eduKateSG · Why Science?
Follow every graph edge back to its source, species, relation type and release before turning a connected path into a biological story
Choose the closest route, then return to the full index whenever you need the wider evidence chain.
Science learning becomes powerful when students can tell a measured pattern from a model, a statistical decision from a mechanism, and a useful lead from a finished conclusion. The Monarch Initiative develops open ontologies, semantic data models, tools and a knowledge graph that integrates phenotype, gene, disease, variant and organism information from many biomedical resources. Its official Knowledge Graph overview describes a combined resource built from dozens of sources and ontologies. Integration makes relationships easier to search and compare across species, but the graph is not one experiment and an edge is not automatically causal. Responsible use retains entity identifiers, relation type, source database, supporting publications, species, evidence and release context. This guide supports Primary Science, PSLE Science, Secondary Science, O-Level Science and STEM education as a longform evidence-reading route; it is not medical advice, a laboratory protocol, an admission promise or a career guarantee.
Related eduKate reading: Human Phenotype Ontology and structured clinical phenotypes; ORDO and structured rare-disease concepts; GWAS Catalog and curated variant–trait associations; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 11 October 2026: Official Monarch Knowledge Graph overview; Official Monarch citation guide; The Monarch Initiative in 2024; Official Monarch Initiative repositories; 2026 Singapore–Cambridge O-Level Biology syllabus; MOE G2/G3 Lower Secondary Science teaching and learning syllabus; 2026 MOE G2 Computing syllabus. Package interfaces and recommendations can change, so reproduce an analysis only with the documentation for the installed release.
| Stage | Question | Evidence | Responsible output |
|---|---|---|---|
| Ask | Which biological question and entity types are relevant? | Phenotype, gene, disease, variant and taxon identifiers | A scoped graph query |
| Trace | Where did each node and edge originate? | Source database, evidence, relation and release | A provenance-aware path |
| Compare | Are cross-species phenotypes and genes genuinely comparable? | Ontology alignment, homology and experimental context | A ranked mechanistic hypothesis |
| Validate | What independent work can test the proposed path? | Literature, functional experiments and expert review | A bounded conclusion, not graph-derived causation |
Inside this guide
1–12 · Foundations and design
- 1. Start with an integrated graph
- 2. Understand nodes and edges
- 3. Keep identifiers with names
- 4. Treat integration as a transformation
- 5. Read the source inventory
- 6. Distinguish asserted from inferred links
- 7. Do not count paths as independent evidence
- 8. Keep species explicit
- 9. Use homology carefully
- 10. Align phenotypes through ontologies
- 11. Read phenotype similarity as a model
- 12. Inspect evidence codes
13–24 · Models and estimation
- 13. Keep disease scope visible
- 14. Separate genes from variants
- 15. Separate association from mechanism
- 16. Trace publications
- 17. Watch source update schedules
- 18. Handle conflicts openly
- 19. Use graph queries reproducibly
- 20. Beware high-degree hubs
- 21. Challenge missing edges
- 22. Use multi-hop paths cautiously
- 23. Design negative controls
- 24. Separate discovery from validation
25–36 · Diagnostics and interpretation
- 25. Protect sensitive inputs
- 26. Cite the platform and data
- 27. Use bounded clinical language
- 28. Connect to Primary Science
- 29. Connect to Secondary Science
- 30. Connect to mathematics
- 31. Connect to computing
- 32. Connect to pathways
- 33. Did you know?
- 34. Write a reader-ready result
- 35. Build a validation ladder
- 36. Finish with a reproducible bundle
Section 1 of 36
1. Start with an integrated graph
Monarch connects structured biomedical entities and associations contributed by many resources so users can explore relationships across datasets.
Working checkpoint — knowledge graph: Define the node and edge types needed for the question. Keep the independent sample, design, annotation and software version beside this decision so the evidence can be reconstructed.
For a classroom model, use invented counts and clearly label the activity as an analogy rather than a sequencing experiment.
Section 2 of 36
2. Understand nodes and edges
Nodes represent entities such as genes, diseases and phenotypes; edges represent typed associations between them.
Working checkpoint — biomedical entity: Read both endpoints and the predicate. Read the numerical result together with sample-level plots, effect size and a sensitivity check capable of weakening the headline.
For future study, connect the idea to biology, mathematics and computing while checking current official pathway information separately.
Section 3 of 36
3. Keep identifiers with names
Human-readable labels can be ambiguous or change. Curated identifiers preserve entity identity across sources.
Working checkpoint — evidence edge: Store identifier namespace, label and release. Preserve the complete tested family and every excluded feature; a polished shortlist is not a substitute for provenance.
For scientific writing, prefer “supports”, “is associated with” and “is predicted to” when stronger causal verbs are not justified.
Section 4 of 36
4. Treat integration as a transformation
Source records are normalised and mapped into a common model. Transformation improves interoperability but may simplify source-specific detail.
Working checkpoint — cross-species phenotype: Retain links back to original records. Separate what the model estimates from the biological story proposed afterward, then name the extra evidence that story would require.
For students, the transferable habit is to ask: What was measured? Compared with what? What alternative explanation remains?
Section 5 of 36
5. Read the source inventory
The official graph overview identifies the biomedical resources and ontologies integrated by the current platform. Coverage differs by entity and species.
Working checkpoint — semantic similarity: Check whether the relevant source is present. Treat identifiers, denominators and factor levels as scientific quantities: a silent mismatch can reverse an otherwise correct calculation.
For parents, a calm checkpoint is to ask how many independent samples support the claim and which assumption matters most.
Section 6 of 36
6. Distinguish asserted from inferred links
Some edges come directly from curated sources; others may be inferred through ontology or graph logic.
Working checkpoint — provenance trail: Record the provenance category for each edge. Keep the independent sample, design, annotation and software version beside this decision so the evidence can be reconstructed.
For a classroom model, use invented counts and clearly label the activity as an analogy rather than a sequencing experiment.
Section 7 of 36
7. Do not count paths as independent evidence
Several paths may trace back to the same publication or imported assertion. Graph density can exaggerate apparent replication.
Working checkpoint — knowledge graph: Deduplicate by underlying evidence source. Read the numerical result together with sample-level plots, effect size and a sensitivity check capable of weakening the headline.
For future study, connect the idea to biology, mathematics and computing while checking current official pathway information separately.
Section 8 of 36
8. Keep species explicit
A gene or phenotype association from mouse, zebrafish or another organism is not a human observation.
Working checkpoint — biomedical entity: Retain the taxon on every node and edge. Preserve the complete tested family and every excluded feature; a polished shortlist is not a substitute for provenance.
For scientific writing, prefer “supports”, “is associated with” and “is predicted to” when stronger causal verbs are not justified.
Section 9 of 36
9. Use homology carefully
Homologous genes can support mechanistic hypotheses across species, but duplication and functional divergence complicate one-to-one transfer.
Working checkpoint — evidence edge: Name the homology relation rather than saying “same gene”. Separate what the model estimates from the biological story proposed afterward, then name the extra evidence that story would require.
For students, the transferable habit is to ask: What was measured? Compared with what? What alternative explanation remains?
Section 10 of 36
10. Align phenotypes through ontologies
Cross-species phenotype comparison relies on anatomy, quality and phenotype ontologies that bridge different vocabularies.
Working checkpoint — cross-species phenotype: Inspect the mapped concepts driving similarity. Treat identifiers, denominators and factor levels as scientific quantities: a silent mismatch can reverse an otherwise correct calculation.
For parents, a calm checkpoint is to ask how many independent samples support the claim and which assumption matters most.
Section 11 of 36
11. Read phenotype similarity as a model
Semantic similarity ranks resemblance under chosen ontologies and annotations. It does not establish disease identity.
Working checkpoint — semantic similarity: Report method, release and top contributing terms. Keep the independent sample, design, annotation and software version beside this decision so the evidence can be reconstructed.
For a classroom model, use invented counts and clearly label the activity as an analogy rather than a sequencing experiment.
Section 12 of 36
12. Inspect evidence codes
An association supported by direct experiment differs from one imported from a computational or high-throughput source.
Working checkpoint — provenance trail: Use evidence metadata in filtering and reporting. Read the numerical result together with sample-level plots, effect size and a sensitivity check capable of weakening the headline.
For future study, connect the idea to biology, mathematics and computing while checking current official pathway information separately.
Section 13 of 36
13. Keep disease scope visible
Broad disease classes and narrow subtypes can connect to overlapping phenotypes or genes. A path can change meaning when scope shifts.
Working checkpoint — knowledge graph: Record exact disease identifiers. Preserve the complete tested family and every excluded feature; a polished shortlist is not a substitute for provenance.
For scientific writing, prefer “supports”, “is associated with” and “is predicted to” when stronger causal verbs are not justified.
Section 14 of 36
14. Separate genes from variants
A gene-level association does not say that every variant in that gene is relevant, and a variant assertion has its own evidence.
Working checkpoint — biomedical entity: Keep gene and variant nodes distinct. Separate what the model estimates from the biological story proposed afterward, then name the extra evidence that story would require.
For students, the transferable habit is to ask: What was measured? Compared with what? What alternative explanation remains?
Section 15 of 36
15. Separate association from mechanism
A path between phenotype, gene and disease proposes a relationship to investigate. It does not reveal molecular mechanism by itself.
Working checkpoint — evidence edge: Write a testable hypothesis, not a causal conclusion. Treat identifiers, denominators and factor levels as scientific quantities: a silent mismatch can reverse an otherwise correct calculation.
For parents, a calm checkpoint is to ask how many independent samples support the claim and which assumption matters most.
Section 16 of 36
16. Trace publications
A useful edge should lead to supporting evidence or an originating resource. Citation counts alone do not grade quality.
Working checkpoint — cross-species phenotype: Open decisive publications and assess design. Keep the independent sample, design, annotation and software version beside this decision so the evidence can be reconstructed.
For a classroom model, use invented counts and clearly label the activity as an analogy rather than a sequencing experiment.
Section 17 of 36
17. Watch source update schedules
Integrated sources release on different calendars. A Monarch graph snapshot can lag or lead an individual portal.
Working checkpoint — semantic similarity: Record graph version and source access dates. Read the numerical result together with sample-level plots, effect size and a sensitivity check capable of weakening the headline.
For future study, connect the idea to biology, mathematics and computing while checking current official pathway information separately.
Section 18 of 36
18. Handle conflicts openly
Different sources can assert different relationships or scopes. Integration should preserve disagreement rather than flatten it.
Working checkpoint — provenance trail: List conflicting edges and their provenance. Preserve the complete tested family and every excluded feature; a polished shortlist is not a substitute for provenance.
For scientific writing, prefer “supports”, “is associated with” and “is predicted to” when stronger causal verbs are not justified.
Section 19 of 36
19. Use graph queries reproducibly
APIs and downloads support repeated analysis, but queries depend on schema, endpoints and releases.
Working checkpoint — knowledge graph: Save the query, parameters and response version. Separate what the model estimates from the biological story proposed afterward, then name the extra evidence that story would require.
For students, the transferable habit is to ask: What was measured? Compared with what? What alternative explanation remains?
Section 20 of 36
20. Beware high-degree hubs
Well-studied genes and broad phenotypes attract many edges, creating popularity bias in naive rankings.
Working checkpoint — biomedical entity: Compare against degree-aware or sensitivity analyses. Treat identifiers, denominators and factor levels as scientific quantities: a silent mismatch can reverse an otherwise correct calculation.
For parents, a calm checkpoint is to ask how many independent samples support the claim and which assumption matters most.
Section 21 of 36
21. Challenge missing edges
No edge can mean no evidence, no ingestion, no mapping or a release lag. Absence in a graph is not biological absence.
Working checkpoint — evidence edge: Check the source database directly. Keep the independent sample, design, annotation and software version beside this decision so the evidence can be reconstructed.
For a classroom model, use invented counts and clearly label the activity as an analogy rather than a sequencing experiment.
Section 22 of 36
22. Use multi-hop paths cautiously
Longer paths can generate creative hypotheses but multiply mapping and evidence uncertainty.
Working checkpoint — cross-species phenotype: Limit path length and inspect every step. Read the numerical result together with sample-level plots, effect size and a sensitivity check capable of weakening the headline.
For future study, connect the idea to biology, mathematics and computing while checking current official pathway information separately.
Section 23 of 36
23. Design negative controls
A graph method should not rank unrelated phenotypes or shuffled evidence as strongly as the intended signal.
Working checkpoint — semantic similarity: Include benchmark positives and negatives. Preserve the complete tested family and every excluded feature; a polished shortlist is not a substitute for provenance.
For scientific writing, prefer “supports”, “is associated with” and “is predicted to” when stronger causal verbs are not justified.
Section 24 of 36
24. Separate discovery from validation
The graph can prioritise candidate genes or models; independent data must test them.
Working checkpoint — provenance trail: Reserve untouched evidence for validation. Separate what the model estimates from the biological story proposed afterward, then name the extra evidence that story would require.
For students, the transferable habit is to ask: What was measured? Compared with what? What alternative explanation remains?
Section 25 of 36
25. Protect sensitive inputs
A detailed patient phenotype profile can be identifying even when names are removed.
Working checkpoint — knowledge graph: Use approved, minimised inputs and access controls. Treat identifiers, denominators and factor levels as scientific quantities: a silent mismatch can reverse an otherwise correct calculation.
For parents, a calm checkpoint is to ask how many independent samples support the claim and which assumption matters most.
Section 26 of 36
26. Cite the platform and data
The Monarch citation guide distinguishes the platform paper, graph and component resources.
Working checkpoint — biomedical entity: Credit both integration and original evidence owners. Keep the independent sample, design, annotation and software version beside this decision so the evidence can be reconstructed.
For a classroom model, use invented counts and clearly label the activity as an analogy rather than a sequencing experiment.
Section 27 of 36
27. Use bounded clinical language
Monarch supports research and hypothesis generation; it does not diagnose or direct treatment.
Working checkpoint — evidence edge: Say “prioritised for review” rather than “identified as causal”. Read the numerical result together with sample-level plots, effect size and a sensitivity check capable of weakening the headline.
For future study, connect the idea to biology, mathematics and computing while checking current official pathway information separately.
Section 28 of 36
28. Connect to Primary Science
A map joining animals by shared features shows how connections help exploration but do not prove ancestry or cause.
Working checkpoint — cross-species phenotype: Use fictional species and transparent rules. Preserve the complete tested family and every excluded feature; a polished shortlist is not a substitute for provenance.
For scientific writing, prefer “supports”, “is associated with” and “is predicted to” when stronger causal verbs are not justified.
Section 29 of 36
29. Connect to Secondary Science
Model organisms can illuminate conserved pathways while species differences test the limits of analogy.
Working checkpoint — semantic similarity: State which evidence is human and which is cross-species. Separate what the model estimates from the biological story proposed afterward, then name the extra evidence that story would require.
For students, the transferable habit is to ask: What was measured? Compared with what? What alternative explanation remains?
Section 30 of 36
30. Connect to mathematics
Graphs, path lengths, neighbourhoods and similarity scores turn integrated knowledge into analysable structure.
Working checkpoint — provenance trail: Compare a raw path count with a provenance-deduplicated count. Treat identifiers, denominators and factor levels as scientific quantities: a silent mismatch can reverse an otherwise correct calculation.
For parents, a calm checkpoint is to ask how many independent samples support the claim and which assumption matters most.
Section 31 of 36
31. Connect to computing
ETL pipelines, ontologies, identifiers, APIs and versioned graphs make many resources interoperable.
Working checkpoint — knowledge graph: Validate entity types and source mappings. Keep the independent sample, design, annotation and software version beside this decision so the evidence can be reconstructed.
For a classroom model, use invented counts and clearly label the activity as an analogy rather than a sequencing experiment.
Section 32 of 36
32. Connect to pathways
Knowledge-graph science joins biology, data engineering, ontology design and statistics and can illuminate STEM education and career pathways.
Working checkpoint — biomedical entity: Check current official programme information separately. Read the numerical result together with sample-level plots, effect size and a sensitivity check capable of weakening the headline.
For future study, connect the idea to biology, mathematics and computing while checking current official pathway information separately.
Section 33 of 36
33. Did you know?
The official Monarch Knowledge Graph combines knowledge from dozens of biomedical resources, yet a single displayed path can still trace back to one original assertion.
Working checkpoint — evidence edge: Follow the path to its evidence before counting support. Preserve the complete tested family and every excluded feature; a polished shortlist is not a substitute for provenance.
For scientific writing, prefer “supports”, “is associated with” and “is predicted to” when stronger causal verbs are not justified.
Section 34 of 36
34. Write a reader-ready result
A strong result names query entities, graph release, path predicates, species, sources, evidence, conflicts and what remains hypothetical.
Working checkpoint — cross-species phenotype: Let every important edge be inspected. Separate what the model estimates from the biological story proposed afterward, then name the extra evidence that story would require.
For students, the transferable habit is to ask: What was measured? Compared with what? What alternative explanation remains?
Section 35 of 36
35. Build a validation ladder
Verify identifiers, inspect edge provenance, deduplicate evidence, challenge cross-species mappings, run controls, then test candidates independently.
Working checkpoint — semantic similarity: Strengthen the weakest edge first. Treat identifiers, denominators and factor levels as scientific quantities: a silent mismatch can reverse an otherwise correct calculation.
For parents, a calm checkpoint is to ask how many independent samples support the claim and which assumption matters most.
Section 36 of 36
36. Finish with a reproducible bundle
Deliver graph version, query, schema, identifiers, edge table, source links, evidence deduplication, sensitivity results and validation plan.
Working checkpoint — provenance trail: Make every graph-based hypothesis reconstructable. Keep the independent sample, design, annotation and software version beside this decision so the evidence can be reconstructed.
For a classroom model, use invented counts and clearly label the activity as an analogy rather than a sequencing experiment.
Contents · Previous section · Continue to the Science Learning Hub