Technology Overview¶
The spatial omics landscape spans dozens of technologies that differ along three fundamental axes: resolution (what spatial granularity is captured), throughput (how many genes/proteins are measured), and tissue compatibility (fresh-frozen vs. FFPE). Understanding these trade-offs is essential for technology selection.
Resolution vs. throughput landscape¶
| Technology | Resolution | Genes/Proteins | Tissue type | Commercial? |
|---|---|---|---|---|
| Visium | 55 um (multi-cell) | Whole transcriptome | FF, FFPE (CytAssist) | Yes (10x) |
| Visium HD | 2 um (bins) | Whole transcriptome | FF, FFPE | Yes (10x) |
| Slide-seq V2 | 10 um | Whole transcriptome | FF only | No (academic) |
| Stereo-seq | ~500 nm | Whole transcriptome | FF | Yes (BGI/STOmics) |
| Open-ST | Subcellular | Whole transcriptome | FF | No (open-source) |
| Seq-Scope | Subcellular | Whole transcriptome | FF | No (academic) |
| Xenium | Subcellular | 100-5,000 (panel) | FF, FFPE | Yes (10x) |
| MERSCOPE | Subcellular | 100-1,000 (panel) | FF, FFPE | Yes (Vizgen/10x) |
| CosMx SMI | Subcellular | 1,000-6,000 (panel) | FF, FFPE | Yes (Bruker) |
| GeoMx DSP | ROI-level | Whole transcriptome or ~100 proteins | FF, FFPE | Yes (Bruker) |
| PhenoCycler | Subcellular | ~100 proteins | FF, FFPE | Yes (Akoya) |
| MIBI | ~260 nm | ~40 proteins | FF, FFPE | Discontinued |
| IMC/Hyperion | 1 um | ~40 proteins | FF, FFPE | Yes (Standard BioTools) |
| MALDI-MSI | 5-50 um | Untargeted metabolites | FF, FFPE | Yes (Bruker, others) |
The fundamental trade-off¶
Sequencing-based vs. imaging-based
- Sequencing-based methods (Visium, Visium HD, Stereo-seq, Slide-seq) capture the whole transcriptome but at lower spatial resolution (multi-cell spots or small bins that require aggregation). Discovery-oriented: no need to pre-select a gene panel.
- Imaging-based methods (Xenium, MERFISH, CosMx, seqFISH) achieve subcellular resolution with single-molecule sensitivity but require a pre-designed gene panel (typically 100-5,000 genes). Hypothesis-confirming or targeted discovery.
- Protein-based methods (PhenoCycler, IMC, MIBI) measure protein abundance at subcellular resolution but are limited to ~40-100 markers by antibody availability. Best for immune phenotyping and tissue architecture.
Commercial consolidation (2023-2025)¶
The spatial omics vendor landscape has undergone rapid consolidation:
| Event | Year | Impact |
|---|---|---|
| 10x Genomics acquires Vizgen (MERSCOPE) | 2024 | 10x now controls both the leading sequencing-based (Visium/HD) and imaging-based (Xenium, MERSCOPE) platforms. MERSCOPE's future product roadmap is uncertain. |
| NanoString acquired by Bruker | 2024 | CosMx SMI and GeoMx DSP continue under the Bruker Spatial Biology brand. Financial stability improved but R&D direction may shift. |
| Ionpath shuts down | 2023 | MIBIscope (MIBI) is no longer commercially available. Existing instruments continue to operate but without vendor support. A significant loss for spatial proteomics. |
| Akoya Biosciences financial pressure | 2024-25 | PhenoCycler remains available but Akoya's long-term viability is uncertain. |
Vendor risk
Technology selection now requires evaluating commercial viability alongside scientific merit. Choosing a platform from a vendor under financial pressure creates risk for long-term projects. 10x Genomics and Bruker are currently the most stable vendors, though both face their own challenges.
Technology selection guide¶
If the goal is discovery (no prior gene list)¶
Use a sequencing-based method:
- Standard tissue, established workflow needed: Visium (most tutorials, most tools, largest community)
- Higher resolution needed: Visium HD (2 um bins, but tooling is still maturing)
- Largest possible tissue area: Stereo-seq (centimeter-scale, but data management is challenging)
- Academic, budget-conscious: Open-ST or Seq-Scope (open-source protocols, no instrument purchase)
If the goal is targeted validation (known gene panel)¶
Use an imaging-based method:
- Largest panel needed: CosMx SMI (up to ~6,000 genes) or Xenium (up to ~5,000 genes)
- Established MERFISH expertise: MERSCOPE (mature platform, but monitor 10x integration plans)
- FFPE clinical samples: Xenium or CosMx (both handle FFPE well)
- Subcellular transcript localization matters: Any imaging-based method (all provide molecule coordinates)
If the goal is protein phenotyping¶
Use a protein-based method:
- High-plex protein on standard microscope: PhenoCycler (formerly CODEX)
- Highest spatial resolution for protein: IMC/Hyperion (1 um, metal-tagged antibodies)
- Combined RNA + protein: CosMx SMI (supports both in the same experiment)
If the goal is chromatin accessibility or multi-omic¶
See Spatial Multi-Omics -- emerging technologies with limited but growing tool support.
If the goal is metabolite mapping¶
See Spatial Metabolomics -- mass spectrometry imaging approaches with fundamentally different data structures.
Cross-technology comparison: practical considerations¶
| Consideration | Sequencing-based | Imaging-based | Protein-based |
|---|---|---|---|
| Cost per sample | $$ | $$$ | $$ |
| Instrument cost | Requires sequencer | Dedicated instrument ($500K-1M+) | Varies (microscope to dedicated) |
| Hands-on time | Low (standard library prep) | Medium-High (imaging runs hours-days) | Medium (iterative staining) |
| FFPE compatibility | Yes (CytAssist, some platforms) | Yes (Xenium, CosMx) | Yes (most) |
| Data size per sample | 1-50 GB | 50-500 GB (images) | 10-100 GB (images) |
| Computational expertise needed | Medium | High (segmentation-dependent) | Medium-High |
| Deconvolution needed? | Yes (except Visium HD with Bin2Cell) | No (single-cell resolution) | No |
| Reference scRNA-seq needed? | Usually (for deconvolution) | Helpful (for annotation) | No |
For detailed coverage of each technology category, see: