Why Science? | CellComm, Receptor-to-Regulator Paths and Developmental Crosstalk Evidence

Section 1 of 36

1. Start with developmental crosstalk

CellComm was built to ask how cell types in a developing microenvironment may influence one another. It begins with cell-type expression and curated ligand–receptor relationships, then follows candidate receptor signals towards transcriptional regulators. The result is a traceable hypothesis chain, not a recording of molecules moving through tissue.

Contents · Next section

Section 2 of 36

2. Keep the 2022 paper boundary

The primary Nature Cell Biology study analysed the aorta–gonad–mesonephros environment during haematopoietic stem and progenitor cell emergence. Its validations in zebrafish, mouse and human models support conclusions in that developmental programme. They do not automatically validate every CellComm prediction in another organ, age or disease.

Contents · Previous section · Next section

Section 3 of 36

3. Find co-expressed ligand–receptor pairs

CellComm identifies candidate communication by examining curated ligand–receptor pairs whose members are expressed in source and receiver cell types. Co-expression supplies compatibility. It cannot show that a ligand was processed or released, that a receptor reached the cell surface, or that binding occurred.

Contents · Previous section · Next section

Section 4 of 36

4. Use spatial evidence when available

When spatial transcriptomics is available, potential interactions can be constrained by location. Compatible cell types that occupy the same niche offer stronger opportunity than distant groups. Spatial proximity remains opportunity, not molecular transfer, and the appropriate distance depends on whether a signal is contact-bound or diffusible.

Contents · Previous section · Next section

Section 5 of 36

5. Search receptor-to-regulator paths

CellComm uses an optimisation procedure to search a weighted protein-interaction network for paths from cell-surface receptors to downstream transcriptional regulators. A high-ranking path connects prior molecular knowledge with sample expression. It does not directly observe every intermediate protein or biochemical step.

Contents · Previous section · Next section

Section 6 of 36

6. Prioritise regulators by regulons

Candidate transcriptional regulators are evaluated using enrichment of their regulons in cell-type-specific expression signatures. This adds a downstream-response layer. Regulon enrichment is still an inference based on gene sets, not direct evidence of regulator protein activity or DNA binding.

Contents · Previous section · Next section

Section 7 of 36

7. Separate inference layers

Ligand and receptor RNA, spatial opportunity, network paths and regulon enrichment are different forms of evidence. Label them separately. The attraction of CellComm is that the layers can be connected; the scientific discipline is refusing to merge them into one unqualified claim of communication.

Contents · Previous section · Next section

Section 8 of 36

8. Create a specimen manifest

Record organism, developmental stage, tissue, condition, biological specimen, assay, preprocessing, cell annotations and spatial technology. Development changes quickly, so stage and anatomical compartment are part of the question rather than optional metadata.

Contents · Previous section · Next section

Section 9 of 36

9. Audit cell identities

Check markers, doublets, transitional populations and specimen balance for every source and receiver group. The AGM contains closely related developmental states. A fragile label can create an apparently specific route that disappears when cells are grouped more conservatively.

Contents · Previous section · Next section

Section 10 of 36

10. Freeze the ligand–receptor resource

Archive the exact pair table, species, identifiers, direction and complex rules. Database updates can add or remove eligible edges without any change to the tissue. A versioned pair universe keeps knowledge-base change separate from biological change.

Contents · Previous section · Next section

Section 11 of 36

11. Freeze the protein network

Preserve the protein–protein interaction resource, weights, direction assumptions and filters used for receptor-to-regulator path finding. Highly studied pathways are richer than neglected ones, so network coverage can shape which regulators appear reachable.

Contents · Previous section · Next section

Section 12 of 36

12. Freeze the regulon resource

Save the transcriptional regulators, target sets and enrichment background. Report target-set size for leading regulators. A regulator with a large, broadly annotated regulon can be enriched more readily than one with sparse evidence.

Contents · Previous section · Next section

Section 13 of 36

13. Protect biological replication

Cells are nested inside embryos, animals, donors or samples. Estimate route support within each independent specimen and report prevalence. Thousands of single cells improve resolution but do not create thousands of independent developmental systems.

Contents · Previous section · Next section

Section 14 of 36

14. Archive the full path

Keep expression summaries, cell labels, spatial coordinates, pair mappings, path-search settings, network edges, regulon tests, thresholds and complete ranked results. A network figure cannot reveal which layer supplied the decisive evidence.

Contents · Previous section · Next section

Section 15 of 36

15. Practise with an invented evidence table

This fictional table is for reasoning practice, not CellComm output. Identify the weakest link before choosing an experiment.

Contents · Previous section · Next section

Section 16 of 36

16. Read pair abundance cautiously

A highly expressed ligand–receptor pair can dominate a network while remaining biologically inactive. Publish distributions and detection fractions rather than means alone. Check whether the signal is spread across cells and specimens or driven by a small, noisy subset.

Contents · Previous section · Next section

Section 17 of 36

17. Inspect cell-type specificity

Compare a route with other possible source and receiver groups. A pair expressed everywhere may be important but is not specific evidence for one developmental interaction. Specificity should be demonstrated, not inferred from an attractive cell-pair label.

Contents · Previous section · Next section

Section 18 of 36

18. Inspect every intermediate node

For a receptor-to-regulator path, list the intermediate proteins, edge provenance and sample expression. Missing expression does not always mean an inactive protein, but repeated gaps weaken a sample-specific path and should remain visible.

Contents · Previous section · Next section

Section 19 of 36

19. Read regulon enrichment transparently

Publish the target genes, background universe, statistic and correction. Broad developmental signatures can enrich many regulators. A useful regulator claim shows which genes carry the result and whether the enrichment repeats across specimens.

Contents · Previous section · Next section

Section 20 of 36

20. Compare pathways, not just ranks

Two receptor-to-regulator paths can reach the same regulator through different molecular intermediates. Compare pathway plausibility, expression support and experimental discriminability. The top numerical rank is not automatically the best biological explanation.

Contents · Previous section · Next section

Section 21 of 36

21. Use developmental time

A route proposed during cell emergence should match stage order. Measure ligand availability, receptor-proximal events and regulator response across justified time points. A single snapshot can mix cells that occupy different moments along a developmental trajectory.

Contents · Previous section · Next section

Section 22 of 36

22. Audit spatial constraints

Repeat the analysis with biologically reasonable neighbourhood definitions and anatomical compartments. Contact-dependent signals should require close opportunity; diffusible factors may act farther away. A single generic radius cannot represent every mediator.

Contents · Previous section · Next section

Section 23 of 36

23. Audit path-search choices

Vary planned edge confidence, maximum path length and weight settings. Long paths can connect many receptors to familiar regulators by chance. Stable conclusions should persist under defensible alternatives and should not depend on one low-confidence bridge.

Contents · Previous section · Next section

Section 24 of 36

24. Audit regulon sensitivity

Compare compatible regulator resources or high-confidence target subsets. Mark predictions that vanish when uncertain targets are removed. Resource sensitivity is not a failure; it tells experimentalists which evidence link is weakest.

Contents · Previous section · Next section

Section 25 of 36

25. Compare simpler baselines

Compare CellComm with ligand–receptor expression alone and with regulator enrichment alone. The integrated method adds value when its linked path creates a stable, discriminating prediction that the two components do not reveal separately.

Contents · Previous section · Next section

Section 26 of 36

26. Publish informative negatives

Show routes rejected by absent spatial opportunity, unstable cell labels, missing receptor protein, weak regulator enrichment or specimen disagreement. These cases demonstrate how the evidence chain filters hypotheses instead of simply drawing more edges.

Contents · Previous section · Next section

Section 27 of 36

27. Design a staged perturbation

Block a leading ligand or receptor and measure an early pathway event, the nominated regulator and a later developmental outcome. Include a competing route. This separates molecular specificity from a general stress or maturation effect.

Contents · Previous section · Next section

Section 28 of 36

28. Write a bounded CellComm claim

Prefer: ‘CellComm prioritised this ligand–receptor–regulator path across three specimens, with spatial and regulon support.’ Avoid saying CellComm proved signal transfer. The bounded statement preserves the method’s real contribution and makes validation straightforward.

Contents · Previous section · Next section

Section 29 of 36

29. Turn development into a classroom chain

Give students cards for source cell, ligand, receptor, pathway, regulator and target genes. Ask which cards are observations, database links or predictions. Then remove one card and decide how the claim changes.

Contents · Previous section · Next section

Section 30 of 36

30. Begin with Primary Science habits

Observation, classification and careful sequencing are the roots of a developmental evidence chain. Young learners can already distinguish what happened first, what was measured and which explanation remains possible.

Contents · Previous section · Next section

Section 31 of 36

31. Use PSLE Science for fair tests

Students can compare two routes and choose one variable to change. Predicting the early and late outcomes of blocking a ligand practises cause-and-effect reasoning without requiring advanced genomics.

Contents · Previous section · Next section

Section 32 of 36

32. Connect Secondary and O-Level Science

Cells, receptors, coordination, gene expression and development provide the biological language. CellComm adds networks and evidence integration. The aim is to evaluate a pathway, not memorise software names.

Contents · Previous section · Next section

Section 33 of 36

33. Let Computing support the path

Computing contributes graphs, optimisation, weights, enrichment, version control and reproducible code. Biology determines whether nodes and directions are meaningful. A correct shortest path can still be biologically implausible.

Contents · Previous section · Next section

Section 34 of 36

34. Choose opportunities by verified fit

For school choices or science enrichment, check current official details and look for sustained opportunities in biology, data handling, computing and research. Programme names do not guarantee admission, equipment or a particular outcome.

Contents · Previous section · Next section

Section 35 of 36

35. See connected career families

The reasoning connects to developmental biology, haematology, bioinformatics, regenerative medicine, statistics, research software and experimental science. Education routes vary, so verify current institutional and employer requirements directly.

Contents · Previous section · Next section

Section 36 of 36

36. Finish with the chain intact

A defensible CellComm conclusion names the specimens, pair resource, spatial rule, protein network, path settings and regulon evidence, then separates the reconstructed route from molecular and developmental validation. The path is valuable because every link can be tested.

CellComm is most useful when its output is read as a chain of evidence rather than a single colourful network. The first link is cell-type-level ligand–receptor co-expression. The second is a candidate path through a weighted protein–protein interaction network from a receptor towards a transcriptional regulator. The third is regulon enrichment, which asks whether genes associated with that regulator are supported in the receiver state. Each link answers a different question, and none directly measures signal transmission.

Did you know? The CellComm method was presented in a developmental study spanning zebrafish, mouse and human evidence. That breadth strengthened the paper’s claims about its reported haematopoietic system, but it does not make every CellComm prediction in a new tissue experimentally validated. Reuse the method; do not copy the validation status.

Start by preserving the biological unit. Cells from one specimen are not independent replicates of the treatment. Summarise candidate routes by donor, embryo, animal or other true unit, then ask whether the route recurs. A high score driven by one large specimen is a lead for inspection, not a population-level result.

Co-expression also needs a visible denominator. Report the fraction of cells expressing each partner, the number of eligible cells and the aggregation rule used for each cluster. A ligand detected in three cells out of three has a different evidence base from one detected in three cells out of three thousand. Keep low-count groups and dropout sensitivity in the table.

Spatial constraints are optional only when spatial evidence is unavailable. If coordinates or anatomical regions exist, define a contact or diffusion opportunity before ranking the biological story. Membrane-bound interactions usually require a tighter neighbourhood than secreted signals. Repeat the result across reasonable distance thresholds and show whether the claimed source–receiver pairing survives.

The receptor-to-regulator search depends on a prior network. Freeze its release, identifier namespace, species mapping and edge weights. Inspect the complete path behind a headline route. A short path composed of well-supported edges and a long path assembled from generic hubs should not receive the same prose. Hub proteins can connect many endpoints without establishing context-specific activity.

Regulon enrichment provides response-side evidence, but a regulon is itself an inference resource. Show the target genes supporting a regulator, their direction and their specimen prevalence. Recalculate after removing one dominant target. If the ranking collapses, report that fragility instead of presenting the regulator as a stable endpoint.

Validation should climb a ladder. First confirm cell identities and expression in independent specimens. Next localise ligand and receptor proteins. Then assay a receptor-proximal event or regulator activation. Finally perturb the ligand, receptor or route and test the predicted receiver programme. Each rung addresses a different gap; one successful immunostain does not validate the whole path.

Negative controls can be beautifully specific. Shuffle sender labels within specimen while preserving abundance. Test a receptor-free receiver group. Compare the nominated regulator with an expression-matched alternative. Use an anatomical compartment where the pair should lack opportunity. A method that keeps ranking the same route under these controls may be following broad expression or network connectivity.

For students, CellComm joins familiar scientific habits. Primary Science begins with observation and classification. PSLE Science strengthens fair comparison. Secondary and O-Level Biology add receptors, regulation and development. Computing adds graphs, shortest paths and reproducible data handling. The advanced workflow still depends on the simple question: which parts were observed, and which were inferred?

A good report includes an evidence ledger for each headline path: ligand expression, receptor expression, source and receiver counts, spatial opportunity, prior-network path, regulator targets, replicate support, protein evidence and perturbation status. Mark each field as observed, inferred, unsupported or not tested. This makes the result navigable and prevents an absent experiment from being mistaken for a negative result.

Compare CellComm with a simpler baseline. If ligand–receptor co-expression alone produces the same ranking, the downstream path may add explanation rather than discrimination. If regulator evidence changes the priority, show exactly why. The comparison helps readers understand what the extra modelling contributes.

Version the analysis as carefully as the data. Archive the CellComm/FUSCA code revision, dependency environment, interaction resource, network object, regulon set, parameters and random seeds. A future database update may reroute a receptor to another regulator. Provenance tells readers that the analytical knowledge changed; the original sample did not.

The bounded conclusion is already valuable: CellComm can prioritise traceable ligand–receptor and receptor-to-regulator hypotheses for developmental or other single-cell studies. It does not directly observe ligand release, receptor binding, biochemical activation, direction of information flow or causal regulation. Keeping those verbs precise turns a network into a better experiment plan.

Before publication, run one final pathway audit. Trace every displayed arrow to either the sample data, a named prior resource or a clearly labelled calculation. Remove isolated nodes that do not affect the interpretation. Check that the article never shifts from “co-expressed”, “connected” or “prioritised” to “activated” without additional evidence. This small language review is a powerful defence against accidental overclaiming.

Plan the next experiment around the weakest link rather than the most familiar technique. If spatial opportunity is uncertain, improve localisation. If the receptor-to-regulator path depends on a generic hub, assay a proximal event. If the regulon is unstable, measure targets after controlled perturbation. Science progresses fastest when the validation step is chosen to discriminate among explanations.

Finally, retain alternatives. A receiver programme may be produced by several receptors, and a ligand may act through several receivers. Rank plausible competing paths and state what observation would separate them. A network becomes scientifically useful when it supports a decision about what to test next, not merely when it contains many arrows.

Contents · Previous section · Continue to the Science Learning Hub