Organ-Scale Spatial Transcriptomics With Stereo-seq: Study Design Guide

Inquiry

Figure 1. Stereo-seq large-format chip with whole-organ tissue mapping — subcellular resolution across a centimeter-scale continuous capture area.

Most spatial transcriptomics platforms ask you to accept a trade-off: high resolution or large field of view, but not both. Stereo-seq is the exception — its 500 nm DNB spacing and capture areas up to 13.2 × 13.2 cm make whole-organ, subcellular-resolution mapping feasible on a single continuous section. But that capability comes with its own set of planning decisions: which chip size to use, how to orient the tissue, whether to run serial sections for 3D reconstruction, and how to choose a bin size that matches the biological question. This guide walks through the design choices that determine whether an organ-scale Stereo-seq project produces a publication-quality atlas or an expensive pilot that missed its target.

Key Takeaways

• Stereo-seq is the only platform that combines subcellular resolution with centimeter-scale capture area. Maximum 13.2 × 13.2 cm field of view — roughly 500× larger than a standard Visium capture area — enables whole mouse embryos, entire human organ cross-sections, and large model organisms on a single chip.
• Serial sections enable true 3D spatial transcriptomics. Stereo-seq's coordinate-based barcoding (CID) provides precise spatial addresses that make cross-slice alignment more accurate than bead-based methods, demonstrated in macaque cortex (161 slices) and Drosophila embryo (full developmental series) atlases.
• Chip size drives every downstream decision. Standard 1 × 1 cm chips suit routine sections; large-format chips (up to 13 × 13 cm) are custom-order items that require advance planning with the service provider.
• Bin size is a biological choice, not just a computational one. Bin20 (~10 μm) approximates single-cell resolution for cell-type mapping; Bin50–100 suits tissue-region analysis; Bin200 is for rapid visualization. Each step up in bin size trades resolution for computational speed.
• Multi-tissue blocks follow strict layout rules. Same tissue type required for uniform permeabilization; minimum 1 mm spacing between tissues; coverage must stay below 80% of the chip surface.

Why Organ-Scale Matters

In spatial transcriptomics, field of view is not just a convenience — it determines which biological questions you can ask. If your capture area is 6.5 × 6.5 mm (a standard Visium HD capture area), a whole mouse embryo sagittal section must be reconstructed from multiple captures stitched together. Each stitch introduces registration uncertainty. Fine spatial boundaries between adjacent anatomical regions — the very features that distinguish one developing structure from another — are the first casualties of stitching error.

Stereo-seq eliminates this problem at the hardware level. The DNA nanoball (DNB) array is manufactured as a continuous patterned surface on a silicon chip, with each DNB carrying a unique Coordinate Identity (CID) barcode decoded by sequencing before tissue is applied. There are no gaps between capture areas, no beads to redistribute, no stitching required — the entire 13.2 × 13.2 cm surface is one continuous coordinate system. For a researcher studying a whole mouse embryo, a complete human liver cross-section, or a tumor with its full microenvironment, this means the spatial relationships between every cell and every structure are captured natively in a single experiment.

The practical implication for study design is straightforward: if your biological question involves spatial relationships that span more than ~1 cm of continuous tissue, Stereo-seq is not merely the best option — it is the only subcellular-resolution option. For guidance on whether your project fits Stereo-seq in the first place, including resolution and tissue-type considerations, see the Stereo-seq platform decision guide.

Chip Sizes and Capture Area Planning

Stereo-seq chips are available in multiple formats, and the choice of chip size is the first concrete decision in project planning. It determines how much tissue you can capture, how many sections you can run in parallel, and — for custom large-format chips — the project timeline.

Custom large-format chips are not off-the-shelf items — they require advance coordination with the service provider and longer lead times. If your project needs a chip larger than the standard 1 × 1 cm format, discuss chip availability early in the planning process, ideally before tissue collection begins.

Tissue placement rules. Regardless of chip size, three placement rules apply. First, the tissue must not exceed 80% of the chip surface — leaving a margin around the tissue edge prevents artifacts during permeabilization and washing. Second, if embedding multiple tissue pieces in the same OCT block for capture on a single chip, all pieces must be of the same tissue type — uniform permeabilization requires uniform tissue characteristics. Third, maintain at least 1 mm of space between adjacent tissue pieces; closer spacing risks cross-contamination during permeabilization and makes computational separation of the two regions unnecessarily difficult.

Orientation matters. The tissue section should be placed so that the longest anatomical axis of interest aligns with the chip's longer dimension when using rectangular formats. For irregularly shaped tissues, photograph the chip after mounting and before processing — this reference image is invaluable during analysis for confirming that spatial gene expression patterns correspond to anatomical landmarks rather than tissue-handling artifacts.

Serial Sections and 3D Reconstruction

The most ambitious Stereo-seq studies go beyond single-section analysis to full 3D spatial transcriptomics. Because each DNB carries a known coordinate in a continuous coordinate system, serial sections from the same tissue block can be computationally aligned with higher precision than bead-based methods, where bead positions are unique to each capture area and must be cross-registered probabilistically.

Three landmark studies illustrate the range of what serial-section Stereo-seq can achieve:

• Mouse organogenesis (Chen et al., 2022): 53 sagittal sections across 8 developmental timepoints generated the Mouse Organogenesis Spatiotemporal Transcriptomic Atlas (MOSTA), capturing whole-embryo gene expression gradients at subcellular resolution. This remains the foundational demonstration of Stereo-seq for developmental biology.
• Macaque cortex 3D atlas (Lei et al., 2023): 161 consecutive 10 μm coronal sections from a single macaque brain were profiled on large-format (5 × 3 cm) chips, producing a comprehensive 3D single-cell spatial transcriptomic atlas of the entire cerebral cortex with 264 transcriptome-defined cell types mapped across 143 cortical regions.
• Drosophila embryogenesis (Wang et al., 2022): Serial sections across developmental stages from embryo to larva were reconstructed into 3D spatiotemporal transcriptomic maps, identifying functional subregions of the midgut and tracking cell-state transitions in the larval testis in their native spatial context.

Planning for serial sections. The key planning parameters are section thickness, spacing, and total section count. For fresh frozen tissue, sections are typically cut at 10 μm; for FFPE, at 5 μm. At 10 μm thickness, a 1 cm tissue block yields up to 1,000 sections — far more than any project can reasonably profile. Most 3D studies use a sampling strategy: profile every Nth section (e.g., every 5th or 10th section, yielding 100–200 μm spacing between profiled planes) and interpolate the intervening tissue computationally. The sampling interval should be decided based on the size of the structures you need to resolve — if you need to track individual glomeruli (~100 μm diameter), spacing should be no more than ~50 μm to avoid missing structures between sections.

Alignment strategy. Stereo-seq's CID coordinate system provides an absolute spatial reference within each section, but cross-section alignment requires additional computational steps. The standard approach uses landmark-based registration: identify common anatomical features visible across adjacent sections, compute a rigid or affine transformation, and apply it to align the spatial coordinate systems. Dedicated tools for 3D Stereo-seq reconstruction are available through the SAW (Stereo-seq Analysis Workflow) pipeline and associated software.

Choosing Your Bin Size

Stereo-seq raw data resolves individual DNB spots at ~500 nm spacing, but single-spot analysis is rarely practical — individual DNBs capture too few transcripts for reliable gene detection. Instead, DNBs are aggregated into square bins for analysis. The choice of bin size is a biological decision that balances spatial resolution against per-bin transcript counts.

The trade-off is straightforward: smaller bins give you higher spatial resolution but fewer transcripts per bin, which means more dropout and noisier gene expression estimates. Larger bins give you more robust per-bin counts but blur fine spatial boundaries. A practical workflow is to perform initial QC and clustering at Bin50 or Bin100, identify regions of interest, and then re-analyze those regions at Bin20 for cell-type-resolution detail.

For organ-scale projects, bin size interacts with total data volume. A 1 × 1 cm chip at Bin20 generates roughly 250,000 spatial units — each requiring expression quantification across the transcriptome. Computational requirements scale accordingly: Bin100 analysis runs comfortably on a standard workstation; Bin20 analysis of a full 1 × 1 cm chip benefits from high-performance computing resources. If computational resources are a constraint, plan your bin size accordingly and communicate expectations to your bioinformatics team before sequencing begins.

Figure 2. Serial section workflow for Stereo-seq 3D reconstruction — from tissue block to aligned 3D spatial transcriptome model.

Section Orientation and Placement

How a tissue section is oriented on the chip may seem like a minor detail, but it has outsized effects on data interpretability. A section placed at an oblique angle relative to anatomical axes produces spatial gene expression gradients that run diagonally across the coordinate grid — analyzable, but harder to map onto known anatomical structures. A section placed with its primary anatomical axis aligned to the chip edge produces gradients that map cleanly onto the x or y coordinate, making biological interpretation far more straightforward.

Practical guidelines for tissue placement:

• Align the primary anatomical axis of interest (anterior-posterior, dorsal-ventral, or proximal-distal) with one edge of the chip. For a sagittal brain section, for example, align the anterior-posterior axis with the long edge of the chip.
• Photograph every chip after mounting and before processing. Include a ruler or scale reference in the image. This photograph becomes the ground-truth reference for verifying that computational spatial mappings correspond to actual tissue morphology.
• Avoid placing the most critical anatomical region near the chip edge — the outer ~0.5 mm of the capture area can show edge effects from uneven permeabilization or washing.
• For multi-section studies (serial sections for 3D), maintain consistent orientation across all sections. Inconsistent orientation across a series complicates cross-section alignment and can introduce rotational artifacts into the reconstructed 3D model.

Multi-region capture. If you need to profile multiple discrete regions from the same tissue type — for example, tumor core and invasive margin from the same surgical specimen — you can mount multiple tissue pieces on a single chip, provided the pieces are of the same tissue type and spaced at least 1 mm apart. This approach conserves chips and sequencing costs while allowing direct comparison of spatial gene expression patterns across regions processed under identical conditions.

From Study Plan to Project Readiness

Organ-scale Stereo-seq projects have more moving parts than standard single-section experiments. The following checklist covers the planning decisions that should be finalized before tissue is sectioned for the capture chip.

Pre-project checklist:

• Biological question defined in spatial terms — what structures, gradients, or boundaries must be resolved?
• Chip size selected and confirmed as available through the service provider
• Tissue type and preservation method (fresh frozen vs. FFPE) determined; sample QC completed per Stereo-seq requirements (RIN ≥ 7 or DV200 ≥ 30%)
• Section orientation planned and documented with a reference diagram
• For serial section studies: sampling interval determined (every Nth section), total section count estimated, alignment strategy selected
• Bin size chosen based on the spatial scale of the biological question
• Sequencing depth estimated — approximately 1.5 billion reads per standard 1 × 1 cm chip; scale proportionally for larger formats
• Computational resources confirmed: Bin20 full-chip analysis requires HPC or cloud compute; Bin50–100 runs on standard workstations
• Data storage planned — raw sequencing data, processed expression matrices, and visualization files for a full organ-scale project can exceed several terabytes

For researchers preparing tissue for Stereo-seq, sample QC requirements and handling protocols for both fresh frozen and FFPE workflows are covered in detail in the Stereo-seq sample preparation and QC guide. General project scoping, feasibility assessment, and service intake information are available through CD Genomics Spatial Transcriptomics Services.

Figure 3. Organ-scale Stereo-seq project planning checklist — six decision categories to finalize before tissue sectioning.

FAQ

Q: How large a tissue can Stereo-seq actually accommodate?

The maximum chip size is 13.2 × 13.2 cm — large enough for a whole mouse embryo, an entire human liver cross-section, or a complete sagittal section of a macaque brain hemisphere. In practice, most organ-scale projects use chips in the 2 × 2 cm to 5 × 3 cm range. Chips larger than 1 × 1 cm are custom-order items; contact the service provider for current availability and lead times before building a project timeline around a specific large format.

Q: How many serial sections do I need for a convincing 3D reconstruction?

The number depends on the size of the structures you need to resolve. For tracking cortical layers (~1 mm thick), spacing sections every 200 μm (every 20th section at 10 μm thickness, or ~5 sections per mm of tissue depth) is usually sufficient. For resolving individual glomeruli or follicles (~100 μm diameter), aim for section spacing no more than 50 μm (every 5th section). A pilot test with 3–5 sections at your planned spacing will confirm whether structures of interest are traceable across the interval before you commit to a full series.

Q: Can I mix fresh frozen and FFPE sections in the same 3D reconstruction study?

Technically possible but strongly discouraged. Fresh frozen and FFPE workflows use different capture chemistries (poly(A) vs. random primers), different section thicknesses (10 μm vs. 5 μm), and different permeabilization protocols. The resulting data have different gene detection sensitivities and spatial resolutions, making cross-workflow alignment unreliable. If your study requires both fresh frozen and FFPE data, analyze them as complementary datasets rather than attempting to merge them into a single 3D reconstruction.

Q: What bin size should I start with for a new tissue type?

Start with Bin50 (~25 μm). It provides enough spatial resolution to identify major tissue compartments and enough transcripts per bin for robust clustering, without the computational overhead of Bin20. After initial analysis at Bin50, re-analyze regions of particular interest at Bin20. Bin100 is most useful for a first-pass overview of data quality and broad tissue architecture before committing to finer-grained analysis.

Q: How should I handle tissue pieces that are slightly larger than the chip?

If the tissue exceeds the chip dimensions, you have two options. Option 1: trim the tissue block before sectioning to fit within the active capture area — preferred when the excess tissue is not essential to the biological question. Option 2: run two adjacent sections on two separate chips and computationally align them after processing — viable but requires careful documentation of the cut line and adds registration uncertainty. In general, it is better to select a chip size that accommodates the full region of interest in a single capture than to split a structure across two chips.

References

1. Chen A, Liao S, Cheng M, et al. Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays. Cell. 2022;185(10):1777-1792.e21. doi:10.1016/j.cell.2022.04.003
2. Lei Y, Li C, Liao S, et al. Single-cell spatial transcriptome reveals cell-type organization in the macaque cortex. Cell. 2023;186(17):3726-3743.e24. doi:10.1016/j.cell.2023.06.009
3. Wang M, Hu Q, Lv T, et al. High-resolution 3D spatiotemporal transcriptomic maps of developing Drosophila embryos and larvae. Developmental Cell. 2022;57(10):1271-1283.e4. doi:10.1016/j.devcel.2022.04.006
4. Ren P, Zhang R, Luo C, et al. Systematic benchmarking of high-throughput subcellular spatial transcriptomics platforms across human tumors. Nature Communications. 2025;16:9232. doi:10.1038/s41467-025-64292-3
5. You Y, Fu Y, Li L, et al. Systematic comparison of sequencing-based spatial transcriptomic methods. Nature Methods. 2024;21:1743-1754. doi:10.1038/s41592-024-02325-3

The information provided in this article is for research use only and is not intended for use in diagnostic or therapeutic procedures. CD Genomics provides sequencing and bioinformatics services for research purposes. Researchers should consult the appropriate regulatory guidelines for their specific applications.