Spatial Transcriptomics Study Design: Key Considerations for Robust Results

Spatial transcriptomics preserves the in situ spatial context of RNA molecules within intact tissues, enabling localization of cell types and their associated gene expression patterns. Unlike conventional transcriptome profiling that loses spatial information during tissue homogenization or dissociation, spatial transcriptomics methods capture where genes are expressed, also what is expressed. This article outlines the critical design decisions for spatial transcriptomics experiments, including sample preparation, platform selection, replication strategy, controls, and validation approaches. The practical outcome is a study design checklist that helps researchers plan experiments to avoid common pitfalls and ensure statistical power.

At a Glance

Defining the Biological Question Before Platform Selection

The first decision in spatial transcriptomics study design is not technical but biological. Researchers must specify whether the question requires single-cell resolution, subcellular resolution, or spot-level resolution at the scale of several cells. This decision determines which platforms are appropriate and how many samples are needed.

Spatial transcriptomics methods differ fundamentally in their resolution and throughput. Sequencing-based approaches capture gene expression from spatially barcoded spots that may contain multiple cells, while imaging-based approaches such as Xenium, MERSCOPE, and CosMx can resolve transcripts at single-cell or subcellular resolution. A systematic benchmarking study of imaging spatial transcriptomics platforms in formalin-fixed paraffin-embedded tissues found that these platforms can recover cell-to-cell interactions, groups of spatially covarying genes, and gene signatures associated with pathological features. The same study reported that Xenium consistently generated higher transcript counts per gene without sacrificing specificity, and that Xenium and CosMx measured RNA transcripts in concordance with orthogonal single-cell transcriptomics. All three platforms performed spatially resolved cell typing with varying degrees of sub-clustering capabilities, with Xenium and CosMx finding slightly more clusters than MERSCOPE, albeit with different false discovery rates and cell segmentation error frequencies.

For researchers working with precious clinical samples, the choice between imaging and sequencing platforms carries practical consequences. Imaging platforms typically measure a predefined set of genes, which constrains discovery but enables higher resolution. Sequencing platforms capture the whole transcriptome but at lower spatial resolution. A practical guide informed by the processing and analysis of over 1000 spatial samples across multiple platforms emphasizes that platform selection, sample quality, and experimental scalability are common barriers to implementation. The guide recommends that researchers at all levels, from those designing their first spatial experiment to groups integrating spatial transcriptomics into large-scale studies, prioritize tissue handling and computational analysis workflows.

Sample Preparation and Tissue Quality

Fresh Frozen Versus Formalin-Fixed Paraffin-Embedded Tissue

Tissue preservation is the most consequential pre-analytical variable in spatial transcriptomics. RNA quality proves to be a critical factor for the reliability of analyses, especially in formalin-fixed paraffin-embedded and postmortem samples. Standardized sample preparation and RNA quality control are essential for valid results.

Fresh frozen tissue generally preserves RNA integrity better than FFPE tissue, but FFPE tissue is the standard format for clinical pathology archives. The choice between these formats affects platform compatibility. Imaging-based spatial transcriptomics platforms are particularly well-suited for FFPE tissues, as demonstrated in the benchmarking study of Xenium, MERSCOPE, and CosMx on tissue microarrays containing 17 tumor and 16 normal tissue types. For sequencing-based platforms, FFPE compatibility varies by product generation, and researchers should verify the specific requirements of their chosen platform before committing precious samples.

For postmortem tissue, RNA quality is an even greater concern. The evolution of gene expression analysis from classical methods such as Northern blot and in situ hybridization to modern spatial transcriptomics has enabled simultaneous analysis of histology and subcellular gene expression, but RNA quality remains the limiting factor in FFPE and postmortem samples. Close collaboration with a pathologist is essential for interpreting tissue quality and selecting appropriate regions for analysis.

Tissue Collection and Storage Conditions

The window between tissue collection and preservation determines RNA integrity. For fresh frozen tissue, the sample should be embedded in optimal cutting temperature compound and snap-frozen as quickly as possible after resection. For FFPE tissue, the fixation time should be standardized because over-fixation crosslinks RNA and proteins, reducing signal.

Researchers should record the following for every sample:

• Time from tissue collection to preservation
• Fixation method and duration for FFPE samples
• Storage temperature and duration
• RNA integrity number or DV200 value if measured
• Pathological assessment of tissue composition

These records become essential when interpreting failed experiments or unexpected results. A sample with poor RNA quality may produce data that passes platform-specific quality metrics but fails to recapitulate known biology.

Sectioning and Mounting

The orientation of tissue sections affects the spatial information captured. Serial sections from the same tissue block can be used for different assays, but adjacent sections are not identical. The benchmarking study of imaging platforms used serial sections from tissue microarrays, which enabled direct comparison of platforms on matched tissue. For studies comparing platforms or validating findings across assays, serial sectioning is a practical strategy, but researchers must account for section-to-section variation.

Section thickness affects RNA capture efficiency and imaging quality. Thicker sections retain more RNA but may compromise imaging resolution. Researchers should follow the platform-specific recommendations for section thickness and document any deviations.

Platform Selection and Resolution Requirements

Imaging-Based Platforms

Imaging-based spatial transcriptomics platforms use fluorescence in situ hybridization to detect transcripts at single-cell or subcellular resolution. These platforms are limited to panels of about a thousand genes, which constrains researchers to build panels from marker genes of different cell types and forgo other genes of interest, such as genes encoding ligand-receptor complexes or those in specific pathways.

The choice of imaging platform affects data quality and analytical options. The benchmarking study comparing Xenium, MERSCOPE, and CosMx found that all three platforms can perform spatially resolved cell typing, but with different false discovery rates and cell segmentation error frequencies. Cell segmentation is a critical analytical step for imaging platforms because inaccurate segmentation propagates errors through all downstream analyses. The study also found that Xenium and CosMx found slightly more clusters than MERSCOPE, suggesting differences in the granularity of cell state detection.

For researchers planning imaging-based experiments, the gene panel design is a primary design consideration. Probe set selection for targeted spatial transcriptomics requires selecting the most informative yet minimal set of genes to profile. Current selections often rely on marker genes, which precludes detecting continuous spatial signals or new states. The Spapros pipeline optimizes both gene set specificity for cell type identification and within-cell type expression variation to resolve spatially distinct populations while considering prior knowledge as well as probe design and expression constraints. Spapros outperformed other selection approaches in both cell type recovery and recovering expression variation beyond cell types.

An alternative approach, scGIST, uses constrained feature selection to design panels that prioritize user-specified genes without compromising cell type detection accuracy. This is valuable when researchers need to include genes beyond canonical markers, such as ligand-receptor pairs or pathway components.

Sequencing-Based Platforms

Sequencing-based spatial transcriptomics platforms capture RNA from spatially barcoded spots on a slide. The resolution is determined by the spot size and spacing, which varies by platform and product generation. These platforms capture the whole transcriptome, enabling unbiased discovery of spatially variable genes.

The tradeoff is that spot-level data represent mixtures of cells, complicating cell type identification. Integration with single-cell RNA sequencing data is a common analytical strategy to deconvolve spot-level expression into cell type contributions. A review of spatial transcriptomics methods summarizes data analysis approaches, including integration with single-cell transcriptomics data, and discusses applications in dermatologic research. These tools offer a path toward understanding niche patterning and cell-cell interactions within heterogeneous tissues.

Matching Platform to Question

The following decision framework helps match platform to biological question:

• If the question requires cell type identification at single-cell resolution in FFPE tissue, imaging platforms are appropriate.
• If the question requires whole transcriptome discovery of spatially variable genes, sequencing platforms are appropriate.
• If the question requires subcellular localization of specific transcripts, imaging platforms with subcellular resolution are required.
• If the question requires analysis of large tissue sections at moderate resolution, sequencing platforms may be more cost-effective.

A practical guide based on over 1000 spatial samples emphasizes that many groups struggle with platform selection. The guide recommends that researchers consider the tradeoffs between resolution, throughput, and cost in the context of their specific biological question.

Gene Panel Design for Targeted Approaches

Marker Genes Are Not Sufficient

Targeted spatial transcriptomic methods capture the topology of cell types and states in tissues at single-cell and subcellular resolution by measuring the expression of a predefined set of genes. The selection of an optimal set of probed genes is crucial for capturing the spatial signals present in a tissue. Current selections often rely on marker genes, which precludes detecting continuous spatial signals or new states.

Marker genes identify canonical cell types but may not distinguish cell states within a type. For example, macrophages in different activation states may share canonical markers but differ in functional gene expression. A panel limited to markers would miss these state differences.

Computational Panel Design Tools

Spapros is an end-to-end probe set selection pipeline that optimizes both gene set specificity for cell type identification and within-cell type expression variation to resolve spatially distinct populations. The pipeline considers prior knowledge as well as probe design and expression constraints. In evaluation, Spapros outperformed other selection approaches in both cell type recovery and recovering expression variation beyond cell types. The pipeline was used to design a single-cell resolution in situ hybridization experiment of adult lung tissue, demonstrating how probes selected with Spapros identify cell types of interest and detect spatial variation even within cell types.

scGIST is a constrained feature selection tool that designs panels prioritizing user-specified genes without compromising cell type detection accuracy. This approach is valuable when researchers need to include genes beyond canonical markers, such as ligand-receptor pairs or pathway components. scGIST has been demonstrated in diverse use cases, highlighting its value in the spatial transcriptomics algorithmic toolbox.

Panel Size and Cost Tradeoffs

Panel size directly affects cost and experimental complexity. Larger panels provide more information but increase the cost per sample and may reduce the number of samples that can be processed within a budget. Researchers should design panels that are informative yet minimal, balancing the need for cell type identification with the need to detect spatial variation and biological processes of interest.

For studies with limited samples, a well-designed panel that captures the key biological axes is more valuable than a larger panel that dilutes coverage of relevant genes. The panel design should be documented and justified in the study protocol, including the rationale for gene inclusion and exclusion.

Replication Strategy and Statistical Power

Biological Versus Technical Replicates

Spatial transcriptomics experiments are expensive, which creates pressure to minimize sample numbers. However, inadequate replication undermines the statistical power of downstream analyses and limits the generalizability of findings.

Biological replicates are samples from different individuals or animals within the same condition. Technical replicates are repeated measurements from the same biological sample. For spatial transcriptomics, biological replication is essential because spatial patterns vary between individuals. Technical replication is less critical because the platforms are generally reproducible, but technical replicates can help identify platform-specific artifacts.

The number of biological replicates needed depends on the biological variability of the system, the magnitude of the expected effect, and the analytical approach. Researchers should perform a power analysis based on pilot data or published studies in similar systems. For studies of heterogeneous tissues such as tumors, more replicates are needed to capture inter-tumor variability.

Spatial Sampling Within Samples

Within a single tissue section, the number of spots or cells captured is large, but these are not independent observations. Spatial autocorrelation means that nearby spots are more similar than distant spots, which affects statistical inference. Analytical methods must account for spatial dependence to avoid inflated significance.

The practical guide based on over 1000 samples emphasizes experimental scalability as a common barrier. Researchers should consider the number of sections per sample, the number of regions per section, and the total number of samples when planning experiments. A design that captures multiple regions from each sample can increase statistical power without proportionally increasing cost.

Aggregated Experimental Designs

The CIPHER framework for designing optimized aggregated spatial transcriptomics experiments addresses the challenge of balancing sample number, section number, and cost. Aggregated designs pool resources across samples to maximize information gain within a budget. Researchers planning large-scale studies should consider whether an aggregated design is appropriate for their question.

Controls and Quality Checks

Positive and Negative Controls

Spatial transcriptomics experiments require controls to distinguish biological signal from technical noise. Positive controls are genes with known spatial patterns in the tissue being studied. Negative controls are genes not expected to be expressed or probes designed against sequences not present in the sample.

For imaging platforms, negative control probes detect non-specific binding and background fluorescence. The number and type of negative controls vary by platform. Researchers should review the platform-specific quality metrics and report them in the study methods.

RNA Quality Assessment

RNA quality should be assessed before library preparation or imaging. For FFPE samples, DV200 is a common metric that measures the percentage of RNA fragments longer than 200 nucleotides. For fresh frozen samples, the RNA integrity number is used. The relationship between RNA quality metrics and spatial transcriptomics data quality is platform-specific, and researchers should establish thresholds based on their platform and tissue type.

The review of spatial transcriptomics in autopsy tissue emphasizes that RNA quality proves to be a critical factor for the reliability of analyses, especially in FFPE and postmortem samples. Standardized sample preparation and RNA quality control are essential for valid results.

Platform-Specific Quality Metrics

Each platform provides quality metrics that should be reviewed before downstream analysis:

• Total transcripts detected per spot or cell
• Number of genes detected per spot or cell
• Fraction of reads or transcripts mapping to the panel
• Negative control probe counts
• Segmentation quality metrics for imaging platforms

The benchmarking study of imaging platforms provides a reference for expected performance across tissue types. Researchers should compare their data quality to these benchmarks and investigate any substantial deviations.

Data Analysis Workflow

Preprocessing and Quality Control

The first analysis steps are platform-specific. For sequencing-based platforms, reads are mapped to the transcriptome and assigned to spatial barcodes. For imaging platforms, transcripts are detected from fluorescence images and assigned to cells through segmentation.

Quality control at the spot or cell level typically includes filtering based on total transcript counts, number of genes detected, and fraction of mitochondrial transcripts. The thresholds for these filters depend on the platform and tissue type. Researchers should document the filtering criteria and the number of spots or cells retained after each step.

Cell Type Annotation

Cell type annotation is a crucial step for downstream analyses. For imaging-based spatial data, the small gene panel makes annotation challenging. A benchmarking study compared five reference-based methods (SingleR, Azimuth, RCTD, scPred, and scmapCell) with marker-gene-based manual annotation on imaging-based Xenium data of human breast cancer. SingleR was the best performing reference-based cell type annotation tool for the Xenium platform, being fast, accurate, and easy to use, with results closely matching those of manual annotation.

The study demonstrated a practical workflow for preparing a high-quality single-cell RNA reference, evaluating accuracy, and estimating running time for reference-based cell type annotation tools. Researchers should prepare a reference dataset from matched or similar tissue and evaluate annotation accuracy before proceeding to downstream analyses.

Integration with Single-Cell RNA Sequencing

Integration with single-cell RNA sequencing data is a common analytical strategy for spatial transcriptomics. Single-cell data provide high-resolution cell type definitions that can be mapped to spatial data. The best practices for single-cell analysis across modalities provide guidance on robust data analysis, including choices of best-performing tools from benchmarking studies.

For imaging-based spatial data, the reference-based annotation tools developed for single-cell and sequencing-based spatial data may not perform optimally due to the small gene panel. Researchers should evaluate multiple tools and select the approach that best matches their data characteristics.

Spatial Domain Identification

Spatial domains are regions of the tissue with distinct gene expression programs. Identifying these domains is a common analytical goal. Methods for spatial domain identification use gene expression and spatial coordinates to segment the tissue into coherent regions.

The GATCL framework uses an adaptive contrastive learning approach based on multi-head graph attention for spatial domain identification. This method represents a class of deep learning approaches that leverage spatial graphs to improve domain detection. Researchers should evaluate multiple domain identification methods and validate the resulting domains against histological annotations.

Spatially Variable Gene Detection

Identifying genes with spatial expression patterns is a primary goal of many spatial transcriptomics studies. Methods for detecting spatially variable genes use spatial coordinates and expression values to identify genes whose expression varies across the tissue in a non-random manner.

The choice of method affects the number and identity of detected spatially variable genes. Researchers should use methods appropriate for their platform and data characteristics and validate findings with orthogonal approaches.

Validation Strategies

Orthogonal Validation Methods

Spatial transcriptomics findings should be validated with independent methods. Common validation approaches include:

• Immunohistochemistry for protein expression
• RNA in situ hybridization for RNA localization
• Quantitative PCR for expression levels
• Single-cell RNA sequencing for cell type composition

The choice of validation method depends on the finding being validated. For cell type localization, immunohistochemistry or RNA in situ hybridization is appropriate. For gene expression differences, quantitative PCR or NanoString is appropriate.

Cross-Platform Validation

Validating findings across spatial transcriptomics platforms provides strong evidence for biological robustness. The benchmarking study of imaging platforms found that Xenium and CosMx measured RNA transcripts in concordance with orthogonal single-cell transcriptomics, supporting the validity of cross-platform comparisons.

For studies with precious samples, a practical approach is to use a sequencing-based platform for discovery and an imaging-based platform for validation of specific findings. This approach leverages the strengths of each platform type.

Integration with Other Omics Data

Spatial transcriptomics data can be integrated with other omics data types to provide a more complete picture of tissue biology. The contextual activity score framework enables inference of ADAR-associated transcriptional activity from gene expression data across diverse transcriptomic technologies, including spatial transcriptomics. This approach demonstrates the value of integrating spatial data with functional annotations.

The spatial ecotypes framework integrates over 10 million single-cell and spot-level spatial transcriptomes from diverse human carcinomas and melanomas to identify multicellular programs in the tumor microenvironment. This approach demonstrates how spatial data can be integrated across studies to identify conserved biological programs.

Common Failure Patterns

RNA Degradation

RNA degradation is the most common cause of failed spatial transcriptomics experiments. Symptoms include low transcript counts, poor signal-to-noise ratio, and failure to detect known marker genes. Prevention requires rapid tissue processing, proper storage, and RNA quality assessment before library preparation.

Inadequate Panel Design

Panels that rely solely on marker genes may fail to detect biologically relevant spatial variation. Symptoms include inability to distinguish cell states, missing known spatial patterns, and poor integration with single-cell data. Prevention requires computational panel design tools that optimize for both cell type identification and expression variation.

Segmentation Errors

For imaging platforms, inaccurate cell segmentation propagates errors through all downstream analyses. Symptoms include inflated or deflated cell counts, mixed cell type signals, and poor concordance with histological annotations. Prevention requires careful review of segmentation quality and adjustment of segmentation parameters.

Batch Effects

Spatial transcriptomics experiments processed in multiple batches can exhibit batch effects that confound biological differences. Symptoms include clustering by batch instead of condition and poor reproducibility across batches. Prevention requires balanced experimental designs and computational batch correction.

Overinterpretation of Correlations

Spatial correlations between genes or cell types do not establish causal relationships. Researchers should be cautious in interpreting spatial co-localization as evidence of cell-cell communication. Validation with perturbation experiments or other functional assays is needed to establish causality.

Limitations and Interpretation

Resolution Limits

Spatial transcriptomics platforms have inherent resolution limits. Sequencing-based platforms capture RNA from spots that may contain multiple cells, complicating cell type identification. Imaging platforms may miss transcripts expressed at low levels or in dense tissue regions.

RNA Quality Dependence

All spatial transcriptomics methods depend on RNA quality. FFPE and postmortem tissues have variable RNA quality, which affects data reliability. The review of spatial transcriptomics in autopsy tissue emphasizes that RNA quality proves to be a critical factor for the reliability of analyses.

Panel Constraints

Targeted approaches are limited to predefined gene panels. Genes not included in the panel cannot be detected, which constrains discovery. Researchers should design panels carefully and consider whether whole transcriptome approaches are needed for their question.

Cost and Scalability

Spatial transcriptomics experiments are expensive, which limits sample numbers and experimental scale. The practical guide based on over 1000 samples emphasizes that experimental scalability is a common barrier to implementation. Researchers should design experiments that maximize information gain within their budget.

Data Management and Reproducibility

Data Storage and Sharing

Spatial transcriptomics data are large and complex. Raw data, processed data, and analysis code should be stored and shared according to community standards. The FAIR Guiding Principles provide a framework for making data findable, accessible, interoperable, and reusable.

The National Institutes of Health Genomic Data Sharing Policy outlines expectations for data sharing from NIH-funded research. Researchers should review the policy requirements for their funding source and plan data management accordingly.

Reproducible Analysis Workflows

Analysis workflows should be documented and version-controlled. Containerized analysis environments improve reproducibility by capturing software versions and dependencies. The best practices for single-cell analysis across modalities provide guidance on robust data analysis workflows.

Data Repositories

Spatial transcriptomics data should be deposited in appropriate repositories. The National Center for Biotechnology Information provides data resources for genomic data, including spatial transcriptomics data. The European Bioinformatics Institute provides training and data resources for bioinformatics.

Professional Escalation Criteria

Researchers should seek expert consultation when encountering the following situations:

• RNA quality metrics fall below platform-specific thresholds
• Data quality metrics deviate substantially from platform benchmarks
• Cell type annotation results conflict with histological assessment
• Spatial domain identification produces biologically implausible results
• Cross-platform validation fails to confirm key findings
• Batch effects cannot be resolved with computational correction

Consultation with a pathologist is essential for interpreting tissue morphology and validating spatial findings. Consultation with a bioinformatician is essential for complex analytical decisions and troubleshooting.

Frequently Asked Questions

How many biological replicates are needed for a spatial transcriptomics experiment?

The number of biological replicates depends on the biological variability of the system, the magnitude of the expected effect, and the analytical approach. Researchers should perform a power analysis based on pilot data or published studies in similar systems. For heterogeneous tissues such as tumors, more replicates are needed to capture inter-tumor variability. A practical approach is to start with three to five biological replicates per condition and increase the number if pilot data show high variability.

What is the difference between imaging-based and sequencing-based spatial transcriptomics?

Imaging-based platforms use fluorescence in situ hybridization to detect transcripts at single-cell or subcellular resolution but are limited to predefined gene panels. Sequencing-based platforms capture the whole transcriptome from spatially barcoded spots but at lower spatial resolution. The choice between platform types depends on whether the biological question requires single-cell resolution or whole transcriptome discovery.

How should I choose genes for a targeted spatial transcriptomics panel?

Marker genes are not sufficient for detecting continuous spatial signals or new cell states. Computational panel design tools such as Spapros and scGIST optimize gene selection for both cell type identification and expression variation. These tools consider prior knowledge, probe design constraints, and expression constraints to select informative yet minimal gene sets.

What RNA quality metrics should I assess before a spatial transcriptomics experiment?

For FFPE samples, DV200 measures the percentage of RNA fragments longer than 200 nucleotides. For fresh frozen samples, the RNA integrity number is used. RNA quality proves to be a critical factor for the reliability of analyses, especially in FFPE and postmortem samples. Researchers should establish platform-specific thresholds based on their tissue type and validate these thresholds with pilot experiments.

How do I validate spatial transcriptomics findings?

Findings should be validated with orthogonal methods such as immunohistochemistry, RNA in situ hybridization, quantitative PCR, or single-cell RNA sequencing. Cross-platform validation provides strong evidence for biological robustness. The benchmarking study of imaging platforms found that Xenium and CosMx measured RNA transcripts in concordance with orthogonal single-cell transcriptomics.

What are the most common causes of failed spatial transcriptomics experiments?

RNA degradation is the most common cause of failed experiments. Inadequate panel design, segmentation errors, and batch effects also contribute to poor data quality. Prevention requires rapid tissue processing, computational panel design, careful review of segmentation quality, and balanced experimental designs.

How should I handle batch effects in spatial transcriptomics data?

Batch effects can be minimized through balanced experimental designs that distribute conditions across batches. Computational batch correction methods can help, but they cannot fully compensate for poor experimental design. Researchers should document batch information and test for batch effects before proceeding to downstream analyses.

What data management practices support reproducibility in spatial transcriptomics?

Raw data, processed data, and analysis code should be stored and shared according to community standards. The FAIR Guiding Principles provide a framework for making data findable, accessible, interoperable, and reusable. Containerized analysis environments improve reproducibility by capturing software versions and dependencies. Data should be deposited in appropriate repositories such as those provided by the National Center for Biotechnology Information.

Related Bioinformatics Guides

• Spatial Transcriptomics: Mapping the Cellular Atlas
• Computational Strategies in Structure Based Drug Design
• Spatial Transcriptomics Alignment and Cellular Neighborhood Analysis
• Modern Transcriptomics: From Bulk RNA-Seq to Single-Cell and Spatial Resolution
• Ethical Considerations in Computational Genomics

References and Further Reading

• EMBL-EBI Training. European Bioinformatics Institute.
• NCBI Data Resources. National Center for Biotechnology Information.
• Genomic Data Sharing Policy. National Institutes of Health.
• The FAIR Guiding Principles. Scientific Data.
• Non-invasive profiling of the tumour microenvironment with spatial ecotypes.. Nature, 2026.
• Systematic benchmarking of imaging spatial transcriptomics platforms in FFPE tissues.. Nature communications, 2025.
• Dynamic urinary proteomics integrates single-cell and spatial transcriptomics to reveal tumour microenvironment and predict immunotherapy response in biliary tract cancer.. Gut, 2026.
• Probe set selection for targeted spatial transcriptomics.. Nature methods, 2024.
• Assessment of a HER2 scoring system for colorectal cancer: results from a validation study.. Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc, 2015.
• scGIST: gene panel design for spatial transcriptomics with prioritized gene sets.. Genome biology, 2024.
• Sialic acid aptamer and RNA in situ hybridization-mediated proximity ligation assay for spatial imaging of glycoRNAs in single cells.. Nature protocols, 2025.
• Research Techniques Made Simple: Spatial Transcriptomics.. The Journal of investigative dermatology, 2022.
• Protocol for spatially resolved pathology scores using optimal transport on spatial transcriptomics data.. 2026.
• A contextual activity score (CAS) for inferring ADAR-associated transcriptional activity across RNA-seq, single-cell, and spatial transcriptomics.. 2026.
• Multi-Dimensional Transcriptomics Reveals the Prominent Role of Neuroinflammation in Alzheimer's Disease.. 2026.
• Characterizing Infiltrating Macrophages in Intracranial Aneurysm with IA Animal Models and Spatial Transcriptomics.. 2026.
• [Half a century of gene expression analysis : The path to modern spatial transcriptomics in autopsy tissue].. 2026.
• circVDJ-seq for T cell clonotype detection in single-cell and spatial multi-omics.. 2026.
• A practical guide to spatial transcriptomics: lessons from over 1000 samples.. Trends in Biotechnology, 2025.
• Best practices for single-cell analysis across modalities. Nature reviews genetics, 2023.
• Optimizing Xenium In Situ data utility by quality assessment and best-practice analysis workflows. bioRxiv, 2023.
• SEPAL: Spatial Gene Expression Prediction from Local Graphs. 2023 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), 2023.
• Benchmarking cell type annotation methods for 10x Xenium spatial transcriptomics data. BMC Bioinformatics, 2025.
• GATCL: An Adaptive Contrastive Learning Framework Based on MHGAT for Spatial Domain Identification in Spatial Transcriptomics. Proceedings of the Aaai Conference on Artificial Intelligence, 2026.
• Advances and applications in spatial transcriptomics. Journal of Molecular Science, 2024.
• PGST: A prototype-guided parameter-efficient network for spatial transcriptomics prediction. IEEE Journal of Biomedical and Health Informatics, 2026.
• Microarray integrated spatial transcriptomics (MIST) for affordable and robust digital pathology. Npj Systems Biology and Applications, 2024.
• CIPHER: An end-to-end framework for designing optimized aggregated spatial transcriptomics experiments. Plos One, 2026.

This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.