eduKateSG · Why Science?
Connect a genomic change to carefully described phenotypes, sharing status and provenance—then compare cases without erasing consent, uncertainty or difference
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. DECIPHER is a free web platform for sharing and interpreting rare-disease phenotype-linked variant data. Its public paper describes secure deposition, consent-aware sharing, browsing, variant and phenotype comparison, gene and syndrome resources, and patient matching. The current public site identifies its release as v11.41. DECIPHER can help users locate overlapping genomic changes or similar phenotypes and contact responsible professionals through governed routes. A database match remains a hypothesis: shared coordinates, gene or HPO terms do not by themselves establish the same mechanism, variant pathogenicity or diagnosis. 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; ClinVar variant classifications and evidence conflicts; Monarch Initiative phenotype–gene–disease evidence; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 11 October 2026: Official DECIPHER platform, v11.41 observed 11 October 2026; DECIPHER rare-disease data sharing and interpretation paper; Open-access DECIPHER platform paper; EMBL-EBI announcement that DECIPHER joined EMBL-EBI; 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 |
|---|---|---|---|
| Deposit | What phenotype-linked variant record is permitted to be stored? | Variant, phenotype, consent, submitter and provenance | A governed case record |
| Compare | Which records overlap in coordinate, gene or phenotype? | Genome build, consequence, HPO profile and sharing level | A candidate match |
| Contact | Can responsible teams review a plausible match together? | Vetted contact route, consent and contextual data | A collaborative evidence check |
| Validate | Does the combined evidence support one interpretation? | Independent cases, segregation, function and expert review | A bounded conclusion, not a database verdict |
Inside this guide
1–12 · Foundations and design
- 1. Start with linked data
- 2. Understand sharing levels
- 3. Keep identifiers pseudonymous
- 4. Record the genome assembly
- 5. Distinguish sequence and copy-number variants
- 6. Describe phenotype with HPO
- 7. Add age and onset context
- 8. Retain submitter provenance
- 9. Use overlapping variants cautiously
- 10. Use gene matches cautiously
- 11. Use phenotype similarity cautiously
- 12. Check functional consequence
13–24 · Models and estimation
- 13. Inspect population context
- 14. Inspect inheritance and segregation
- 15. Compare independent cases
- 16. Distinguish a match from replication
- 17. Use syndrome resources carefully
- 18. Trace gene information to sources
- 19. Read version and update context
- 20. Audit coordinate liftover
- 21. Audit missingness
- 22. Challenge ascertainment bias
- 23. Challenge visibility bias
- 24. Challenge matching thresholds
25–36 · Diagnostics and interpretation
- 25. Use contact routes responsibly
- 26. Protect consent over time
- 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 linked data
DECIPHER connects genomic variation with structured phenotypes so users can compare the change and the observed features together.
Working checkpoint — phenotype-linked variant: Do not separate the variant from its case context. Keep consent state, case identifier, genome build, variant representation, phenotype profile and release together.
For a classroom model, use fictional records and explicit permission tokens rather than personal or genomic data.
Section 2 of 36
2. Understand sharing levels
Deposited data can have different access and consent states; not every record is public or usable for every purpose.
Working checkpoint — consent status: Read the sharing status before analysis. Read every similarity result beside the exact overlap, consequence, HPO terms and missing data that produced it.
For future study, connect case comparison to biology, computing and ethics while checking official pathways separately.
Section 3 of 36
3. Keep identifiers pseudonymous
A case identifier reduces direct exposure but does not eliminate re-identification risk in rare disease.
Working checkpoint — genomic coordinate: Minimise combinations that are not necessary for the approved purpose. Preserve plausible non-matches and case differences; a polished matching pair is not the whole evidence record.
For scientific writing, call a result a candidate match until independent evidence justifies stronger language.
Section 4 of 36
4. Record the genome assembly
A coordinate is meaningful only on a named reference assembly. The same numbers can point elsewhere after build changes.
Working checkpoint — patient similarity: Store assembly, chromosome, start, end and allele representation. Separate what DECIPHER displays from what an assay measured and what clinical review concludes afterward.
For students, ask: What is truly shared between these cases, and which important differences remain?
Section 5 of 36
5. Distinguish sequence and copy-number variants
Single-nucleotide changes, indels and copy-number variants have different representations, evidence and comparison logic.
Working checkpoint — copy-number variant: Retain the variant class. Treat coordinates, assemblies and permission levels as scientific data: a silent mismatch can create a false match.
For parents, a calm checkpoint is to ask whether a database similarity has been independently confirmed and responsibly reviewed.
Section 6 of 36
6. Describe phenotype with HPO
Structured terms make case comparison computable while original notes preserve nuance.
Working checkpoint — data provenance: Keep present, absent and unknown observations distinct. Keep consent state, case identifier, genome build, variant representation, phenotype profile and release together.
For a classroom model, use fictional records and explicit permission tokens rather than personal or genomic data.
Section 7 of 36
7. Add age and onset context
The same phenotype profile can change across development, and a missing feature may be age dependent.
Working checkpoint — phenotype-linked variant: Record observation date and onset where known. Read every similarity result beside the exact overlap, consequence, HPO terms and missing data that produced it.
For future study, connect case comparison to biology, computing and ethics while checking official pathways separately.
Section 8 of 36
8. Retain submitter provenance
The responsible centre and professional context support clarification and controlled collaboration.
Working checkpoint — consent status: Keep who entered and reviewed each field. Preserve plausible non-matches and case differences; a polished matching pair is not the whole evidence record.
For scientific writing, call a result a candidate match until independent evidence justifies stronger language.
Section 9 of 36
9. Use overlapping variants cautiously
Two copy-number variants may overlap yet differ in size, breakpoints, genes, inheritance and mechanism.
Working checkpoint — genomic coordinate: Calculate reciprocal overlap and inspect interval content. Separate what DECIPHER displays from what an assay measured and what clinical review concludes afterward.
For students, ask: What is truly shared between these cases, and which important differences remain?
Section 10 of 36
10. Use gene matches cautiously
Variants in one gene can have different consequences, inheritance models and diseases.
Working checkpoint — patient similarity: Compare alleles and mechanism, not only the gene symbol. Treat coordinates, assemblies and permission levels as scientific data: a silent mismatch can create a false match.
For parents, a calm checkpoint is to ask whether a database similarity has been independently confirmed and responsibly reviewed.
Section 11 of 36
11. Use phenotype similarity cautiously
Similarity scores organise candidate cases under an ontology and algorithm; incomplete profiles can change rankings.
Working checkpoint — copy-number variant: Inspect the terms driving the score. Keep consent state, case identifier, genome build, variant representation, phenotype profile and release together.
For a classroom model, use fictional records and explicit permission tokens rather than personal or genomic data.
Section 12 of 36
12. Check functional consequence
A consequence annotation predicts transcript impact under a selected transcript set; it is not direct proof of molecular effect.
Working checkpoint — data provenance: Record annotation tool, transcript and release. Read every similarity result beside the exact overlap, consequence, HPO terms and missing data that produced it.
For future study, connect case comparison to biology, computing and ethics while checking official pathways separately.
Section 13 of 36
13. Inspect population context
A variant common in an appropriate population is less compatible with some rare high-penetrance models.
Working checkpoint — phenotype-linked variant: Use ancestry-aware frequency evidence without treating absence as proof. Preserve plausible non-matches and case differences; a polished matching pair is not the whole evidence record.
For scientific writing, call a result a candidate match until independent evidence justifies stronger language.
Section 14 of 36
14. Inspect inheritance and segregation
De novo, recessive, dominant and mosaic explanations require different family observations.
Working checkpoint — consent status: Keep family structure and tested relatives visible. Separate what DECIPHER displays from what an assay measured and what clinical review concludes afterward.
For students, ask: What is truly shared between these cases, and which important differences remain?
Section 15 of 36
15. Compare independent cases
A second truly independent case can strengthen a hypothesis when genotype, phenotype and mechanism align.
Working checkpoint — genomic coordinate: Deduplicate related or republished individuals. Treat coordinates, assemblies and permission levels as scientific data: a silent mismatch can create a false match.
For parents, a calm checkpoint is to ask whether a database similarity has been independently confirmed and responsibly reviewed.
Section 16 of 36
16. Distinguish a match from replication
A database match may be partial, noisy or based on a broad phenotype; replication requires comparable evidence and independent ascertainment.
Working checkpoint — patient similarity: Write the similarities and differences side by side. Keep consent state, case identifier, genome build, variant representation, phenotype profile and release together.
For a classroom model, use fictional records and explicit permission tokens rather than personal or genomic data.
Section 17 of 36
17. Use syndrome resources carefully
Curated syndrome descriptions can orient review, but an individual may have an atypical presentation or a different cause.
Working checkpoint — copy-number variant: Treat the syndrome page as context, not a diagnostic shortcut. Read every similarity result beside the exact overlap, consequence, HPO terms and missing data that produced it.
For future study, connect case comparison to biology, computing and ethics while checking official pathways separately.
Section 18 of 36
18. Trace gene information to sources
Gene pages integrate several types of knowledge that can have different update schedules.
Working checkpoint — data provenance: Cite the decisive underlying evidence. Preserve plausible non-matches and case differences; a polished matching pair is not the whole evidence record.
For scientific writing, call a result a candidate match until independent evidence justifies stronger language.
Section 19 of 36
19. Read version and update context
The platform and its annotations evolve; saved interpretations need the release and access date.
Working checkpoint — phenotype-linked variant: Pin the DECIPHER version and external annotation releases. Separate what DECIPHER displays from what an assay measured and what clinical review concludes afterward.
For students, ask: What is truly shared between these cases, and which important differences remain?
Section 20 of 36
20. Audit coordinate liftover
Moving variants between genome assemblies can fail or produce ambiguous mappings, especially in repetitive regions.
Working checkpoint — consent status: Validate the lifted interval and reference allele. Treat coordinates, assemblies and permission levels as scientific data: a silent mismatch can create a false match.
For parents, a calm checkpoint is to ask whether a database similarity has been independently confirmed and responsibly reviewed.
Section 21 of 36
21. Audit missingness
A blank phenotype, inheritance or validation field is unknown—not negative evidence.
Working checkpoint — genomic coordinate: Preserve explicit missingness reasons. Keep consent state, case identifier, genome build, variant representation, phenotype profile and release together.
For a classroom model, use fictional records and explicit permission tokens rather than personal or genomic data.
Section 22 of 36
22. Challenge ascertainment bias
Submitted rare-disease cases are not a random population sample, so database proportions are not prevalence estimates.
Working checkpoint — patient similarity: Do not infer population risk from case counts. Read every similarity result beside the exact overlap, consequence, HPO terms and missing data that produced it.
For future study, connect case comparison to biology, computing and ethics while checking official pathways separately.
Section 23 of 36
23. Challenge visibility bias
Open records are only a subset of deposited knowledge and can differ from restricted records.
Working checkpoint — copy-number variant: Describe the access slice used. Preserve plausible non-matches and case differences; a polished matching pair is not the whole evidence record.
For scientific writing, call a result a candidate match until independent evidence justifies stronger language.
Section 24 of 36
24. Challenge matching thresholds
Changing overlap or similarity thresholds changes who appears as a match.
Working checkpoint — data provenance: Run a small sensitivity analysis and keep rejected near-matches. Separate what DECIPHER displays from what an assay measured and what clinical review concludes afterward.
For students, ask: What is truly shared between these cases, and which important differences remain?
Section 25 of 36
25. Use contact routes responsibly
A promising match may justify professional contact, but communication must respect submitter expectations, consent and local governance.
Working checkpoint — phenotype-linked variant: Use the platform’s approved contact process. Treat coordinates, assemblies and permission levels as scientific data: a silent mismatch can create a false match.
For parents, a calm checkpoint is to ask whether a database similarity has been independently confirmed and responsibly reviewed.
Section 26 of 36
26. Protect consent over time
Consent, withdrawal and permitted uses can change while copied data may already have travelled.
Working checkpoint — consent status: Check current permission before reuse or contact. Keep consent state, case identifier, genome build, variant representation, phenotype profile and release together.
For a classroom model, use fictional records and explicit permission tokens rather than personal or genomic data.
Section 27 of 36
27. Use bounded clinical language
DECIPHER supports interpretation and research; it does not itself diagnose or recommend care.
Working checkpoint — genomic coordinate: Say “similar record identified for review”. Read every similarity result beside the exact overlap, consequence, HPO terms and missing data that produced it.
For future study, connect case comparison to biology, computing and ethics while checking official pathways separately.
Section 28 of 36
28. Connect to Primary Science
Comparing fictional creatures by features and one coded change shows why both similarity and difference matter.
Working checkpoint — patient similarity: Use invented data and simple consent rules. Preserve plausible non-matches and case differences; a polished matching pair is not the whole evidence record.
For scientific writing, call a result a candidate match until independent evidence justifies stronger language.
Section 29 of 36
29. Connect to Secondary Science
Chromosomes, mutations, inheritance and phenotype become a connected evidence chain.
Working checkpoint — copy-number variant: Separate observed trait, genomic finding and proposed cause. Separate what DECIPHER displays from what an assay measured and what clinical review concludes afterward.
For students, ask: What is truly shared between these cases, and which important differences remain?
Section 30 of 36
30. Connect to mathematics
Intervals, overlap, similarity scores and denominators make case matching quantitative.
Working checkpoint — data provenance: Compare one-way overlap with reciprocal overlap. Treat coordinates, assemblies and permission levels as scientific data: a silent mismatch can create a false match.
For parents, a calm checkpoint is to ask whether a database similarity has been independently confirmed and responsibly reviewed.
Section 31 of 36
31. Connect to computing
Secure databases, ontologies, APIs, genome coordinates and access controls make international collaboration possible.
Working checkpoint — phenotype-linked variant: Validate build, identifier and permission at every transfer. Keep consent state, case identifier, genome build, variant representation, phenotype profile and release together.
For a classroom model, use fictional records and explicit permission tokens rather than personal or genomic data.
Section 32 of 36
32. Connect to pathways
DECIPHER work links clinical genetics, bioinformatics, data stewardship, ethics and software and can illuminate STEM pathways.
Working checkpoint — consent status: Check official training requirements separately. Read every similarity result beside the exact overlap, consequence, HPO terms and missing data that produced it.
For future study, connect case comparison to biology, computing and ethics while checking official pathways separately.
Section 33 of 36
33. Did you know?
A small variant can overlap the same gene as another case yet affect a different transcript or mechanism; “same gene” is only the beginning of comparison.
Working checkpoint — genomic coordinate: Open the exact variant records. Preserve plausible non-matches and case differences; a polished matching pair is not the whole evidence record.
For scientific writing, call a result a candidate match until independent evidence justifies stronger language.
Section 34 of 36
34. Write a reader-ready result
A strong summary names case access level, variant, build, phenotype profile, similarity method, matches, differences, evidence and uncertainty.
Working checkpoint — patient similarity: Make consent and provenance visible. Separate what DECIPHER displays from what an assay measured and what clinical review concludes afterward.
For students, ask: What is truly shared between these cases, and which important differences remain?
Section 35 of 36
35. Build a validation ladder
Verify permissions and identifiers, confirm coordinates and consequences, compare phenotypes and inheritance, contact responsibly, then seek independent validation.
Working checkpoint — copy-number variant: Do not skip governance to reach biology faster. Treat coordinates, assemblies and permission levels as scientific data: a silent mismatch can create a false match.
For parents, a calm checkpoint is to ask whether a database similarity has been independently confirmed and responsibly reviewed.
Section 36 of 36
36. Finish with a reproducible bundle
Deliver permitted case fields, variant representation, phenotype terms, release, matching parameters, results, contact record, evidence review and governance notes.
Working checkpoint — data provenance: Make every comparison reconstructable. Keep consent state, case identifier, genome build, variant representation, phenotype profile and release together.
For a classroom model, use fictional records and explicit permission tokens rather than personal or genomic data.
Contents · Previous section · Continue to the Science Learning Hub