallest spatial units are sparse, so binning or cell segmentation affects the effective biological resolution
Stereo-seq	mRNA is captured on DNA nanoball arrays carrying coordinate-specific barcodes	Whole-transcriptome profiling	~500 nm center-to-center array spacing	Fresh-frozen and FFPE tissues with assay-specific chemistry	Very high spatial sampling density combined with a large field of view	Nanoscale array spacing does not directly represent the transcriptome of an individual cell; downstream binning or segmentation is generally required
Slide-seqV2	Tissue RNA is captured by randomly positioned DNA-barcoded beads whose coordinates are decoded before analysis	Transcriptome-wide	~10 μm beads	Primarily fresh-frozen tissue sections	Near-cellular spatial sampling with relatively high transcript detection efficiency	Smaller capture area and a more specialized experimental workflow than commercial array platforms
DBiT-seq	Two perpendicular microfluidic barcode flows create unique spatial barcode combinations at channel intersections	Transcriptome-wide; compatible with multimodal extensions	Typically 10–50 μm pixels, depending on channel dimensions	Fixed or cryosectioned tissue in published implementations	Flexible in situ spatial barcoding and compatibility with RNA–protein co-profiling	Spatial resolution and field of view are constrained by microfluidic channel geometry and alignment

Note: Platform characteristics are summarized from published literature and publicly available product information. Specifications for commercial platforms may change with product updates; please refer to the manufacturers’ official websites for the latest information.

2.2 Imaging-Based Spatial Transcriptomics

Imaging-based spatial transcriptomics keeps RNA inside the tissue and identifies transcripts through repeated rounds of probe hybridization, imaging, and barcode decoding or through related in situ sequencing chemistries. Because individual RNA molecules are localized optically, these methods can provide single-cell or subcellular coordinates when tissue morphology and cell segmentation are sufficiently resolved. MERFISH, seqFISH+, Xenium, and CosMx are representative examples. seqFISH+ demonstrated transcriptome-scale imaging of 10,000 selected genes in single cells, illustrating the potential of highly multiplexed in situ detection (Eng et al., 2019). Commercial imaging platforms rely on predefined or customizable probe panels, with some platforms now extending this strategy to whole-transcriptome-scale coverage.

The experimental workflow is therefore fundamentally different from sequencing-based arrays. Fixed sections undergo platform-specific probe hybridization and cyclic imaging, and the primary data consist of decoded transcript coordinates together with morphology or segmentation images. Cell segmentation then determines which transcripts are assigned to each cell. This creates a different source of uncertainty: sequencing-based assays are strongly influenced by capture efficiency and spatial binning, while imaging-based assays depend heavily on probe performance, optical background, transcript decoding, and segmentation quality. Imaging methods are particularly useful when precise localization of a defined gene set is more important than broad transcript discovery.

Representative Imaging-Based Spatial Transcriptomics Technologies

Technology	RNA detection strategy	Gene coverage	Spatial information	Sample compatibility	Key strengths	Main considerations
10x Genomics Xenium	Probe-based in situ detection with iterative fluorescence imaging and transcript decoding	Targeted high-plex panels; current Xenium Prime assays support up to ~5,000 genes	Individual transcript coordinates with cell segmentation and subcellular localization	FFPE and fresh-frozen tissue	High gene plex with single-cell and subcellular spatial information; standardized commercial workflow	Requires predefined gene panels, so the measured transcriptome is determined before the experiment
MERFISH / MERSCOPE	Combinatorial RNA barcoding with error-robust sequential fluorescence imaging	Targeted panels up to approximately 1,000 genes in current MERSCOPE workflows	Single-molecule transcript localization at subcellular resolution	FFPE and fresh-frozen tissue	Direct localization of individual transcripts with error-robust molecular decoding	Panel design and repeated imaging cycles increase experimental and computational complexity
CosMx Spatial Molecular Imager	Highly multiplexed cyclic in situ hybridization and imaging	High-plex targeted panels and whole-transcriptome-scale imaging assays	Single-cell and subcellular transcript localization	Strongly established for FFPE; additional workflows depend on assay configuration	Broad RNA profiling with morphology and optional protein measurements in the same tissue context	Imaging burden, segmentation quality, and transcript density influence data generation and interpretation
seqFISH+	Sequential fluorescence in situ hybridization using combinatorial pseudocolor encoding	Up to 10,000 genes demonstrated experimentally	Individual RNA molecules at subcellular resolution	Primarily research tissue preparations	Very broad in situ gene coverage while retaining single-molecule spatial information	Technically complex research workflow with extensive imaging and image-processing requirements

Note: Platform characteristics are summarized from published literature and publicly available product information. Specifications for commercial platforms may change with product updates; please refer to the manufacturers’ official websites for the latest information.

3. Spatial Transcriptomics Workflow and Data Analysis

A spatial transcriptomics study links molecular measurements to tissue coordinates through a sequence of experimental and computational steps. The exact laboratory workflow depends on the platform: sequencing-based assays encode position through spatial barcodes and generate sequencing libraries, whereas imaging-based assays detect transcripts directly in situ. Despite these differences, both routes converge on a spatial expression dataset that must be aligned with tissue morphology before biological interpretation. The workflow can therefore be understood in two broad stages: generating spatially resolved molecular data and converting those data into interpretable tissue maps (Williams et al., 2022). 

Single cell and spatial transcriptomics workflow. (A) Droplet-based single cell RNA sequencing. (B) Sequencing-based spatial transcriptomics. (C) Image-based spatial transcriptomics. Reproduced from Tchatchoua Ngassam et al. (2026), under the Creative Commons Attribution 4.0 International License (CC BY 4.0).

3.1 Spatial Transcriptomics Data Generation

Generating spatial transcriptomic data begins with tissue preparation that preserves RNA quality and the anatomical structures needed for spatial interpretation. A section is mounted or processed according to the selected assay, and morphology is recorded so molecular measurements can later be registered to the tissue. Sequencing-based platforms then assign positional information to captured RNA or probe-derived molecules through spatial barcodes before library construction and sequencing. Imaging-based platforms use gene-specific probes, repeated hybridization or imaging cycles, and barcode decoding to localize RNA molecules directly within the section. These differences mean there is no single laboratory protocol for spatial transcriptomics; section thickness, fixation, permeabilization, probe chemistry, imaging, and readout are platform-specific. What is shared is the requirement to preserve the relationship between molecular signal and tissue position. The resulting primary data may consist of sequencing reads associated with spatial barcodes or decoded transcript coordinates from images, together with histological or fluorescence images that define tissue structure.

3.2 Spatial Transcriptomics Data Analysis

Spatial transcriptomics data analysis converts these platform-specific readouts into expression patterns that can be interpreted within tissue architecture. Primary processing first assigns molecular signals to coordinates and registers them with the tissue image, producing counts for spots, bins, segmented cells, or individual transcript locations. Quality control then evaluates molecular signal, tissue coverage, background, image registration, and—when relevant—binning or cell segmentation. These choices define the biological unit of analysis: larger bins increase transcript counts but may mix neighboring cell types, while cell-level interpretation depends on reliable segmentation and sufficient molecular information.

Once a spatial expression matrix has been established, downstream analysis asks how gene expression is organized across the tissue. Spatial clustering and spatially variable gene analysis can identify molecular domains, gradients, and boundaries associated with anatomy or pathology. Cell types can be mapped directly in sufficiently resolved data or inferred through integration with scRNA-seq references when spatial units contain mixed populations. Neighborhood analysis then evaluates how cell states or tissue domains are arranged relative to one another, and spatially constrained ligand-receptor analysis can prioritize candidate interactions occurring in plausible local contexts. These analyses provide increasingly biological interpretations of the same spatial framework, but proximity and co-expression remain evidence of spatial association rather than direct proof of functional signaling (Zeng et al., 2022).

4. Key Applications of Spatial Transcriptomics

Spatial transcriptomics provides spatially resolved information on gene-expression programs, cell-type and cell-state distributions, tissue domains, molecular gradients, and local cellular organization. By integrating transcriptomic profiles with tissue architecture, it supports the study of how molecular and cellular patterns relate to biological function, development, and disease. The approach has been widely applied in cancer research, neuroscience, developmental biology, and plant science.

4.1 Cancer and the Tumor Microenvironment

Cancer tissues contain malignant, stromal, vascular, and immune populations whose function depends strongly on spatial organization. Spatial transcriptomics can distinguish tumor core, invasive boundary, stromal compartments, and immune niches while linking each region to gene-expression programs. Xun and colleagues integrated spatial transcriptomics with histology and single-cell data to reconstruct a malignant–boundary–nonmalignant axis in tumor tissues. They identified macrophage and fibroblast subtypes concentrated around the tumor boundary and associated this spatial organization with restricted T-cell infiltration and immune exclusion (Xun et al., 2023). The study shows why spatial profiling adds information beyond cell-type abundance: the location of specific stromal and immune states can reveal tissue structures that may shape access of immune cells to malignant regions.

Spatial Characterization of the Tumor Boundary Microenvironment in Colorectal Cancer. Adapted from Xun et al. (2023), under the Creative Commons Attribution 4.0 International License (CC BY 4.0).

4.2 Neuroscience and Brain Architecture

The brain is organized into layers, nuclei, circuits, and molecularly distinct regions, making spatial context essential for interpreting neuronal and glial diversity. Yao and colleagues combined large-scale scRNA-seq with MERFISH to construct a high-resolution transcriptomic and spatial atlas of the adult mouse brain. The integrated atlas organized thousands of transcriptional clusters within defined brain regions and revealed strong correspondence between molecular cell identity and spatial specificity, together with region-dependent patterns of neurotransmitter and neuropeptide expression (Yao et al., 2023). Such maps provide a framework for understanding how cellular diversity is organized across the brain and how spatially restricted cell states relate to neural function and disease.

Transcriptomic Identity and Spatial Organization of Pallial Glutamatergic Neurons. Adapted from Yao et al. (2023), under CC BY 4.0.

4.3 Development and Organogenesis

Development is a spatial process: cell states change while tissues expand, fold, migrate, and form anatomical boundaries. Spatial transcriptomics can therefore connect developmental gene programs to the structures in which they arise. Using DNA nanoball-patterned Stereo-seq arrays, Chen and colleagues generated a spatiotemporal transcriptomic atlas of mouse organogenesis that mapped cell-type heterogeneity and developmental dynamics across embryonic tissues (Chen et al., 2022). The study also linked spatially resolved developmental programs to windows of vulnerability for developmental disorders. This type of analysis provides information that dissociated single-cell profiles cannot supply alone, because similar transcriptional states can have different developmental roles depending on their position within an emerging organ.

4.4 Plant Development and Physiology

Plant organs contain highly structured cell layers, vascular systems, meristems, and specialized interfaces whose organization is central to development and environmental response. In Lotus japonicus, Stereo-seq was used to map the molecular landscape of root-nodule organogenesis and resolve spatially distinct developmental programs associated with infection and peripheral nodule tissues, providing a spatial framework for understanding symbiotic nitrogen fixation (Ye et al., 2024). In pitaya fruit, integrated single-cell and spatial RNA sequencing linked senescence-related transcriptional changes to exocarp and mesocarp regions and reconstructed the spatial progression of fruit senescence (Li et al., 2024). These studies illustrate how spatial transcriptomics can connect gene regulation to plant anatomy across processes ranging from organ formation and plant–microbe interactions to crop quality and stress biology.

Spatial Transcriptomic Mapping of Lotus japonicus Nodule Development. Adapted from Ye et al. (2024), under CC BY 4.0.

5. Limitations and Interpretation of Spatial Transcriptomics

Spatial transcriptomics adds positional information, but spatial coordinates do not remove measurement bias or establish mechanism by themselves. Effective resolution depends on more than the nominal size of a spot, bead, or array feature; transcript capture, molecular density, binning, optical resolution, and cell segmentation determine how confidently expression can be assigned to individual cells or subcellular compartments.

Sample preparation can also alter RNA quality and morphology, and different tissue chemistries may favor different platforms. Cross-platform comparisons therefore require caution because transcriptome scope, sensitivity, spatial units, and segmentation strategies may differ substantially even when results are described with the same term such as “single-cell resolution.” Replication also remains a sample-level property. Thousands of spots or segmented cells from one section do not replace independent biological specimens, and spatial patterns should be evaluated across replicate tissues when the goal is to infer condition-associated biology rather than to describe one specimen.

Interpretation should also distinguish spatial association from functional interaction. Neighboring cell types, co-localized genes, or ligand–receptor pairs identify biologically plausible relationships, but proximity alone does not demonstrate signaling or causality. Mechanistic conclusions require additional support from replicate tissues, morphology, orthogonal protein or imaging assays, perturbation experiments, or complementary single-cell data. These boundaries are especially important in disease studies, where a visually compelling spatial pattern can reflect tissue composition, technical segmentation, or genuine state-specific biology. A robust spatial study therefore treats tissue architecture as quantitative biological evidence while preserving the same standards of replication and validation required in other omics experiments.

6. Frequently Asked Questions About Spatial Transcriptomics

Does spatial transcriptomics provide single-cell resolution?

Some spatial transcriptomics technologies can achieve single-cell or subcellular spatial measurements, but nominal feature size alone does not determine biological resolution. Transcript density, RNA capture efficiency, optical resolution, binning, and cell segmentation all influence how confidently gene expression can be assigned to individual cells. Resolution should therefore be evaluated according to both the measurement platform and the downstream analytical unit.

How do I choose between sequencing-based and imaging-based spatial transcriptomics?

The choice depends on the biological question, required spatial resolution, transcriptome coverage, tissue type, and study scale. Sequencing-based approaches are commonly used for broad or whole-transcriptome profiling across tissue, while imaging-based approaches provide direct transcript localization at cellular or subcellular scales and may use targeted high-plex or broader gene panels. Sample compatibility and the intended downstream analysis should also be considered when selecting a platform.

Can spatial transcriptomics be integrated with single-cell RNA sequencing?

Yes. Single-cell RNA sequencing provides detailed transcriptional profiles of individual cell types and states, while spatial transcriptomics preserves their location within tissue. Integrating the two datasets can improve cell-type mapping and help resolve the cellular composition of spatial regions containing mixed populations. This combination connects detailed cell-state information with tissue architecture and supports the interpretation of spatial domains, cellular neighborhoods, and disease- or development-associated patterns.