Spatial Transcriptomics Analysis: Questions to Ask Before Choosing a Workflow

If you are planning a spatial transcriptomics experiment, the single most important decision is whether your biological question demands subcellular resolution, cellular resolution, or regional patterning. This guide is for principal investigators, postdocs, and graduate students who are evaluating platforms and computational pipelines for the first time. We walk through the key questions that bridge your biological goal with the technical specifications of spatial transcriptomics workflows.

At a Glance

Decision Criteria for Selecting a Spatial Transcriptomics Workflow

1. Match Resolution to Your Biological Question

Spatial transcriptomics methods fall along a resolution continuum. Sequencing based approaches like 10x Visium capture RNA from 55 micron diameter spots, each containing 1,10 cells. This is sufficient for identifying tissue regions and major cell types but cannot resolve individual cells. Imaging based methods such as MERFISH or seqFISH+ achieve subcellular resolution by directly visualizing transcripts in situ. If your question involves ligand receptor pairs between adjacent cells or subcellular localization of RNA, you need the higher resolution. The EMBL EBI Training materials provide excellent overviews of how resolution influences data interpretation [2].

For studies that integrate single cell and spatial data, resolution mismatch becomes a critical concern. A recent study in glioblastoma used integrated single cell and spatial transcriptomics to reveal a lactate driven crosstalk between NFATc4 positive tumor cells and SPP1 positive macrophages [6]. They relied on cellular resolution from imaging to confirm the spatial proximity of the two populations.

2. Consider Tissue Compatibility and Preparation

Not every platform works with every tissue type. Fresh frozen tissue is compatible with most sequencing and imaging methods. FFPE tissue works with some targeted approaches like the Nanostring GeoMx DSP or the updated Visium FFPE assay. If your tissue is archived or from a clinical biobank, check the platform's compatibility first. The NCBI Bookshelf contains detailed protocols on tissue handling for genomic assays [1].

If you plan to use a public dataset for method development, the NCBI Sequence Read Archive (SRA) holds spatial transcriptomics raw data from many platforms [5]. You can download fastq files and test your pipeline before committing to a wet lab experiment.

3. Evaluate Image Data Requirements

Nearly every spatial transcriptomics workflow requires a histological image (H&E or immunofluorescence) for alignment and downstream analysis. The image is used to map transcript capture locations onto tissue morphology. If your platform uses predetermined spots (e.g., 10x Visium), the image is essential for quality control and for defining tissue regions. For imaging based methods, the image is the primary data itself.

Aligning image and transcript data introduces computational challenges. The Galaxy Training Network offers tutorials on image processing and spatial registration that are accessible to researchers without a computational background [3]. Their workflows are also useful for quality checking your alignment.

4. Choose Computational Methods That Match Your Data

After selecting a platform, you need to decide on a computational pipeline. Many analysis steps are shared: normalization, dimensionality reduction, clustering, and spatial domain detection. However, platform specific steps vary. For spot based data, you must account for spot size and multiple cells per spot. For single cell resolution data, you need segmentation algorithms that accurately assign transcripts to cells.

Bioconductor hosts several packages for spatial transcriptomics, including SpatialExperiment, BayesSpace, and SPARK [4]. These packages are well documented and regularly updated. For larger datasets or multi group comparisons, a recently developed method called scalable multi group nonnegative spatial factorization handles cell type heterogeneity across multiple tissue sections [11]. This approach is particularly useful for studies comparing disease versus control.

5. Plan for Integration with Single Cell References

A common workflow is to perform single cell RNA sequencing on dissociated cells from the same tissue and then map those cell types onto the spatial map. This integration step requires careful normalization and batch correction. Several tools exist, but they assume that the single cell reference contains all the cell states present in the spatial data. If your tissue has rare or spatially restricted populations, they may be missing from the reference.

A study on Schistosoma japonicum used single cell and spatial transcriptomics to unveil regulators governing cell differentiation during sexual development [8]. The integration approach helped them identify spatially restricted progenitor cells that were underrepresented in the single cell data.

Practical Workflow or Implementation Sequence

Follow these steps to design and execute a spatial transcriptomics analysis:

1. Define your biological question precisely. Write down the expected spatial resolution required. If you need to localize RNA to subcellular compartments, choose imaging. If you need a whole transcriptome view at regional level, choose sequencing.

2. Check tissue availability and compatibility. Consult your core facility or platform documentation to confirm your tissue type is supported.

3. Decide on image modality. Select H&E for morphological landmarks or multiplexed immunofluorescence for protein markers. Ensure your platform can align the chosen image type.

4. Plan for a single cell reference if needed. If your question requires identifying rare cell types, include a single cell component in your experiment.

5. Select a computational framework. Use established pipelines from Galaxy Training [3] or Bioconductor [4] for initial analysis. For advanced statistical modeling, consider the scalable factorization method [11].

6. Perform quality control. Check for tissue damage, RNA degradation, and alignment errors. Remove low quality spots or cells.

7. Identify spatial domains. Use clustering and spatial autocorrelation to find regions with distinct expression patterns.

8. Validate findings. Compare with known tissue anatomy or perform independent experiments like RNAscope or immunofluorescence.

Common Mistakes

Mistake 1: Ignoring the image alignment step. Many researchers treat the image as cosmetic rather than computational. Poor alignment leads to misassignment of transcripts to tissue regions and false spatial patterns.

Mistake 2: Assuming spot resolution equals single cell resolution. With 10x Visium, a spot often contains multiple cells. Calling a spot "cell type A" without deconvolution is misleading.

Mistake 3: Overlooking the need for biological replicates. Spatial transcriptomics experiments are expensive, so researchers often run one sample per condition. This makes it impossible to separate biological variation from technical noise. A study on type 1 diabetes used multiple donors to show macrophage remodeling signatures for diagnosis and risk stratification [10]. Replicates were essential for statistical confidence.

Mistake 4: Using default parameters without understanding the algorithm. Spatial clustering methods have tunable parameters that affect the number and shape of domains. Always test a range of values.

Limits and Uncertainty

Spatial transcriptomics is still a maturing field. Key limitations include:

• Resolution versus depth trade off. Imaging methods capture fewer genes (targeted panels) while sequencing methods capture the whole transcriptome at lower resolution.
• Batch effects. Differences between tissue sections or library preparation batches can obscure true biological patterns. Computational batch correction exists but is not perfect.
• Uncertainty in segmentation. For imaging based methods, each cell's boundary is estimated from the image. Errors in segmentation propagate into expression profiles.
• Cost. High resolution spatial methods remain expensive, limiting sample sizes.
• Statistical power. Spatial data are highly correlated, standard statistical tests may produce inflated significance. Use spatial appropriate statistics.

A study on premature ovarian insufficiency in mice used spatial transcriptomics to reveal that Forkhead box O3 mediated mitochondrial dynamics imbalance drives the condition [9]. The authors acknowledged that their conclusions depended on a single time point and a limited number of sections. Replication in larger cohorts is needed.

Frequently Asked Questions

1. Can I use FFPE tissue for spatial transcriptomics?
Yes, but only certain platforms support it. 10x Visium FFPE and GeoMx DSP are designed for formalin fixed tissue. Check platform documentation for RNA integrity requirements. Fresh frozen tissue remains the gold standard for whole transcriptome assays.

2. How do I choose between sequencing based and imaging based methods?
If your question requires whole transcriptome discovery, use sequencing. If you need single cell or subcellular resolution and can pre select a gene panel, use imaging. Many early stage projects use sequencing for discovery and imaging for validation.

3. Do I need a paired single cell reference?
Not always. If you have a well characterized tissue with known cell type markers, you can use a reference derived from public data. However, for novel tissues or disease states, paired single cell data is strongly recommended to avoid misannotation.

4. How many samples should I replicate?
Aim for at least three biological replicates per condition. Spatial transcriptomics is expensive, but without replicates you cannot distinguish spatial patterns driven by biology from those driven by chance.

References and Further Reading

• NCBI Bookshelf , Free biomedical books covering tissue handling and genomics protocols. https://www.ncbi.nlm.nih.gov/books/ [1]
• EMBL EBI Training , Official training resources for biological data analysis, including spatial transcriptomics. https://www.ebi.ac.uk/training/ [2]
• Galaxy Training Network , Open bioinformatics workflow training materials for spatial data. https://training.galaxyproject.org/ [3]
• Bioconductor , Open source software and documentation for spatial genomic data analysis. https://bioconductor.org/ [4]
• NCBI Sequence Read Archive , Public repository for high throughput sequencing data including spatial transcriptomics. https://www.ncbi.nlm.nih.gov/sra [5]
• Integrative single cell and spatial transcriptomic analysis reveals a lactate driven crosstalk in glioblastoma. J Transl Med (2025). PubMed ID: 42436508. [6]
• Scalable multi group nonnegative spatial factorization for spatial genomics data with cell type heterogeneity. bioRxiv (2025). PubMed ID: 42433294. [11]
• Single cell and spatial transcriptomics unveil key regulators governing cell differentiation for Schistosoma japonicum sexual development. Adv Sci (2025). PubMed ID: 42435768. [8]
• Spatial transcriptomic atlas reveals that Forkhead box O3 mediated mitochondrial dynamics imbalance drives premature ovarian insufficiency in mice. Aging Cell (2025). PubMed ID: 42435003. [9]
• From islet to blood: macrophage remodeling signatures for diagnosis and risk stratification in type 1 diabetes. Front Immunol (2025). PubMed ID: 42433374. [10]

Related Articles

• Variant Calling Workflow: Design Decisions Before You Start
• VCF Annotation: How to Read Variant Consequence Reports
• Genome Browsers for Researchers: A Guide to Inspecting Genomic Evidence
• Sequence Alignment: Choosing the Right Method for Your Biological Question
• Multiple Sequence Alignment: Common Pitfalls and Quality Checks