eduKateSG · Why Science?
Follow every classification back to its submitter, condition, evidence, date and review status before drawing a conclusion
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. ClinVar is NCBI’s freely accessible public archive of submitted reports about human genomic variation classified for inherited diseases, cancer and drug responses, together with supporting evidence. It preserves who submitted an assertion, the condition and classification type, review status, dates and the history of the record. ClinVar aggregates submissions for communication; it does not independently reproduce every classification, and its star-based review status describes review and transparency rather than the biological effect of an allele. A responsible reader opens the underlying submissions and evidence before using a headline classification. 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: ANNOVAR and gene-, region- and filter-based annotation; Ensembl VEP and transcript-aware variant consequences; gnomAD and population allele-frequency context; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 11 October 2026: Official NCBI introduction to ClinVar; Official NCBI ClinVar review-status documentation; Official NCBI representation of classifications in ClinVar; ClinVar primary archive paper in PubMed; 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 |
|---|---|---|---|
| Identify | Which exact variant, condition and classification type are shown? | VCV/RCV pages, allele description and condition record | A correctly scoped record |
| Trace | Who submitted each classification and when? | SCV accessions, submitter, evaluation date and assertion criteria | A provenance-aware evidence list |
| Compare | Do submitters agree and how was each assertion reviewed? | Review status, evidence summaries and conflicts | A transparent account of consensus or disagreement |
| Interpret | What conclusion is supported outside the archive? | Population, phenotype, segregation, functional and professional review | A bounded interpretation, not an automatic diagnosis |
Inside this guide
1–12 · Foundations and design
- 1. Start with an archive of submissions
- 2. Separate the variant from the assertion
- 3. Learn the accession layers
- 4. Keep the condition attached
- 5. Distinguish classification types
- 6. Read review status in words
- 7. Do not treat stars as an effect score
- 8. Name the submitter
- 9. Inspect assertion criteria
- 10. Read the evidence summary
- 11. Treat conflicts as information
- 12. Use evaluation and update dates
13–24 · Models and estimation
- 13. Verify variant identity
- 14. Keep the assembly and transcript visible
- 15. Understand versions
- 16. Read aggregation rules
- 17. Recognise expert panels and guidelines
- 18. Keep germline and somatic logic apart
- 19. Use downloads and APIs reproducibly
- 20. Interpret missing evidence cautiously
- 21. Understand uncertain significance
- 22. Reconcile disagreement methodically
- 23. Map conditions carefully
- 24. Add population evidence separately
25–36 · Diagnostics and interpretation
- 25. Add functional and segregation evidence
- 26. Plan for reanalysis
- 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 archive of submissions
ClinVar stores reports supplied by laboratories, expert panels, researchers and other recognised sources. Its public value is that those assertions can be inspected together.
Working checkpoint — submitted classification: Use “ClinVar contains a submitted classification” rather than implying that the database performed the clinical evaluation itself. 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. Separate the variant from the assertion
A genomic allele is an object; a classification is a claim about that object under a stated condition and classification type. Mixing the two turns evidence into a permanent label.
Working checkpoint — SCV accession: Record allele, condition, classification category and assertion owner in separate fields. 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. Learn the accession layers
Submitted classifications receive SCV accessions. ClinVar also aggregates information into variant-condition RCV records and variant-level VCV records, each with versions and a different scope.
Working checkpoint — VCV record: Cite the accession that matches the claim instead of copying only a Variation ID. 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. Keep the condition attached
The same allele may be assessed for different disorders, cancer contexts or drug responses. A classification without its condition has lost essential meaning.
Working checkpoint — RCV record: Retain the condition identifier and name beside every classification. 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. Distinguish classification types
ClinVar documentation separates germline clinical significance, somatic clinical impact and somatic oncogenicity. Their terms and aggregation rules are not interchangeable.
Working checkpoint — review status: State the classification type before stating the category. 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. Read review status in words
The official review-status page maps stars to descriptions such as practice guideline, expert panel, multiple submitters with criteria, single submitter or no assertion criteria.
Working checkpoint — evidence conflict: Store the written review status as well as the star count. 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 treat stars as an effect score
Stars communicate the level and form of review supporting submissions. They are not a probability that a variant is pathogenic and do not measure effect size.
Working checkpoint — submitted classification: Explain what the stars mean before using them to rank records. 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. Name the submitter
A classification is more interpretable when the reader knows the organisation, expertise, method and public contact behind it.
Working checkpoint — SCV accession: Open the SCV and record the submitter rather than relying only on the aggregate page. 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. Inspect assertion criteria
ClinVar defines assertion criteria as the documented categories and rules used by a submitter. Criteria describe the method, not necessarily every item of evidence for one allele.
Working checkpoint — VCV record: Link the dated criteria document and distinguish it from case-specific evidence. 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. Read the evidence summary
Supporting text may mention case observations, segregation, population frequency, computational predictions or functional work. Detail and completeness vary by submission.
Working checkpoint — RCV record: Extract the evidence types and note which supporting data are absent. 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. Treat conflicts as information
Different submitters can disagree because of evidence dates, condition definitions, transcript choices, thresholds or methods. A conflict is a signal to investigate, not a database failure to hide.
Working checkpoint — review status: List the individual classifications, dates and methods before summarising disagreement. 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. Use evaluation and update dates
The date last evaluated belongs to the submitter’s classification; the record update date describes archive activity. New evidence can make an older assertion less informative.
Working checkpoint — evidence conflict: Keep both dates and check for newer submissions or expert review. 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. Verify variant identity
HGVS expressions, genomic coordinates and identifiers can refer to the same allele at different sequence levels. An incorrect transcript or assembly can retrieve the wrong record.
Working checkpoint — submitted classification: Match reference sequence accessions, versions and alleles, not a gene name alone. 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. Keep the assembly and transcript visible
Genomic coordinates change across assemblies, while coding descriptions depend on transcript accessions. ClinVar may display several equivalent expressions.
Working checkpoint — SCV accession: Anchor the search with a validated genomic allele and retain the transcript used in reporting. 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. Understand versions
VCV, RCV and SCV accessions can carry version numbers as records change. A citation without a version and access date may not reproduce the observed page later.
Working checkpoint — VCV record: Archive the accession version, accessed date and downloaded release. 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. Read aggregation rules
Aggregate classifications are calculated from contributing submissions under documented rules. A headline is therefore a derived summary, not a replacement for every SCV.
Working checkpoint — RCV record: Open the contributing submissions whenever the decision is consequential. 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. Recognise expert panels and guidelines
Expert-panel and practice-guideline submissions receive higher review status because they follow recognised review routes. Their scope still depends on the condition and classification type.
Working checkpoint — review status: Verify that the panel’s disease scope matches the question. 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. Keep germline and somatic logic apart
A hereditary-disease assertion, a tumour oncogenicity claim and a treatment-response tier answer different questions. Reusing one as another can mislead.
Working checkpoint — evidence conflict: Partition the workflow by classification type before aggregation. 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 downloads and APIs reproducibly
ClinVar offers web pages and machine-readable releases. Large analyses should pin the release, file format and parsing logic rather than scrape changing displays.
Working checkpoint — submitted classification: Save the release date, checksums, parser version and filtering rules. 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. Interpret missing evidence cautiously
An empty evidence field can mean evidence was not submitted publicly, not that no evidence existed. Conversely, a classification label does not guarantee detailed public support.
Working checkpoint — SCV accession: Represent missingness explicitly and avoid filling gaps with assumption. 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. Understand uncertain significance
A variant of uncertain significance marks insufficient or conflicting evidence for a more decisive classification. It is not a halfway diagnosis.
Working checkpoint — VCV record: Do not convert uncertainty into risk without qualified review and new evidence. 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. Reconcile disagreement methodically
Compare allele identity, condition, inheritance model, criteria, evidence, dates and submitter expertise. Many apparent conflicts narrow after scope is aligned.
Working checkpoint — RCV record: Build a side-by-side evidence table before choosing any summary. 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. Map conditions carefully
Condition names can be broad, narrow or historically changed. Ontology identifiers help, but clinical phenotypes and inheritance context still matter.
Working checkpoint — review status: Retain identifiers and human-readable names and note mapping uncertainty. 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. Add population evidence separately
Population frequency from a versioned resource can support or challenge plausibility under a disease model. It is external evidence, not a ClinVar star.
Working checkpoint — evidence conflict: Match the exact allele and appropriate population before comparison. 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. Add functional and segregation evidence
Laboratory assays, family segregation and phenotype fit address mechanisms that an archive record alone cannot settle. Their strength depends on design and relevance.
Working checkpoint — submitted classification: Keep each evidence stream separate until a recognised framework combines them. 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. Plan for reanalysis
Classifications can change as new cases, functional studies or expert specifications appear. A one-time lookup can age silently.
Working checkpoint — SCV accession: Record a review date and recheck material findings under current professional guidance. 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
ClinVar facilitates evidence sharing; it does not replace patient-specific assessment by qualified professionals. Educational summaries must not direct care.
Working checkpoint — VCV record: Write “submitted as” and “review status” with the condition and date. 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 class can compare several labelled cards and ask who made each claim, what rule was used and whether the evidence agrees.
Working checkpoint — RCV record: Use fictional objects and non-medical examples. 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
Variation, inheritance and evidence evaluation meet in one archive. Students can see why the same DNA change may support different claims in different contexts.
Working checkpoint — review status: Separate observation, database record and interpretation in every explanation. 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
Aggregation rules, categories and disagreement counts show how summaries compress multiple inputs. A star is ordinal metadata, not a probability.
Working checkpoint — evidence conflict: Ask what information disappears when several SCVs become one headline. 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
Stable accessions, schemas, versions and APIs make a living archive computable. Reproducibility requires exact identifiers and dated releases.
Working checkpoint — submitted classification: Test parsers against records with conflicts, multiple conditions and version changes. 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
Clinical genomics joins biology, evidence synthesis, data stewardship and software engineering and can illuminate STEM education and career pathways.
Working checkpoint — SCV accession: Use current official course and professional requirements for real decisions. 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?
ClinVar preserves both agreement and disagreement. That means a conflict can become a useful map of where evidence, scope or methods diverge.
Working checkpoint — VCV record: Open the underlying SCVs before treating the aggregate as settled. 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 responsible summary names the allele, condition, classification type, aggregate label, review status, contributing submissions, dates and unresolved conflict.
Working checkpoint — RCV record: End with the narrowest claim supported by the evidence. 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
Confirm allele identity, inspect SCVs, align conditions and methods, compare dates, add population and functional evidence, then seek qualified interpretation.
Working checkpoint — review status: Strengthen the weakest link before strengthening the language. 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 accessions and versions, release files, exact query, submitter evidence table, external evidence, conflict notes, parser code and review date.
Working checkpoint — evidence conflict: Make the summary traceable to each contributing classification. 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