Why Science? | GeoMx DSP, Photocleavable Tags and Region-Level RNA–Protein Evidence

Section 1 of 36

1. Start with a labelled tissue landscape

Digital Spatial Profiling begins with a fixed tissue section carrying morphology markers and many oligonucleotide-tagged probes or antibodies. The morphology image helps researchers choose where to measure. It is not decorative: region selection determines which molecules contribute to each count and which tissue remains outside the question.

Contents · Next section

Section 2 of 36

2. Attach a countable tag to each target reagent

RNA probes or antibodies carry oligonucleotide tags through photocleavable linkers. The biological target is recognised in tissue, while the released tag becomes the counted surrogate. Specificity therefore depends on both molecular recognition and tag handling. Counting a clean oligo does not rescue a nonspecific antibody or poorly designed probe.

Contents · Previous section · Next section

Section 3 of 36

3. Use light to release tags from a chosen region

Patterned ultraviolet illumination cleaves tags within a selected region of interest or compartment. Released oligos are collected and quantified by an nCounter system or sequencing, depending on the assay. The measurement belongs to the illuminated mask and collection event, not to every cell individually unless the design and validation truly support that scale.

Contents · Previous section · Next section

Section 4 of 36

4. Understand what digital means here

Digital refers to counting identity-coded molecules rather than estimating colour intensity alone. Counts can be more multiplexed and comparable than visual staining, yet they still include background, efficiency differences and sampling variation. A number is not automatically absolute concentration or biological causation.

Contents · Previous section · Next section

Section 5 of 36

5. Treat morphology as a sampling decision

Researchers may choose tumour, stroma, immune-rich areas or marker-defined compartments. That flexibility is powerful and vulnerable to bias. If regions are selected after viewing outcomes or because they look dramatic, estimates can overstate the tissue. Selection rules should be declared before molecular counts are inspected.

Contents · Previous section · Next section

Section 6 of 36

6. Keep the 2020 validation in scope

The Nature Biotechnology study reported spatial profiling of proteins and RNA in fixed tissues, including FFPE material. It described regions spanning roughly one to about five thousand cells and demonstrated panels from dozens of proteins or genes to larger RNA sets. These results establish feasibility under tested conditions, not universal sensitivity for every target, tissue or archive.

Contents · Previous section · Next section

Section 7 of 36

7. Did You Know? A region can be shaped like biology

Illumination masks do not have to be simple circles. They can follow morphology or split a region into marker-positive and marker-negative compartments. That improves question alignment, but boundary accuracy, segmentation and marker reliability become part of the assay. A sophisticated mask is still a hypothesis about tissue organisation.

Contents · Previous section · Next section

Section 8 of 36

8. Choose controls before choosing regions

Include positive controls, negative probes or isotypes, housekeeping targets and blank areas appropriate to the panel. Controls reveal background, tissue quality and assay performance. They should be distributed across the slide rather than placed only where tissue looks easy.

Contents · Previous section · Next section

Section 9 of 36

9. Validate antibodies and RNA probes

An antibody can bind off-target structures; an RNA probe set can cross-hybridise or fail on degraded material. Review vendor and laboratory validation, expected localisation and independent evidence. Tissue-specific optimisation matters because fixation and antigen retrieval can change accessibility.

Contents · Previous section · Next section

Section 10 of 36

10. Record fixation and pre-analytics

Time to fixation, fixative duration, block age, section thickness and storage influence RNA and protein signals. Balance study groups across batches and record these variables. A difference between two archives may reflect tissue handling rather than biology.

Contents · Previous section · Next section

Section 11 of 36

11. Register the image and the illumination

The mask displayed on a morphology image must align with the physical region exposed to light. Save pre- and post-illumination images, fiducials and instrument logs. Small misregistration matters most at thin boundaries or rare-cell compartments.

Contents · Previous section · Next section

Section 12 of 36

12. Collect released tags without carryover

Each illuminated region produces a tag pool that must be transferred to a well. Carryover, evaporation or indexing error can blur regions. Blank collections and alternating strong and weak controls help measure contamination instead of assuming it away.

Contents · Previous section · Next section

Section 13 of 36

13. Count with platform-aware quality checks

nCounter and sequencing readouts have different workflows, dynamic ranges and artefacts. Report read depth or imaging counts, saturation, alignment, deduplication and negative-control behaviour. Do not merge measurements from different panels or readouts merely because both end as integers.

Contents · Previous section · Next section

Section 14 of 36

14. Normalise at the region level

Regions can differ in area, cell number and tissue content. Normalisation by housekeeping genes, area, nuclei or other factors answers different questions. Show raw counts and the chosen denominator, then test whether conclusions survive plausible alternatives.

Contents · Previous section · Next section

Section 15 of 36

15. Practise with an invented ROI table

All values below are fictional.

Contents · Previous section · Next section

Section 16 of 36

16. Separate area from cell composition

A higher regional count may arise because a target is stronger within cells, because more target-positive cells are present, or because the region contains more cells. Morphology markers and deconvolution can help, but conclusions should distinguish composition from per-cell state.

Contents · Previous section · Next section

Section 17 of 36

17. Use compartments without forcing purity

Marker-defined compartments often contain mixed cells and imperfect boundaries. Publish overlap rules, excluded pixels and ambiguous areas. A compartment is an analytical region, not a guarantee that every released tag came from one cell type.

Contents · Previous section · Next section

Section 18 of 36

18. Interpret low counts against background

Subtracting or modelling negative controls can reveal weak signals, but a corrected value near zero remains uncertain. Report detection thresholds and avoid declaring biological absence when tissue quality, probe efficiency or small region size limits sensitivity.

Contents · Previous section · Next section

Section 19 of 36

19. Compare regions within matched specimens

Patient-to-patient or animal-to-animal variation can exceed within-slide differences. Pair compartments inside each specimen where possible, then summarise effects across independent specimens. Hundreds of regions from one block do not equal hundreds of biological replicates.

Contents · Previous section · Next section

Section 20 of 36

20. Correct multiplex exploration

Large panels invite many comparisons. Predeclare primary targets, control false discovery for exploratory tests and validate selected patterns with new specimens or orthogonal assays. A heat map can contain chance structure even when every count is technically valid.

Contents · Previous section · Next section

Section 21 of 36

21. Preserve the selection audit trail

Save who drew each ROI, which image and markers were visible, whether selection was blinded, and any exclusions. This makes the sampling process reproducible and helps readers assess whether visually obvious outcomes influenced molecular discovery.

Contents · Previous section · Next section

Section 22 of 36

22. Challenge single-cell language

The foundational platform could show sensitive detection in small regions, but most ROI counts aggregate molecules. ‘Single-cell sensitivity’ does not mean every result is a single-cell profile. State region size, nuclei count and segmentation evidence beside any cellular interpretation.

Contents · Previous section · Next section

Section 23 of 36

23. Challenge spatial resolution

Mask geometry, optical projection and collection define an address, while diffusion, tissue thickness and cell mixing define biological specificity. A tiny drawn ROI is not automatically a pure measurement. Validate edge behaviour and report the effective scale supported by controls.

Contents · Previous section · Next section

Section 24 of 36

24. Challenge panel completeness

Targeted panels measure selected genes or proteins. Unmeasured biology cannot be treated as absent. Panel design reflects prior hypotheses, reagent availability and budget. Conclusions must stay inside that vocabulary or be tested with broader assays.

Contents · Previous section · Next section

Section 25 of 36

25. Challenge batch and slide effects

Reagent lot, scanner settings, retrieval, operator and slide position can alter counts. Randomise groups, include reference controls and model batches at the specimen level. If condition and batch coincide, computational correction cannot prove which caused the difference.

Contents · Previous section · Next section

Section 26 of 36

26. Challenge clinical translation

Research tissue profiling can generate biomarkers and mechanistic hypotheses. A clinical test requires analytical validation, locked thresholds, representative cohorts, quality systems and evidence that the result improves a defined decision. One spatial association is not a diagnosis or treatment recommendation.

Contents · Previous section · Next section

Section 27 of 36

27. Keep conflicts and platform history visible

The foundational study involved company researchers and described a commercial platform. That does not invalidate the work; it makes transparent disclosure, independent replication, accessible protocols and comparative evaluation especially important.

Contents · Previous section · Next section

Section 28 of 36

28. Compare DSP with whole-transcriptome arrays

DSP uses morphology-guided, often targeted region measurements; capture arrays collect broader transcriptomes across predetermined spatial addresses. DSP can focus deeply on chosen compartments and fixed archives. Arrays may reduce manual region selection. The right method follows the question and tissue.

Contents · Previous section · Next section

Section 29 of 36

29. Build a safe light-mask activity

Students overlay transparent shapes on a printed tissue map, count coloured target symbols inside each mask and compare circular with morphology-following regions. They see how selection, area, mixed contents and background change conclusions without using biological samples or ultraviolet light.

Contents · Previous section · Next section

Section 30 of 36

30. Connect to Primary Science classification

Children can sort symbols, count carefully and explain why two equal areas may contain different numbers of objects. They learn that a category boundary is chosen for a reason and may include uncertain cases.

Contents · Previous section · Next section

Section 31 of 36

31. Strengthen PSLE Science process skills

Pupils can identify the dependent variable, propose a fair denominator and explain why ROI C in the fictional table is unreliable. They practise distinguishing measurement from inference and choosing controls that answer a specific worry.

Contents · Previous section · Next section

Section 32 of 36

32. Extend through Secondary and O-Level Science

Biology supplies tissues and antibodies; Chemistry supplies hybridisation and photocleavable bonds; Physics supplies optics; Mathematics supplies rates and uncertainty; Computing supplies masks and high-dimensional analysis. The topic integrates classroom concepts into one traceable workflow.

Contents · Previous section · Next section

Section 33 of 36

33. Use the topic for school-choice questions

Check official school information for inquiry, biology, chemistry, computing, microscopy and research opportunities. A paper ROI exercise offers authentic reasoning without a commercial instrument. Do not invent equipment access, programmes, admission rules or outcomes.

Contents · Previous section · Next section

Section 34 of 36

34. See the career network

Digital spatial profiling connects pathology, histotechnology, molecular biology, optical engineering, bioinformatics, statistics, quality assurance, clinical research and data governance. Its strength comes from cooperation between people who see the tissue and people who count, model and challenge tags.

Contents · Previous section · Next section

Section 35 of 36

35. Ask better questions at home

What exactly was illuminated? How many cells were inside? Which controls define background? Were regions chosen before outcomes were known? Is the panel targeted or genome-wide? How many independent specimens agree? These questions travel well to diagnostic claims and product marketing.

Contents · Previous section · Next section

Section 36 of 36

36. Finish with an evidence sentence

A defensible conclusion names tissue, fixation, panel, morphology markers, ROI rule, area, nuclei estimate, readout, controls, normalisation and independent specimens. It reports a region-level association at the supported scale. It does not silently convert tag counts into single-cell causation or clinical advice.

Pre-registration can define specimen eligibility, fixation bounds, panel version, morphology markers, ROI types, number and size range, segmentation rules, background controls, normalisation, primary contrasts and specimen-level statistics. Regions selected for discovery after viewing the tissue remain valuable, but their status should be explicit and their findings confirmed on new material.

ROI sampling needs a denominator. Record eligible tissue area, excluded damage, number of candidate regions, selected regions and the rule that chose them. If a tumour contains many ordinary areas and one unusual immune aggregate, selecting only the aggregate answers a focused question; it does not estimate the average tumour.

Panel governance matters. Keep target sequences or antibody identifiers, lots, validation evidence, probe counts, expected localisation and any retired targets. Changes between panel versions can make a longitudinal dataset look biological. Version labels belong in the analysis table, not only in a laboratory notebook.

Background should be measured at several levels: negative probes or isotypes, blank collection wells, off-tissue regions and biological negatives where defensible. These controls answer different questions. A single universal subtraction can create false precision when background varies by region size, tissue autofluorescence or collection batch.

Region geometry should accompany counts. Release area, nuclei, perimeter, marker-positive area and tissue class. A long thin boundary ROI behaves differently from a compact circle even when area matches. Edge-to-area ratio can influence contamination and should be considered in narrow compartments.

Normalisation is a scientific choice. Per-area scaling asks about molecular density; per-nucleus scaling approximates average cell-associated signal; housekeeping scaling assumes reference stability. Present raw counts and at least one sensitivity analysis. If the biological ranking reverses with a reasonable denominator, the evidence is not settled.

Deconvolution and cell-type scoring are models trained on references. Their output depends on which cell classes and genes are available. Validate important composition claims with morphology or independent assays, and report uncertainty. A mixed ROI should not be rewritten as an exact list of pure cells merely because software returns proportions.

Statistical analyses should use independent specimens as the generalisation unit. Multiple ROIs increase within-specimen coverage, but they share fixation, genetics and history. Pairing regions within a specimen and modelling nested variation are usually more honest than treating every illuminated mask as independent.

Reproducibility requires raw images, mask files, coordinates, collection order, instrument logs, raw tag counts, controls, panel manifest, normalisation, exclusions, code and software versions. Patient-linked metadata require lawful minimisation and access controls. Share enough to audit the assay without exposing identities.

A cautious conclusion might read: ‘In independently processed fixed specimens, predeclared morphology-guided regions showed higher target-tag counts after background and area or nuclei adjustment, with consistent direction across specimens.’ It should name panel and region scale. It should not claim single-cell mechanism, diagnosis or treatment response without the required evidence.

Power planning should reflect specimen-to-specimen variation and the number of primary contrasts, not merely the large number of regions an instrument can collect. A pilot can estimate background, within-block variation and plausible effect sizes. Additional ROIs improve coverage of heterogeneous tissue, but additional independent specimens are what strengthen generalisation.

Rare-cell questions need especially careful masks. A marker-positive compartment may be small, irregular and vulnerable to registration error. Report minimum area, nuclei estimate, marker threshold and whether a second reviewer agreed. When too few cells contribute, pool only according to a predeclared biological rule and retain the original region records.

Cross-platform comparisons should use matched questions rather than headline plex or resolution. Compare the same tissue, targets, controls and biological units where possible. A broad whole-transcriptome assay and a targeted high-sensitivity panel answer different questions. Performance claims should name the tested sample and metric instead of ranking technologies universally.

Figures should place the unedited morphology image, ROI masks, raw control counts, target counts, normalised values and specimen-level estimates in one navigable sequence. Mark tiny or low-quality regions instead of hiding them. Readers should be able to reconstruct why each area was measured and how a biological conclusion emerged from tag collections rather than from colour alone.

Contents · Previous section · Continue to the Science Learning Hub