Spatial biology is the study of how molecules, cells, and structures are organized within tissues, and that physical arrangement turns out to be central to how the body works, how diseases develop, and how treatments succeed or fail. Traditional methods of analyzing tissue have required grinding it up first, which reveals what genes or proteins are active but destroys any record of where they were active. Spatial technologies preserve that “where,” letting researchers see gene expression, protein levels, and even chromatin architecture mapped onto the actual tissue in place. The field has moved fast enough that it is already reshaping cancer immunotherapy, brain mapping, and infectious-disease research, with clinical pathology not far behind.
The Problem With Blending Tissue Into Soup
For decades, the workhorse of molecular biology was bulk sequencing: take a tissue sample, break it apart, and measure the average gene activity across millions of cells. This approach is powerful for detecting broad signals but blind to the fact that different cells sitting side by side can be doing very different things. Single-cell sequencing improved the picture dramatically by reading individual cells one at a time, revealing that tumors, for instance, contain dozens of cell subtypes with distinct molecular profiles. But single-cell methods still require physically separating cells from the tissue, which strips away all information about which cells were neighbors, which were clustered together, and how they were positioned relative to blood vessels, immune structures, or disease foci.
Both bulk and single-cell RNA sequencing lose spatial information in the process of dissociating tissue.
1PubMed Central. From bulk RNA sequencing to spatial transcriptomics: a comparative review of differential gene expression analysis methodsThat loss matters because biology is not random. Cells communicate with their immediate neighbors through signaling molecules that diffuse only short distances. Immune cells patrol specific zones. Stem cells live in defined niches. A tumor’s ability to escape the immune system depends on which immune cells are nearby and how they are physically arranged. Averaging all of that into a single readout is like trying to understand a city’s traffic flow by counting the total number of cars without knowing where the roads are.
How Spatial Technologies Capture Location
Spatial biology tools fall into two broad families, each with strengths that suit different research questions. The choice between them involves a genuine trade-off between how many genes you can measure, how fine the resolution is, how large an area you can cover, and how practical the experiment is to run.
2Computational and Structural Biotechnology Journal. A guidebook of spatial transcriptomic technologies, data resources and analysis approachesImaging-Based Methods
The first family works by labeling individual RNA molecules inside intact tissue with fluorescent probes, then imaging them under a microscope. Technologies like MERFISH and seqFISH use clever barcoding schemes: each gene is assigned a unique pattern of colors across multiple rounds of imaging, so that even with only a handful of fluorescent dyes, thousands of genes can be distinguished by their color sequence. MERFISH adds an error-correction layer borrowed from information theory, using extra bits to catch and fix misidentifications.
3PubMed Central. Introduction to bioimaging‐based spatial multi‐omic novel methodsThese imaging methods achieve single-molecule sensitivity and subcellular resolution, meaning researchers can see not just which cell expresses a gene but where inside the cell the RNA sits. A recent advance called RT&T-AMP, integrated with MERFISH, pushed the throughput to roughly 33,000 distinct RNAs, including about 23,000 genes and 10,000 transcript variants, imaged in mouse brain tissue.
4Cell. Whole-transcriptome spatial and isoform-resolved transcriptomics in tissues by in situ RNA amplificationSequencing-Based Methods
The second family captures RNA by placing tissue on a surface studded with spatially barcoded oligonucleotides, short DNA sequences that grab nearby RNA and tag it with a location code. The tissue is then sequenced as usual, but the barcodes let you map each read back to its position on the tissue. Platforms like Visium (from 10x Genomics) and Slide-seq belong to this camp. Slide-seqV2, for example, improved its RNA capture efficiency to roughly half that of droplet-based single-cell methods, about a tenfold jump over its predecessor, making it sensitive enough to detect mRNAs localized to specific parts of neurons in the mouse hippocampus.
5PubMed Central. Highly sensitive spatial transcriptomics at near-cellular resolution with Slide-seqV2Sequencing-based approaches naturally capture the whole transcriptome without requiring a pre-selected gene panel, but their spatial resolution has historically been coarser. Recent open-source tools like Open-ST have pushed sequencing-based resolution to the subcellular level, accurately separating nuclear-retained transcripts from cytoplasmic ones within individual cells.
6Cell. Open-ST high-resolution spatial transcriptomics in 3DWhy Cancer Research Adopted Spatial Biology First
Tumors are not uniform masses of identical cancer cells. They are ecosystems containing cancer cells, immune cells, blood vessels, fibroblasts, and signaling molecules, all organized in ways that determine whether the immune system can attack the tumor or whether the tumor escapes. Understanding that spatial organization has become especially urgent in the era of immunotherapy, where checkpoint inhibitor drugs work by releasing the brakes on the patient’s own immune cells. Whether those drugs work for a given patient depends heavily on how immune cells are positioned within and around the tumor.
7PubMed Central. Decoding the tumor microenvironment with spatial technologiesSpatial technologies are now being used to develop biomarkers that predict which patients will respond to immunotherapy. In non-small cell lung cancer, a study of 234 patients used spatial proteomics and spatial transcriptomics to profile the tumor immune microenvironment. The researchers identified a resistance signature, including proliferating tumor cells, granulocytes, and blood vessels, and a response signature dominated by macrophages and CD4 T cells. Translating these spatial signatures into gene-level predictors produced hazard ratios that were consistent across three independent patient groups.
8Nature Genetics. Spatial signatures for predicting immunotherapy outcomes using multi-omics in non-small cell lung cancerSimilar spatial biomarker work in triple-negative breast cancer found that the physical proximity of specific immune cell subtypes to cancer cells early during treatment was a strong predictor of whether checkpoint inhibitors would work, with proliferating CD8+ T cells and cancer cells expressing certain surface markers emerging as dominant predictors of response.
9Nature. Spatial predictors of immunotherapy response in triple-negative breast cancerThe clinical promise here is straightforward: rather than giving an expensive therapy to every patient and waiting to see who responds, spatial biomarkers could guide treatment selection upfront.
10PubMed Central. The current landscape of spatial biomarkers for prediction of response to immune checkpoint inhibitionMapping the Brain at Cellular Resolution
The brain is arguably the organ that benefits most from spatial approaches, because its function depends so directly on the precise arrangement of cells. Neurons in adjacent cortical layers express different genes and perform different computations, and understanding that architecture requires knowing which cells sit where. Spatial transcriptomics applied to the human cortex mapped nearly 60,000 cells into 75 previously defined subtypes, creating a detailed atlas of cell types in their native positions. Excitatory neurons showed strong layer-specific distributions that matched predicted locations, while inhibitory neurons were more broadly scattered across layers.
11Communications Biology. Comprehensive in situ mapping of human cortical transcriptomic cell typesIn the developing brain, spatial methods have mapped the embryonic mouse brain well enough that unsupervised analysis faithfully recovered the major anatomical regions, including distinct zones where excitatory and inhibitory neurons are born.
12Biology Open. Spatial transcriptomics map of the embryonic mouse brain – a tool to explore neurogenesisThese atlases are not just academic exercises. They serve as reference maps for studying neurological disease, drug effects, and brain development at a resolution that was previously available only through painstaking manual dissection and staining.
Embryonic Development and Body Patterning
During embryonic development, cells make fate decisions based on signals from their neighbors, and those signals vary depending on position along the body’s axes. Spatial transcriptomics can capture this process at scale. A method called sci-Space profiled about 120,000 nuclei from developing mouse embryos, identifying thousands of genes with anatomically patterned expression and tracking how differentiating neurons migrate through space as they mature.
13PubMed Central. Embryo-scale, single-cell spatial transcriptomicsWhole-embryo imaging in zebrafish has taken this further, mapping 495 genes at subcellular resolution from the earliest stages of body formation through organ development. One finding that would have been invisible without spatial data: sharp boundaries between gene expression domains in the embryo develop through changes in gene expression within cells rather than through physical sorting of cell populations.
14PubMed. Whole-embryo spatial transcriptomics at subcellular resolution from gastrulation to organogenesisInfectious Disease and Host-Pathogen Geography
When a pathogen infects tissue, the host response is not uniform. Some cells mount an inflammatory defense, others change their metabolism, and still others remain relatively unaffected, all depending on how close they are to the site of infection. Spatial transcriptomics has made it possible to map these local responses directly. In malaria-infected mouse livers, spatial profiling revealed that lipid metabolism changed in cells near parasites, that different zones of the liver lobule activated distinct inflammation programs, and that certain regions became “inflammatory hotspots” enriched with specific immune cell types.
15PubMed Central. Host-pathogen interactions in the Plasmodium-infected mouse liver at spatial and single-cell resolutionIn human COVID-19 lung tissue, a dual spatial transcriptomics approach simultaneously captured both the host and the SARS-CoV-2 transcriptomes from the same tissue section, enabling researchers to identify which host genes were active specifically in cells colocalized with viral RNA.
16PubMed Central. Dual spatially resolved transcriptomics for human host-pathogen colocalization studies in FFPE tissue sectionsThat kind of analysis was essentially impossible with conventional methods, which could tell you that a certain gene was upregulated in an infected lung but not whether the upregulation happened in cells sitting right next to the virus or in cells far from it.
Autoimmune Disease and Tissue Remodeling
Chronic autoimmune conditions like rheumatoid arthritis involve localized immune cell infiltration and tissue destruction, processes that spatial methods are uniquely suited to study. Spatial transcriptomics of rheumatoid arthritis synovium has mapped where different types of infiltrating immune cells gather relative to organized lymphoid structures in the joint lining.
17Communications Biology. Three-dimensional spatial transcriptomics uncovers cell type localizations in the human rheumatoid arthritis synoviumMore recently, deep spatial profiling of synovial tissue identified a network of tissue-resident macrophages, located within perivascular niches alongside fibroblasts and blood vessel cells, that appears to play a protective, anti-inflammatory role. This network was disrupted in active disease but restored in patients who responded well to standard therapy, with specific cell-cell interactions and molecular pathways mediating the restoration.
18Annals of the Rheumatic Diseases. Spatial mapping of rheumatoid arthritis synovial niches reveals a LYVE1+ macrophage network associated with response to therapyFindings like these point toward using spatial tissue profiling not just for research but eventually to monitor treatment response and guide therapy choices in autoimmune disease.
19PubMed Central. Spatial transcriptomics in autoimmune rheumatic disease: potential clinical applications and perspectivesBeyond RNA: Spatial Proteomics and Multi-Omics
Spatial biology is not limited to measuring RNA. Spatial proteomics visualizes and quantifies protein expression within tissues at single-cell resolution.
20PubMed Central. scProAtlas: an atlas of multiplexed single-cell spatial proteomics imaging in human tissuesSince proteins are the molecules that actually carry out most cellular functions, measuring them in place adds a layer of information that RNA alone cannot provide. High-plex spatial protein panels can now profile hundreds of proteins simultaneously: one study of head and neck cancer used a 580-protein panel alongside an 18,000-gene spatial transcriptome panel, both applied to the same tissue samples.
21npj precision oncology. The development of a high-plex spatial proteomic methodology for the characterisation of the head and neck tumour microenvironmentEven further upstream, spatial epigenomics methods now map chromatin accessibility, the open-versus-closed state of DNA that controls which genes can be turned on, in tissue sections. A method called SPACE-seq captures chromatin accessibility, gene expression, and cell lineage information from the same tissue section, connecting the regulatory layer of the genome to its functional output in space.
22PubMed Central. Unified molecular approach for spatial epigenome, transcriptome, and cell lineagesThe most ambitious experiments combine multiple data types from the exact same tissue section. Integrating spatial transcriptomics and spatial proteomics from the same slice of human lung cancer tissue, registered computationally so that coordinates align, allows researchers to ask questions that neither modality could answer alone.
23PubMed Central. An integrated approach for analyzing spatially resolved multi-omics datasets from the same tissue sectionGoing Subcellular and Three-Dimensional
The resolution frontier keeps advancing. Several recent methods achieve spatial transcriptomic profiling at the subcellular level, meaning they can distinguish RNA in the nucleus from RNA in the cytoplasm or in specific organelles like mitochondria and stress granules. PHOTON, for example, captures the transcriptome of specific subcellular compartments, including nucleoli and mitochondria, within intact tissue.
24Nature Communications. Subcellular level spatial transcriptomics with PHOTONThe push into three dimensions is equally active. A deep-tissue imaging method achieved whole-embryo transcriptomic imaging across a specimen depth of about 310 micrometers, and when combined with expansion microscopy, it revealed an intricate network of ten subcellular structures within individual cells.
25PubMed Central. Deep-tissue transcriptomics and subcellular imaging at high spatial resolutionThree-dimensional reconstruction is valuable because tissues are not flat. A two-dimensional slice of a tumor might show immune cells near cancer cells, but a 3D reconstruction can reveal whether those immune cells are actually infiltrating the tumor mass or sitting on its surface.
The Computational Bottleneck
Spatial biology generates staggering amounts of data. A single MERFISH experiment might produce images of millions of individual RNA molecules across a tissue section, and each one needs to be assigned to the correct cell. That cell segmentation step, deciding which transcripts belong to which cell, is one of the biggest computational challenges in the field. Deep learning models, including transformer-based architectures and foundation models, have shown strong results for this task, though they require substantial training data and computing power.
26PubMed Central. Transforming subcellular spatial transcriptomics: deep learning models for cell segmentationSeveral tools are tackling this from different angles. Bering uses graph deep learning to jointly segment cells and annotate transcript types in both 2D and 3D data, with pre-trained models that can transfer to new tissues.
27PubMed Central. Bering: joint cell segmentation and annotation for spatial transcriptomics with transferred graph embeddingsQuality control also requires rethinking. Standard QC methods borrowed from single-cell sequencing do not account for the fact that gene expression varies naturally across space, so a “low-quality” spot might actually just be in a region of tissue with lower cell density. SpotSweeper was developed specifically as a spatially aware QC tool that uses local neighborhoods to distinguish genuine outliers from expected spatial variation.
28Nature Methods. SpotSweeper: spatially aware quality control for spatial transcriptomicsAnother active area is modeling cell-cell communication from spatial data. Because spatial transcriptomics tells you which cells are close to each other and what signaling molecules they express, computational tools can now infer directed communication networks, predicting which cells are sending signals, which are receiving them, and how distance affects the interaction.
29PubMed Central. Reconstructing cell-cell interaction network in single-cell spatial transcriptomics via directed heterogeneous graph autoencoderMoving Toward the Clinic
For spatial biology to influence patient care, it needs to work with the samples hospitals already have. Most diagnostic tissue samples are formalin-fixed and paraffin-embedded (FFPE), a preservation method that has been standard in pathology for over a century. FFPE tissue is durable and widely archived, but the fixation process degrades and fragments RNA, making it harder to profile. Recent platforms have been specifically designed to handle FFPE material. Patho-DBiT, for example, uses in situ polyadenylation to recover whole-transcriptome spatial data from FFPE sections.
30Cell. Patho-DBiT enables spatial whole transcriptome sequencing and multi-omic profiling in clinical FFPE tissuesFFPE compatibility opens the door to retrospective studies using tissue banks that hospitals have accumulated over decades, connecting spatial molecular data to long-term patient outcomes that are already recorded in medical charts.
31PubMed Central. Integrating spatial omics with routine haematoxylin and eosin in formalin-fixed paraffin-embedded: a step-by-step clinical workflowWorkflows are being developed to integrate spatial omics with routine histology staining on the same section, so pathologists can overlay molecular maps onto the tissue morphology they already use for diagnosis.
Cost, Access, and the Open-Source Question
Spatial biology experiments remain expensive and technically demanding. Proprietary systems and reagents create barriers that limit who can run these experiments and who can reproduce the results. The risk of monopoly-driven pricing is a real concern in a field where the technology is advancing faster than the market can keep up.
32PubMed Central. Enablers and challenges of spatial omics, a melting pot of technologiesOpen-source efforts are pushing back against this. Open-ST, for example, was designed as an open-source experimental and computational platform for subcellular spatial transcriptomics, lowering the entry barrier for labs that cannot afford or do not want to depend on proprietary systems.
6Cell. Open-ST high-resolution spatial transcriptomics in 3DBenchmarking across platforms is also gaining attention: a recent comparison of spatial transcriptomics technologies across six cancer types quantified the trade-offs in resolution versus tissue context, helping labs choose the right tool for their specific question rather than defaulting to the most marketed one.
33PubMed Central. A technical comparison of spatial transcriptomics platforms across six cancer typesPlants, Agriculture, and the World Beyond Mammals
Most spatial biology headlines focus on human and mouse tissue, but the field is expanding into plant biology and agriculture. Plants present unique technical challenges: thick cell walls make tissue dissociation difficult, and the antibody reagents common in animal research are far less developed for plant systems. Despite these hurdles, single-cell and spatial transcriptomics have been adopted in plant biology, providing new views of how tissues develop and respond to environmental stress. Building more complete reference atlases for crop species and developing better computational tools tailored to plant data remain active priorities.
34PubMed Central. Exploring the untapped potential of single-cell and spatial omics in plant biologyFor agriculture, the practical implications are real. Understanding how gene expression is spatially organized in roots, leaves, or developing seeds could inform breeding strategies, disease resistance, and crop optimization in ways that bulk sequencing cannot. The technology is still early for plant systems, but it represents one of the clearest paths for spatial biology to affect fields well beyond medicine.