Why Science? | cydar, Hypersphere Counts and Spatial FDR in Mass Cytometry

Section 1 of 36

1. Begin with protein-rich single cells

Mass cytometry can measure many protein markers per cell using metal-isotope-labelled antibodies. It reveals phenotypes that may be missed by a few conventional fluorescence channels.

cydar places hyperspheres through high-dimensional marker space and counts how many cells from each biological sample fall inside each local region.

Archive samples, labels, preprocessing, feature choices, graph parameters, random seeds, software versions and every exclusion so another analyst can reproduce the decision path.

Contents · Next section

Section 2 of 36

2. Understand the measurement boundary

The assay observes chosen epitopes after staining and processing. It does not directly measure every transcript, function or spatial interaction.

Mass cytometry measures metal-tagged antibodies, so the geometry reflects the selected protein-marker panel, staining quality and preprocessing—not the entire cell state.

Keep biological samples as the unit of replication. Thousands of cells improve resolution, but they do not turn one donor, animal or culture into thousands of independent experiments.

Contents · Previous section · Next section

Section 3 of 36

3. Avoid cluster-first blindness

A condition-associated population may occupy only part of a broad cluster or lie between familiar labels. Local testing can search without committing to one partition.

Negative-binomial models can test hypersphere counts with design covariates while retaining sample-level variability instead of treating cells as replicates.

Use negative controls, label permutations and reasonable alternative settings. A trustworthy signal should exceed a design-matched null and survive choices that were not selected for attractiveness.

Contents · Previous section · Next section

Section 4 of 36

4. Define marker space

Choose lineage and functional markers, transformations and scaling. Distances are meaningful only when the panel and preprocessing are technically sound.

Nearby hyperspheres overlap; spatial false-discovery control acknowledges that discoveries cover volumes of marker space rather than independent isolated bins.

Report effect direction, uncertainty, sample support and sensitivity—not only a colourful embedding or adjusted value. The display is a map of evidence, not the evidence itself.

Contents · Previous section · Next section

Section 5 of 36

5. Place hyperspheres

A hypersphere is a neighbourhood around a point in high-dimensional marker space. Cells inside share proximity under the selected distance and radius.

A significant hypersphere can be merged or visualised with neighbours for interpretation, but its biological name requires markers and external knowledge.

Create an evidence card naming the contrast, measured modality, statistical unit, result, validation status, unresolved alternatives and the narrowest defensible claim.

Contents · Previous section · Next section

Section 6 of 36

6. Count cells by sample

For each hypersphere, cydar counts cells from every biological sample. The resulting matrix supports replicate-aware abundance testing.

Thresholds controlling low-density hyperspheres, transformation and marker scaling set sensitivity to rare versus noisy regions and must be reported.

Write association, enrichment or differential abundance when that is what was tested. Reserve cause, mechanism, diagnosis and treatment benefit for experiments that directly support those claims.

Contents · Previous section · Next section

Section 7 of 36

7. Read the 2017 evidence in scope

Lun and colleagues introduced the strategy in Nature Methods for differential abundance in mass cytometry and released it through Bioconductor.

cydar places hyperspheres through high-dimensional marker space and counts how many cells from each biological sample fall inside each local region.

Archive samples, labels, preprocessing, feature choices, graph parameters, random seeds, software versions and every exclusion so another analyst can reproduce the decision path.

Contents · Previous section · Next section

Section 8 of 36

8. Build the sample manifest

Record donor, condition, batch, tissue, processing time, acquisition run, panel lot, total events and exclusions before fitting models.

Mass cytometry measures metal-tagged antibodies, so the geometry reflects the selected protein-marker panel, staining quality and preprocessing—not the entire cell state.

Keep biological samples as the unit of replication. Thousands of cells improve resolution, but they do not turn one donor, animal or culture into thousands of independent experiments.

Contents · Previous section · Next section

Section 9 of 36

9. Audit antibody quality

Titration, specificity, spillover-like effects, oxidation and lot variation can alter marker geometry. Computational testing cannot rescue a poor panel.

Negative-binomial models can test hypersphere counts with design covariates while retaining sample-level variability instead of treating cells as replicates.

Use negative controls, label permutations and reasonable alternative settings. A trustworthy signal should exceed a design-matched null and survive choices that were not selected for attractiveness.

Contents · Previous section · Next section

Section 10 of 36

10. Transform intensities consistently

Arcsinh transformation and cofactor choices affect distances. Apply a justified rule to all samples and document it.

Nearby hyperspheres overlap; spatial false-discovery control acknowledges that discoveries cover volumes of marker space rather than independent isolated bins.

Report effect direction, uncertainty, sample support and sensitivity—not only a colourful embedding or adjusted value. The display is a map of evidence, not the evidence itself.

Contents · Previous section · Next section

Section 11 of 36

11. Remove technical events

Beads, debris, doublets and dead cells can form dense artefactual regions. Gate with predeclared quality criteria and keep audit plots.

A significant hypersphere can be merged or visualised with neighbours for interpretation, but its biological name requires markers and external knowledge.

Create an evidence card naming the contrast, measured modality, statistical unit, result, validation status, unresolved alternatives and the narrowest defensible claim.

Contents · Previous section · Next section

Section 12 of 36

12. Choose analysis markers

Exclude channels that encode batch or technical controls from phenotype distance unless there is a clear reason to include them.

Thresholds controlling low-density hyperspheres, transformation and marker scaling set sensitivity to rare versus noisy regions and must be reported.

Write association, enrichment or differential abundance when that is what was tested. Reserve cause, mechanism, diagnosis and treatment benefit for experiments that directly support those claims.

Contents · Previous section · Next section

Section 13 of 36

13. Set hypersphere radius

A small radius resolves narrow phenotypes but yields sparse counts; a large radius gains stability while blending neighbouring states.

cydar places hyperspheres through high-dimensional marker space and counts how many cells from each biological sample fall inside each local region.

Archive samples, labels, preprocessing, feature choices, graph parameters, random seeds, software versions and every exclusion so another analyst can reproduce the decision path.

Contents · Previous section · Next section

Section 14 of 36

14. Filter low-density regions

Hyperspheres with too few cells provide weak count information and increase the testing burden. State density thresholds and their effect on rare populations.

Mass cytometry measures metal-tagged antibodies, so the geometry reflects the selected protein-marker panel, staining quality and preprocessing—not the entire cell state.

Keep biological samples as the unit of replication. Thousands of cells improve resolution, but they do not turn one donor, animal or culture into thousands of independent experiments.

Contents · Previous section · Next section

Section 15 of 36

15. Practise with a fictional cydar table

This classroom table is invented and not a result from the cydar paper.

Negative-binomial models can test hypersphere counts with design covariates while retaining sample-level variability instead of treating cells as replicates.

Use negative controls, label permutations and reasonable alternative settings. A trustworthy signal should exceed a design-matched null and survive choices that were not selected for attractiveness.

Contents · Previous section · Next section

Section 16 of 36

16. Fit a count model

Model hypersphere counts across samples with condition and relevant covariates. Overdispersion represents variability beyond a simple Poisson count.

Nearby hyperspheres overlap; spatial false-discovery control acknowledges that discoveries cover volumes of marker space rather than independent isolated bins.

Report effect direction, uncertainty, sample support and sensitivity—not only a colourful embedding or adjusted value. The display is a map of evidence, not the evidence itself.

Contents · Previous section · Next section

Section 17 of 36

17. Read log fold change

The coefficient describes abundance change in the analysed samples under the model and normalisation. It is not automatically a tissue-wide absolute change.

A significant hypersphere can be merged or visualised with neighbours for interpretation, but its biological name requires markers and external knowledge.

Create an evidence card naming the contrast, measured modality, statistical unit, result, validation status, unresolved alternatives and the narrowest defensible claim.

Contents · Previous section · Next section

Section 18 of 36

18. Control spatial false discoveries

Overlapping hyperspheres create related tests. Spatial FDR weights discoveries by covered high-dimensional volume instead of pretending each centre is independent.

Thresholds controlling low-density hyperspheres, transformation and marker scaling set sensitivity to rare versus noisy regions and must be reported.

Write association, enrichment or differential abundance when that is what was tested. Reserve cause, mechanism, diagnosis and treatment benefit for experiments that directly support those claims.

Contents · Previous section · Next section

Section 19 of 36

19. Map related discoveries

Visualisations can project significant hyperspheres or connect overlaps. Projection aids navigation but can distort high-dimensional distances.

cydar places hyperspheres through high-dimensional marker space and counts how many cells from each biological sample fall inside each local region.

Archive samples, labels, preprocessing, feature choices, graph parameters, random seeds, software versions and every exclusion so another analyst can reproduce the decision path.

Contents · Previous section · Next section

Section 20 of 36

20. Name populations cautiously

Use marker combinations to describe regions, then compare with reference biology. Mixed or transitional phenotypes deserve descriptive language.

Mass cytometry measures metal-tagged antibodies, so the geometry reflects the selected protein-marker panel, staining quality and preprocessing—not the entire cell state.

Keep biological samples as the unit of replication. Thousands of cells improve resolution, but they do not turn one donor, animal or culture into thousands of independent experiments.

Contents · Previous section · Next section

Section 21 of 36

21. Inspect every sample

Plot counts and marker distributions by sample. A significant model coefficient can still be dominated by one specimen or batch.

Negative-binomial models can test hypersphere counts with design covariates while retaining sample-level variability instead of treating cells as replicates.

Use negative controls, label permutations and reasonable alternative settings. A trustworthy signal should exceed a design-matched null and survive choices that were not selected for attractiveness.

Contents · Previous section · Next section

Section 22 of 36

22. Test radius sensitivity

Repeat nearby radii and filtering thresholds. Robust biology should persist as a coherent marker-defined region, not one isolated coordinate.

Nearby hyperspheres overlap; spatial false-discovery control acknowledges that discoveries cover volumes of marker space rather than independent isolated bins.

Report effect direction, uncertainty, sample support and sensitivity—not only a colourful embedding or adjusted value. The display is a map of evidence, not the evidence itself.

Contents · Previous section · Next section

Section 23 of 36

23. Use panel and batch controls

Reference samples, shared controls and balanced runs help separate condition changes from staining or acquisition drift.

A significant hypersphere can be merged or visualised with neighbours for interpretation, but its biological name requires markers and external knowledge.

Create an evidence card naming the contrast, measured modality, statistical unit, result, validation status, unresolved alternatives and the narrowest defensible claim.

Contents · Previous section · Next section

Section 24 of 36

24. Compare with manual gating

Expert gates provide a familiar baseline. cydar is most informative when it finds reproducible local changes without abandoning interpretable markers.

Thresholds controlling low-density hyperspheres, transformation and marker scaling set sensitivity to rare versus noisy regions and must be reported.

Write association, enrichment or differential abundance when that is what was tested. Reserve cause, mechanism, diagnosis and treatment benefit for experiments that directly support those claims.

Contents · Previous section · Next section

Section 25 of 36

25. Compare with Milo

Milo uses graph neighbourhoods in single-cell data, often transcriptomic; cydar was designed around hyperspheres and count testing in high-dimensional cytometry.

cydar places hyperspheres through high-dimensional marker space and counts how many cells from each biological sample fall inside each local region.

Archive samples, labels, preprocessing, feature choices, graph parameters, random seeds, software versions and every exclusion so another analyst can reproduce the decision path.

Contents · Previous section · Next section

Section 26 of 36

26. Compare with scCODA

scCODA models predefined compositional categories. cydar searches local marker-space regions and therefore faces overlapping-test geometry.

Mass cytometry measures metal-tagged antibodies, so the geometry reflects the selected protein-marker panel, staining quality and preprocessing—not the entire cell state.

Keep biological samples as the unit of replication. Thousands of cells improve resolution, but they do not turn one donor, animal or culture into thousands of independent experiments.

Contents · Previous section · Next section

Section 27 of 36

27. Build a hypersphere evidence card

Record marker centre, radius, cell and sample counts, model contrast, fold change, spatial FDR, sensitivity and validation.

Negative-binomial models can test hypersphere counts with design covariates while retaining sample-level variability instead of treating cells as replicates.

Use negative controls, label permutations and reasonable alternative settings. A trustworthy signal should exceed a design-matched null and survive choices that were not selected for attractiveness.

Contents · Previous section · Next section

Section 28 of 36

28. Design the next experiment

Repeat the panel with balanced batches and predeclared marker combinations, then add sorting or functional assays for the candidate phenotype.

Nearby hyperspheres overlap; spatial false-discovery control acknowledges that discoveries cover volumes of marker space rather than independent isolated bins.

Report effect direction, uncertainty, sample support and sensitivity—not only a colourful embedding or adjusted value. The display is a map of evidence, not the evidence itself.

Contents · Previous section · Next section

Section 29 of 36

29. Connect cydar to school science

Antibodies, controls, fair comparisons and repeated measurements show how a laboratory signal becomes evidence only through careful design.

A significant hypersphere can be merged or visualised with neighbours for interpretation, but its biological name requires markers and external knowledge.

Create an evidence card naming the contrast, measured modality, statistical unit, result, validation status, unresolved alternatives and the narrowest defensible claim.

Contents · Previous section · Next section

Section 30 of 36

30. Try a sphere-count activity

Place coloured points on a two-marker grid, count neighbours inside overlapping circles and discuss why several significant circles may describe one population.

Thresholds controlling low-density hyperspheres, transformation and marker scaling set sensitivity to rare versus noisy regions and must be reported.

Write association, enrichment or differential abundance when that is what was tested. Reserve cause, mechanism, diagnosis and treatment benefit for experiments that directly support those claims.

Contents · Previous section · Next section

Section 31 of 36

31. Connect to mathematics

High-dimensional distance, spheres, count distributions and multiple-testing control give geometric meaning to statistical inference.

cydar places hyperspheres through high-dimensional marker space and counts how many cells from each biological sample fall inside each local region.

Archive samples, labels, preprocessing, feature choices, graph parameters, random seeds, software versions and every exclusion so another analyst can reproduce the decision path.

Contents · Previous section · Next section

Section 32 of 36

32. Connect to computing and AI literacy

Automated local searches can reveal overlooked regions, but the output inherits panel design, transformations and sample quality.

Mass cytometry measures metal-tagged antibodies, so the geometry reflects the selected protein-marker panel, staining quality and preprocessing—not the entire cell state.

Keep biological samples as the unit of replication. Thousands of cells improve resolution, but they do not turn one donor, animal or culture into thousands of independent experiments.

Contents · Previous section · Next section

Section 33 of 36

33. Connect to school and career pathways

Students drawn to cells and instruments can explore immunology, biotechnology, laboratory medicine, statistics, engineering or computational biology.

Negative-binomial models can test hypersphere counts with design covariates while retaining sample-level variability instead of treating cells as replicates.

Use negative controls, label permutations and reasonable alternative settings. A trustworthy signal should exceed a design-matched null and survive choices that were not selected for attractiveness.

Contents · Previous section · Next section

Section 34 of 36

34. Create a family evidence habit

When a study reports an immune-cell increase, ask which markers defined the cells, how many biological samples were tested and how technical batches were controlled.

Nearby hyperspheres overlap; spatial false-discovery control acknowledges that discoveries cover volumes of marker space rather than independent isolated bins.

Report effect direction, uncertainty, sample support and sensitivity—not only a colourful embedding or adjusted value. The display is a map of evidence, not the evidence itself.

Contents · Previous section · Next section

Section 35 of 36

35. Use precise cydar language

Say differentially abundant hypersphere, marker-defined region and spatial FDR. Avoid new cell type or mechanism unless orthogonal evidence supports it.

A significant hypersphere can be merged or visualised with neighbours for interpretation, but its biological name requires markers and external knowledge.

Create an evidence card naming the contrast, measured modality, statistical unit, result, validation status, unresolved alternatives and the narrowest defensible claim.

Contents · Previous section · Next section

Section 36 of 36

36. Finish with a panel-to-validation bundle

Deliver panel metadata, QC gates, transformations, hypersphere settings, sample models, spatial FDR, marker maps, sensitivity results and functional follow-up.

Thresholds controlling low-density hyperspheres, transformation and marker scaling set sensitivity to rare versus noisy regions and must be reported.

Write association, enrichment or differential abundance when that is what was tested. Reserve cause, mechanism, diagnosis and treatment benefit for experiments that directly support those claims.

## Source date and scope note Primary papers and official software documentation were checked for this article on 10 October 2026. Software interfaces and recommendations can change, so readers should consult the linked current documentation before reproducing an analysis.

## Final reader checklist Before accepting a claim, confirm the biological sample count, condition definition, measured modality, neighbourhood construction, statistical unit, uncertainty, software version, negative controls, sensitivity checks, independent validation and the exact boundary between association and causation.

Contents · Previous section · Continue to the Science Learning Hub