eduKateSG · Why Science?
Turn a gene–disease claim into a structured model: disease, inheritance, molecular mechanism, allelic requirement, evidence and curation date all belong together
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. Gene2Phenotype, or G2P, is an expert-curated resource that disseminates structured models of monogenic gene–disease relationships. Its public material and papers describe curation from scientific literature, with fields for disease, gene, molecular mechanism, allelic requirement, variant information, phenotypes and confidence. The resource began with developmental disorders and has expanded into additional domain panels. A G2P model is a traceable assertion about a defined relationship; it does not classify every allele in the gene, guarantee completeness or replace case-level diagnostic review. 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: ClinGen gene–disease validity evidence framework; ORDO and structured rare-disease concepts; Human Phenotype Ontology and structured clinical phenotypes; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 11 October 2026: Official Gene2Phenotype portal; EMBL-EBI Gene2Phenotype update, 27 February 2025; Gene2Phenotype structured curation process; Gene2Phenotype structured monogenic disease database; 2026 Singapore–Cambridge O-Level Biology syllabus; MOE G2/G3 Lower Secondary Science teaching and learning syllabus; 2026 MOE G2 Computing syllabus. Resource interfaces and recommendations can change, so preserve the release, access date and documentation used.
| Stage | Question | Evidence | Responsible output |
|---|---|---|---|
| Define | Which gene, disease and inheritance model are asserted? | Gene identifier, disease concept and allelic requirement | A scoped model |
| Curate | What literature supports the relationship and mechanism? | Cases, segregation, functional work and publications | A traceable evidence set |
| Classify | How confident is the relationship under G2P criteria? | Confidence category, reviewer and date | A dated model strength |
| Apply | Does a particular case and variant fit the model? | Variant consequence, phenotype and independent interpretation | A candidate for review, not a diagnosis |
Inside this guide
1–12 · Foundations and design
- 1. Start with a structured relationship
- 2. Understand expert curation
- 3. Keep the gene identifier stable
- 4. Define the disease concept
- 5. Read allelic requirement
- 6. Read mutation consequence
- 7. Read molecular mechanism
- 8. Read confidence as a category
- 9. Distinguish definitive from emerging evidence
- 10. Trace cases to publications
- 11. Inspect segregation evidence
- 12. Inspect de novo evidence
13–24 · Models and estimation
- 13. Inspect case-control evidence
- 14. Inspect functional evidence
- 15. Use phenotype annotations carefully
- 16. Separate relationship from variant classification
- 17. Separate confidence from penetrance
- 18. Separate mechanism from treatment
- 19. Expect panel-specific coverage
- 20. Use downloads with versions
- 21. Map external identifiers transparently
- 22. Challenge contradictory evidence
- 23. Challenge publication bias
- 24. Challenge circular evaluation
25–36 · Diagnostics and interpretation
- 25. Protect sensitive case detail
- 26. Credit the evidence chain
- 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 a structured relationship
G2P represents a gene linked to a defined disease through an inheritance and molecular model rather than publishing an unqualified gene list.
Working checkpoint — gene–disease model: Write each entity and relation before collecting evidence. Keep gene, disease, allelic requirement, mechanism, confidence and curation date together so the model can be reconstructed.
For a classroom model, use fictional pedigrees and mechanism cards rather than real medical histories.
Section 2 of 36
2. Understand expert curation
Domain experts evaluate published clinical, bioinformatic and functional evidence and encode it into reusable fields.
Working checkpoint — confidence category: Keep curator and curation date visible. Read the confidence category beside the independent cases, segregation, functional work and contradictory evidence that support it.
For future study, connect literature curation to genetics, statistics and computing while checking official requirements separately.
Section 3 of 36
3. Keep the gene identifier stable
Symbols can change or collide across species, so a current identifier anchors the gene entity.
Working checkpoint — allelic requirement: Store namespace, identifier, symbol and release. Preserve all reviewed families and publications; repeated reports of one pedigree are not new replication.
For scientific writing, report the exact disease model rather than the vague claim that a gene “causes disease”.
Section 4 of 36
4. Define the disease concept
A broad syndrome and a molecularly specific disorder may justify different gene–disease assertions.
Working checkpoint — disease mechanism: Record the disease identifier and scope statement. Separate the curated relationship model from the patient variant and from any proposed treatment consequence.
For students, ask: Which part is the observed case, which part is the inheritance pattern and which part is the proposed mechanism?
Section 5 of 36
5. Read allelic requirement
The model can specify whether one or two altered copies, mosaicism or another configuration is expected.
Working checkpoint — variant consequence: Compare the case genotype with the curated allelic requirement. Treat identifiers, inheritance and molecular mechanism as scientific fields: a silent scope change can transform the claim.
For parents, a calm checkpoint is to ask whether the result concerns a gene–disease relationship or one specific variant.
Section 6 of 36
6. Read mutation consequence
Loss of function, altered function, gain of function and other mechanisms predict different relevant variant classes.
Working checkpoint — literature curation: Do not assume every rare consequence fits the model. Keep gene, disease, allelic requirement, mechanism, confidence and curation date together so the model can be reconstructed.
For a classroom model, use fictional pedigrees and mechanism cards rather than real medical histories.
Section 7 of 36
7. Read molecular mechanism
Mechanism explains how genetic variation is proposed to produce disease and can constrain interpretation.
Working checkpoint — gene–disease model: Separate demonstrated mechanism from plausible narrative. Read the confidence category beside the independent cases, segregation, functional work and contradictory evidence that support it.
For future study, connect literature curation to genetics, statistics and computing while checking official requirements separately.
Section 8 of 36
8. Read confidence as a category
G2P confidence labels summarise the strength of the curated relationship under defined criteria; they are not probabilities.
Working checkpoint — confidence category: Report the category name and criteria version. Preserve all reviewed families and publications; repeated reports of one pedigree are not new replication.
For scientific writing, report the exact disease model rather than the vague claim that a gene “causes disease”.
Section 9 of 36
9. Distinguish definitive from emerging evidence
Well-replicated relationships and newly proposed ones belong in different decision contexts.
Working checkpoint — allelic requirement: Keep research candidates separate from established diagnostic models. Separate the curated relationship model from the patient variant and from any proposed treatment consequence.
For students, ask: Which part is the observed case, which part is the inheritance pattern and which part is the proposed mechanism?
Section 10 of 36
10. Trace cases to publications
One paper can describe several individuals, and several papers can reuse the same family.
Working checkpoint — disease mechanism: Deduplicate people and pedigrees. Treat identifiers, inheritance and molecular mechanism as scientific fields: a silent scope change can transform the claim.
For parents, a calm checkpoint is to ask whether the result concerns a gene–disease relationship or one specific variant.
Section 11 of 36
11. Inspect segregation evidence
Co-segregation can support a model but depends on family size, phenotyping, penetrance and alternative explanations.
Working checkpoint — variant consequence: Record informative and non-informative relatives. Keep gene, disease, allelic requirement, mechanism, confidence and curation date together so the model can be reconstructed.
For a classroom model, use fictional pedigrees and mechanism cards rather than real medical histories.
Section 12 of 36
12. Inspect de novo evidence
A confirmed de novo event can be informative, but parentage, mosaicism and background rate matter.
Working checkpoint — literature curation: Preserve validation method and parental testing details. Read the confidence category beside the independent cases, segregation, functional work and contradictory evidence that support it.
For future study, connect literature curation to genetics, statistics and computing while checking official requirements separately.
Section 13 of 36
13. Inspect case-control evidence
Case enrichment requires compatible populations, variant definitions and statistical design.
Working checkpoint — gene–disease model: Report denominators and ancestry context. Preserve all reviewed families and publications; repeated reports of one pedigree are not new replication.
For scientific writing, report the exact disease model rather than the vague claim that a gene “causes disease”.
Section 14 of 36
14. Inspect functional evidence
Cell, animal and biochemical studies can support mechanism, but model relevance and assay validity determine weight.
Working checkpoint — confidence category: Name what the experiment actually measured. Separate the curated relationship model from the patient variant and from any proposed treatment consequence.
For students, ask: Which part is the observed case, which part is the inheritance pattern and which part is the proposed mechanism?
Section 15 of 36
15. Use phenotype annotations carefully
Structured phenotypes support comparison, yet incomplete or age-dependent observations can weaken apparent matches.
Working checkpoint — allelic requirement: Keep absent, unknown and unassessed states distinct. Treat identifiers, inheritance and molecular mechanism as scientific fields: a silent scope change can transform the claim.
For parents, a calm checkpoint is to ask whether the result concerns a gene–disease relationship or one specific variant.
Section 16 of 36
16. Separate relationship from variant classification
A strong gene–disease model does not make every variant pathogenic.
Working checkpoint — disease mechanism: Apply an appropriate variant framework independently. Keep gene, disease, allelic requirement, mechanism, confidence and curation date together so the model can be reconstructed.
For a classroom model, use fictional pedigrees and mechanism cards rather than real medical histories.
Section 17 of 36
17. Separate confidence from penetrance
Confidence that a relationship exists does not tell how often a variant carrier develops disease.
Working checkpoint — variant consequence: Cite penetrance evidence separately. Read the confidence category beside the independent cases, segregation, functional work and contradictory evidence that support it.
For future study, connect literature curation to genetics, statistics and computing while checking official requirements separately.
Section 18 of 36
18. Separate mechanism from treatment
Understanding mechanism can guide research, but it does not establish clinical actionability or benefit.
Working checkpoint — literature curation: Do not infer therapy from the model alone. Preserve all reviewed families and publications; repeated reports of one pedigree are not new replication.
For scientific writing, report the exact disease model rather than the vague claim that a gene “causes disease”.
Section 19 of 36
19. Expect panel-specific coverage
G2P domain panels cover selected disease areas and evidence priorities; absence can reflect scope rather than refutation.
Working checkpoint — gene–disease model: Check the relevant panel and its remit. Separate the curated relationship model from the patient variant and from any proposed treatment consequence.
For students, ask: Which part is the observed case, which part is the inheritance pattern and which part is the proposed mechanism?
Section 20 of 36
20. Use downloads with versions
Structured data are valuable for pipelines only when retrieval date, schema and release are pinned.
Working checkpoint — confidence category: Archive the downloaded file and checksum. Treat identifiers, inheritance and molecular mechanism as scientific fields: a silent scope change can transform the claim.
For parents, a calm checkpoint is to ask whether the result concerns a gene–disease relationship or one specific variant.
Section 21 of 36
21. Map external identifiers transparently
Links to disease ontologies and other resources can change as nomenclature evolves.
Working checkpoint — allelic requirement: Retain the source mapping and date. Keep gene, disease, allelic requirement, mechanism, confidence and curation date together so the model can be reconstructed.
For a classroom model, use fictional pedigrees and mechanism cards rather than real medical histories.
Section 22 of 36
22. Challenge contradictory evidence
Population observations, unaffected carriers, alternative genes or failed functional replication can weaken a model.
Working checkpoint — disease mechanism: Record contradictions, not only supporting studies. Read the confidence category beside the independent cases, segregation, functional work and contradictory evidence that support it.
For future study, connect literature curation to genetics, statistics and computing while checking official requirements separately.
Section 23 of 36
23. Challenge publication bias
Positive novel associations are easier to publish than negative follow-up, which can inflate apparent confidence.
Working checkpoint — variant consequence: Search for refutations and larger reanalyses. Preserve all reviewed families and publications; repeated reports of one pedigree are not new replication.
For scientific writing, report the exact disease model rather than the vague claim that a gene “causes disease”.
Section 24 of 36
24. Challenge circular evaluation
Testing a prioritisation method on cases whose publications informed the curated model can overstate performance.
Working checkpoint — literature curation: Separate curation and benchmark evidence. Separate the curated relationship model from the patient variant and from any proposed treatment consequence.
For students, ask: Which part is the observed case, which part is the inheritance pattern and which part is the proposed mechanism?
Section 25 of 36
25. Protect sensitive case detail
Rare combinations of genotype and phenotype can identify individuals even without names.
Working checkpoint — gene–disease model: Use only approved, minimised data. Treat identifiers, inheritance and molecular mechanism as scientific fields: a silent scope change can transform the claim.
For parents, a calm checkpoint is to ask whether the result concerns a gene–disease relationship or one specific variant.
Section 26 of 36
26. Credit the evidence chain
G2P structures evidence from original research; both the database and decisive papers deserve attribution.
Working checkpoint — confidence category: Cite the model record and primary studies. Keep gene, disease, allelic requirement, mechanism, confidence and curation date together so the model can be reconstructed.
For a classroom model, use fictional pedigrees and mechanism cards rather than real medical histories.
Section 27 of 36
27. Use bounded clinical language
G2P supports interpretation and research; it does not diagnose a person or direct care.
Working checkpoint — allelic requirement: Write “consistent with the curated model” rather than “confirmed”. Read the confidence category beside the independent cases, segregation, functional work and contradictory evidence that support it.
For future study, connect literature curation to genetics, statistics and computing while checking official requirements separately.
Section 28 of 36
28. Connect to Primary Science
A rule card linking fictional traits to inheritance patterns shows why every classification needs evidence and exceptions.
Working checkpoint — disease mechanism: Use invented examples and avoid personal health data. Preserve all reviewed families and publications; repeated reports of one pedigree are not new replication.
For scientific writing, report the exact disease model rather than the vague claim that a gene “causes disease”.
Section 29 of 36
29. Connect to Secondary Science
Genes, alleles, inheritance and protein function become a connected explanation rather than isolated definitions.
Working checkpoint — variant consequence: Mark which step is observation and which is mechanism. Separate the curated relationship model from the patient variant and from any proposed treatment consequence.
For students, ask: Which part is the observed case, which part is the inheritance pattern and which part is the proposed mechanism?
Section 30 of 36
30. Connect to mathematics
Counts of independent families, segregation and case-control evidence show why denominators matter.
Working checkpoint — literature curation: Recalculate evidence after removing a duplicate family. Treat identifiers, inheritance and molecular mechanism as scientific fields: a silent scope change can transform the claim.
For parents, a calm checkpoint is to ask whether the result concerns a gene–disease relationship or one specific variant.
Section 31 of 36
31. Connect to computing
Schemas, identifiers, APIs and versioned exports make expert reasoning reusable by software.
Working checkpoint — gene–disease model: Validate required fields and reject silent nulls. Keep gene, disease, allelic requirement, mechanism, confidence and curation date together so the model can be reconstructed.
For a classroom model, use fictional pedigrees and mechanism cards rather than real medical histories.
Section 32 of 36
32. Connect to pathways
Gene–disease curation joins biology, genetics, medicine, literature review and data engineering and can illuminate STEM pathways.
Working checkpoint — confidence category: Check current programme requirements separately. Read the confidence category beside the independent cases, segregation, functional work and contradictory evidence that support it.
For future study, connect literature curation to genetics, statistics and computing while checking official requirements separately.
Section 33 of 36
33. Did you know?
Two disorders linked to one gene can involve different allelic requirements or molecular mechanisms, so “the gene causes disease” is often too vague.
Working checkpoint — allelic requirement: Name the exact disease model. Preserve all reviewed families and publications; repeated reports of one pedigree are not new replication.
For scientific writing, report the exact disease model rather than the vague claim that a gene “causes disease”.
Section 34 of 36
34. Write a reader-ready result
A strong summary names gene, disease, inheritance, allelic requirement, mechanism, confidence, evidence, contradictions and curation date.
Working checkpoint — disease mechanism: Let every field lead back to evidence. Separate the curated relationship model from the patient variant and from any proposed treatment consequence.
For students, ask: Which part is the observed case, which part is the inheritance pattern and which part is the proposed mechanism?
Section 35 of 36
35. Build a validation ladder
Verify entities and scope, inspect independent cases, test mechanism fit, review contradictions, classify the variant separately and seek expert review.
Working checkpoint — variant consequence: Strengthen the least certain field first. Treat identifiers, inheritance and molecular mechanism as scientific fields: a silent scope change can transform the claim.
For parents, a calm checkpoint is to ask whether the result concerns a gene–disease relationship or one specific variant.
Section 36 of 36
36. Finish with a reproducible bundle
Deliver model identifiers, panel, evidence table, publications, confidence, mappings, release file, variant comparison and review notes.
Working checkpoint — literature curation: Make the structured assertion reconstructable. Keep gene, disease, allelic requirement, mechanism, confidence and curation date together so the model can be reconstructed.
For a classroom model, use fictional pedigrees and mechanism cards rather than real medical histories.
Contents · Previous section · Continue to the Science Learning Hub