Spatial omics is a family of technologies that measure biological molecules, such as RNA, proteins, or metabolites, while preserving exactly where those molecules sit within a tissue. Traditional methods that analyze gene activity in cells require grinding up a tissue sample first, which means the physical arrangement of cells and their neighbors is destroyed in the process. Spatial omics keeps that map intact, and the result is something closer to a molecular photograph of what is happening inside a tissue rather than a blended average of its contents. That capacity to see where biology is happening, not just what is happening, has made spatial omics one of the fastest-moving areas in biomedical research.
Why Location Matters in Biology
A tumor is not a uniform blob. A brain is not a homogeneous mass. In virtually every organ, different cell types sit next to each other in highly organized patterns, and those arrangements are not decorative. Cells behave differently depending on their neighbors, their distance from a blood vessel, or their position within a tissue layer. When researchers use conventional single-cell sequencing, they can identify thousands of individual cells and catalog which genes each one has switched on. But because the tissue had to be broken apart to isolate those cells, the positional information is lost. A 2024 review in Trends in Plant Science put it plainly: positional information is inherently lost during single-cell RNA sequencing, limiting its potential to characterize complex biological systems, and current analysis pipelines cannot completely recover that spatial information, which confounds biological interpretation.1PubMed. Lost in space: what single-cell RNA sequencing cannot tell you
Spatial omics fills that gap. Instead of asking “which genes are active in this cell?” it asks “which genes are active in this cell, and where is this cell sitting relative to everything else?” That distinction matters enormously for understanding disease. A cancer cell at the invasive edge of a tumor, pressing into surrounding normal tissue, may be running entirely different molecular programs than a cancer cell buried deep in the tumor core. Without spatial context, the two look like random variation. With spatial context, they become a story about how tumors spread.
How the Technologies Work
Spatial omics platforms fall into two broad camps: those based on imaging and those based on sequencing. Each has trade-offs, and researchers often pick between them depending on whether they need to see many genes at once or need the finest possible resolution.
Imaging-based methods work by labeling RNA molecules directly in a tissue section so they light up under a microscope. MERFISH (multiplexed error-robust fluorescence in situ hybridization) is one of the best-known approaches. It uses a combinatorial labeling scheme: each RNA species gets tagged with a unique pattern of fluorescent signals read out over several rounds of imaging. Early demonstrations showed that MERFISH could identify and locate hundreds to thousands of distinct RNA species in individual cells while preserving their native spatial context.2PubMed Central. Multiplexed imaging of high-density libraries of RNAs with MERFISH and expansion microscopy The method has since been extended to achieve transcriptome-scale quantification, measuring RNA across the entire gene catalog within single cells.3PubMed Central. Spatial transcriptome profiling by MERFISH reveals subcellular RNA compartmentalization and cell cycle-dependent gene expression
Sequencing-based methods take a different approach. Rather than looking at RNA through a microscope, they capture RNA onto a surface using spatially encoded DNA barcodes and then sequence everything. One technique, called DBiT-seq (deterministic barcoding in tissue), uses two sets of tiny microfluidic channels pressed onto a tissue slice at right angles. The first set lays down one barcode, the second lays down a perpendicular barcode, and together they create a grid of uniquely identified spots, each just 50 micrometers wide.4Nature Communications. Spatial transcriptomics using multiplexed deterministic barcoding in tissue When the captured RNA is sequenced, every transcript carries a coordinate telling researchers exactly where it came from on the tissue.
Imaging methods generally offer finer spatial resolution, sometimes down to the subcellular level, and let you see individual molecules. Sequencing methods tend to capture a broader sweep of genes without needing to preselect targets. Researchers still face a fundamental trade-off among gene throughput, spatial resolution, tissue area, and sensitivity, and no single platform excels at all of these simultaneously.5Computational and Structural Biotechnology Journal. A guidebook of spatial transcriptomic technologies, data resources and analysis approaches
Beyond RNA
Most of the early excitement around spatial omics focused on RNA because measuring gene expression is a natural extension of single-cell transcriptomics. But cells do not run on RNA alone. Proteins carry out most of the work, and small molecules called metabolites are the currency of cellular chemistry. Spatial omics has expanded to map these layers too.
Spatial metabolomics uses imaging mass spectrometry to visualize the distribution of metabolites, lipids, and drugs directly in tissue sections. What was considered impossible two decades ago is now routine enough to support research applications across life sciences and biomedicine.6PubMed Central. Spatial Metabolomics and Imaging Mass Spectrometry in the Age of Artificial Intelligence In cancer research, mass spectrometry imaging has been used to visualize metabolic reprogramming within specific tissue structures, helping to identify biomarkers tied to tumor development.7PubMed Central. Advances in mass spectrometry imaging for spatial cancer metabolomics A study of human kidney tissue, for instance, used high-resolution mass spectrometry imaging at 20-micrometer resolution to map over 40 low-molecular-weight metabolites, showing how individual compounds localize to distinct kidney regions like the cortex, medulla, and renal pelvis.8PubMed. Spatial Metabolomics of the Human Kidney using MALDI Trapped Ion Mobility Imaging Mass Spectrometry
Spatial proteomics has been harder to crack because proteins cannot be amplified the way RNA can. A recent platform called PLATO combines microfluidics with deep learning to map thousands of proteins across whole tissue sections. The approach uses microchannels to digest proteins on the tissue surface at different angles, then feeds the resulting data into a transfer learning algorithm that predicts protein distributions using reference data from neighboring tissue slices.9Cell. High-resolution spatial proteomics across entire tissues using parallel-flow projection and transfer learning The ability to combine RNA, protein, and metabolite maps of the same tissue region is where spatial omics becomes more than a sum of its parts.
Layering Multiple Molecular Views
One of the most active frontiers in the field is spatial multi-omics: measuring several molecular layers from the same piece of tissue at once. A technology called spatial-Mux-seq can simultaneously profile five different modalities from a single tissue section, including two types of histone modifications (which control how tightly DNA is packaged), chromatin accessibility, the whole transcriptome, and a panel of surface proteins.10PubMed Central. Multiplexed spatial mapping of chromatin features, transcriptome, and proteins in tissues That means researchers can ask not just which genes a cell is expressing but also which regions of its genome are open for business, which proteins decorate its surface, and how the epigenetic landscape varies from one tissue zone to another.
Other workflows take a sequential approach. One recently described pipeline maps hundreds of genes using a spatial transcriptomics platform and then profiles more than 40 protein markers on the very same tissue section using a different imaging technology.11PubMed Central. An Integrated Spatial Multi-Omics Workflow for Sequential RNA and Protein Profiling in FFPE Tumor Tissue Combining RNA and protein readouts from the same cells helps cross-validate findings and can reveal discrepancies between what a cell’s genes are saying and what its protein machinery is actually doing.
Cancer Research and the Tumor Microenvironment
Cancer has been one of the most productive proving grounds for spatial omics, largely because tumors are so spatially heterogeneous. The immune cells, blood vessels, connective tissue, and cancer cells within a tumor form a complex ecology called the tumor microenvironment, and how these components are arranged in space often determines whether a patient responds to treatment.
Spatial multi-omics technologies have been used to study transcriptomes, proteomes, and metabolomes of tumor-immune microenvironments across several cancer types, yielding markers of cancer progression.12PubMed Central. Spatial multi-omics analyses of the tumor immune microenvironment A concrete example comes from pancreatic cancer. Using spatial transcriptomics combined with single-nucleus RNA sequencing, researchers found that the leading edge of a primary pancreatic tumor fosters an immunosuppressive environment by accumulating regulatory T cells, a specific type of immune cell that dampens the anti-tumor response. Compared to liver metastases from the same cancer, the primary tumor front showed a distinctly different immune landscape.13PubMed Central. Single-nucleus RNA sequencing and spatial transcriptomics reveal an immunosuppressive tumor microenvironment related to metastatic dissemination during pancreatic cancer liver metastasis Findings like these matter because they suggest that location-specific immune evasion could be a target for therapy, and they would be invisible without spatial data.
Mapping the Brain
The brain is perhaps the organ that benefits most from spatial context. It contains hundreds of distinct cell types arranged in intricate architectures, and the function of a neuron depends heavily on where it lives and what it connects to. Spatial transcriptomics has been used to build functional atlases of the brain, pulling the properties of neural circuits into sharper focus.14Experimental & Molecular Medicine. Spatial transcriptomics in neuroscience
A whole-mouse-brain atlas generated using Stereo-seq and single-nucleus RNA sequencing cataloged over four million cells, assigning them to 308 distinct cell clusters, all with spatial coordinates. The spatial transcriptomic information allowed researchers to define fine anatomical boundaries in brain regions that had not been clearly delineated before.15Neuron. Single-cell spatial transcriptomic atlas of the whole mouse brain This kind of atlas is not just an academic exercise; it provides a reference framework that other researchers can use to pinpoint where disease-associated changes occur, whether in neurodegeneration, psychiatric illness, or brain injury.
Watching an Embryo Take Shape
Embryonic development is another area where spatial information is essential. An embryo transforms from a cluster of nearly identical cells into a body with distinct organs in a matter of days. Understanding how that happens requires knowing which genes switch on where and when. Spatial transcriptomics at the scale of entire embryos has made this possible in ways that were previously unachievable.
The Mouse Organogenesis Spatiotemporal Transcriptomic Atlas (MOSTA) used Stereo-seq to profile 53 sagittal sections from mouse embryos spanning eight days of development, from the onset of organ formation through mid-gestation.16Cell. Stereo-seq and spatial enhanced resolution omics-sequencing of mouse organogenesis Another study used Slide-seq to create three-dimensional spatiotemporal maps of mouse embryos at the very beginning of organogenesis, capturing gene expression dynamics as the body plan was first being laid down.17PubMed Central. Spatiotemporal transcriptomic maps of whole mouse embryos at the onset of organogenesis These datasets are essentially molecular movies of development, showing researchers how gene expression patterns shift across space and time as tissues and organs emerge.
Tracking Host-Pathogen Battles in Tissue
When a pathogen invades a tissue, the host response is not uniform. Immune cells rush to certain zones, the pathogen may cluster in specific niches, and the molecular dialogue between attacker and defender varies from spot to spot. Conventional methods that average gene expression across a whole tissue sample miss this spatial drama entirely. Spatial omics has lowered the barrier to evaluating transcript and protein expression within the subcellular niche, primary cell, or native tissue environment during infection.18PubMed Central. Microbial Pathogenesis in the Era of Spatial Omics
In one striking application, researchers developed a dual spatial transcriptomics approach that captures both the host and pathogen transcriptomes simultaneously in preserved human tissue sections. Applied to lung samples from COVID-19 patients, the method spatially detected SARS-CoV-2 RNA and identified host genes whose expression correlated with the virus’s location, pointing to putative modulators of infection.19PubMed Central. Dual spatially resolved transcriptomics for human host-pathogen colocalization studies in FFPE tissue sections A separate study used spatial metatranscriptomics to resolve interactions among a host plant, bacteria, and fungi all at once. By mapping the gradient of bacterial molecules spreading from an infection site, the researchers discovered that a key immune-response gene tracked the pathogen’s spatial distribution far more precisely than traditional fluorescence imaging could detect.20Nature Biotechnology. Spatial metatranscriptomics resolves host–bacteria–fungi interactomes
The Computational Side
Generating spatial omics data is only half the challenge. Making sense of it requires substantial computational effort, and several problems that do not exist in traditional genomics arise when location enters the picture.
One major challenge is cell-type deconvolution. Many sequencing-based spatial platforms do not have true single-cell resolution; each measurement spot captures RNA from a small cluster of cells rather than one cell. Figuring out which cell types contributed to each spot, and in what proportions, requires algorithms that combine spatial expression data with reference information from single-cell sequencing. Methods like CARD and QR-SIDE address this by borrowing information across neighboring spots, using the fact that cells of the same type tend to cluster together in tissue.21PubMed Central. Robust Spatial Cell-Type Deconvolution with Qualitative Reference for Spatial Transcriptomics When no perfectly matched reference dataset exists, or when batch effects distort the data, these methods adaptively adjust to reduce bias.
Another frontier is cell-cell communication inference. Cells constantly signal to each other through ligand-receptor pairs, molecules sent by one cell that dock onto receptors on another. Spatial transcriptomics makes it possible to ask which cells are close enough to actually communicate and whether the required ligands and receptors are being expressed in the right places. Tools like SpaTalk use a knowledge-graph approach, first building a spatial network of neighboring cells, then filtering for ligand-receptor pairs that are significantly co-expressed, and finally tracing the intracellular signaling cascade downstream from the receptor.22Nature Communications. Knowledge-graph-based cell-cell communication inference for spatially resolved transcriptomic data with SpaTalk SpatialDM takes a statistical approach, using spatial correlation metrics to detect spots where ligand and receptor genes are co-expressed beyond what chance would predict.23Nature Communications. SpatialDM for rapid identification of spatially co-expressed ligand–receptor and revealing cell–cell communication patterns
Moving Toward Clinical Use
For all its scientific power, spatial omics has remained mostly a research tool. Translating it into clinical practice means overcoming several practical hurdles. Cost is one: a single spatial transcriptomics experiment can be expensive enough to limit routine use. Turnaround time is another; some protocols require days of processing and imaging. And standardization remains limited, with different platforms producing data that are not always directly comparable. A recent benchmarking study of three commercial imaging-based platforms found that while all three could perform spatially resolved cell typing, they differed in transcript capture rates, false-discovery rates, and cell segmentation errors.24Nature Communications. Systematic benchmarking of imaging spatial transcriptomics platforms in FFPE tissues
Despite these barriers, the trajectory is clearly toward clinical relevance. Spatial omics has profound implications for differential diagnosis, treatment guidelines, risk assessment, and understanding disease mechanisms.25Life Medicine. Spatiotemporal Omics-Refining the landscape of precision medicine The idea is that spatial biomarkers, markers defined not just by their molecular identity but by where they appear in a tissue, could outperform conventional biomarkers for tasks like predicting which patients will respond to immunotherapy or staging a disease more precisely. These are still early days, and independently validating spatial biomarkers in large patient cohorts is an ongoing effort, but the direction of travel is unmistakable.
Unlocking Archived Tissue Banks
One development that has quietly expanded the reach of spatial omics is the ability to work with formalin-fixed, paraffin-embedded (FFPE) tissue. Hospitals have been archiving tissue samples this way for over a century. These blocks of preserved tissue are the backbone of clinical pathology, and millions of them sit in biobanks worldwide, many linked to detailed patient records and long-term outcomes data. The problem was that the chemical fixation process damages RNA, making molecular profiling difficult.
A method called Patho-DBiT was designed specifically to handle these archived samples. It combines an in-tissue step that adds molecular tails to damaged RNA with a computational pipeline tailored to FFPE artifacts, enabling whole-transcriptome spatial sequencing. The approach has been validated on clinical tumor samples stored for five years, and it goes beyond simple gene expression by co-profiling RNA processing events like region-specific splicing.26PubMed Central. Spatially exploring RNA biology in archival formalin-fixed paraffin-embedded tissues This matters because it means spatial omics does not depend solely on fresh tissue collected prospectively. Researchers can go back to decades-old samples and ask spatial questions that nobody could have posed when those samples were first taken.
Building Three-Dimensional Maps
Most spatial omics experiments produce two-dimensional snapshots, one thin tissue slice at a time. But organs are three-dimensional, and assembling a series of flat slices into a coherent 3D volume introduces its own challenges. Slices can warp during processing, alignment between sections is imperfect, and the gap between consecutive slices means spatial information along the depth axis is sparse.
Two recently published tools tackle this problem from different angles. SINTER3D uses an approach borrowed from computer graphics called implicit neural representations, treating gene expression as a continuous function of 3D coordinates. This allows it to interpolate gene expression between slices, generate virtual sections at arbitrary depths, and identify spatial domains in three dimensions.27PubMed Central. SINTER3D: continuous 3D reconstruction of spatial transcriptomics via implicit neural representations STAIR takes a graph-based approach, using attention mechanisms at both the single-spot and whole-slice level to align and integrate multiple slices into a unified 3D atlas. It can also stitch newly generated slices into an existing atlas, making it practical for growing datasets over time.28PubMed Central. Spatial transcriptomic alignment, integration, and 3D reconstruction by STAIR Reconstructing tissues in 3D adds a layer of biological insight that flat maps miss, revealing features like how a tumor’s immune border wraps around its entire surface rather than just what it looks like in cross-section.
Applications in Plant Biology
Spatial omics is not limited to animal tissues and clinical samples. Plant biology has increasingly adopted these tools, and for good reason. Plant tissues have rigid cell walls that make it harder to dissociate individual cells for conventional single-cell analysis, which means spatial approaches that work directly on intact tissue sections are especially valuable. Comprehensive cell atlases have already been generated in several model plant species using single-cell and spatial transcriptomics, creating reference platforms for understanding how different cell types are arranged within roots, leaves, and other organs.29PubMed Central. Exploring the untapped potential of single-cell and spatial omics in plant biology Understanding the molecular and cellular dynamics of plant tissues at spatial resolution could eventually feed into agricultural applications, from breeding crops that allocate resources more efficiently within their tissues to understanding how plants defend themselves against pathogens at the point of infection.