eduKateSG · Why Science?
Turn a complex case into a portable evidence package—keeping phenotype, time, diagnosis, samples and genomic findings connected without confusing structure with truth
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. GA4GH Phenopackets is a standard for exchanging phenotypic and clinical information in a form that people can read and software can process. The official GA4GH product page lists version 2.0 as the last approved version and describes interoperability with GA4GH Variation Representation and HL7 FHIR. A Phenopacket can connect an individual, structured phenotypic features, diagnoses, measurements, biosamples, medical actions, interpretations and referenced files. The format improves portability and reproducibility; it does not guarantee that a mapping is correct, that a record is complete or that sharing is permitted. 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; GA4GH VRS and computable variant representation; PhEval and reproducible phenotype-tool benchmarking; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 12 October 2026: Official GA4GH Phenopackets product page, version 2.0; Official Phenopacket Schema documentation; Phenopacket-tools reference implementation; Phenopacket Store corpus paper; 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 |
|---|---|---|---|
| Represent | What observations and context belong to this case? | Individual, features, diseases, measurements, samples and interpretations | A structured case package |
| Identify | How are meanings made computable? | Ontology identifiers, resource metadata and stable references | Machine-readable concepts with visible labels |
| Validate | Does the record conform and retain its meaning? | Schema checks, terminology review and provenance | A technically valid, scientifically reviewed packet |
| Govern | What may be shared, with whom and for what purpose? | Consent, minimisation, access control and audit | A permitted use—not automatic openness |
Inside this guide
1–12 · Foundations and design
- 1. Start with the case, not the file
- 2. Understand the Phenopacket standard
- 3. Know the approved version
- 4. Represent the subject
- 5. Capture phenotypic features
- 6. Separate present, excluded and unknown
- 7. Add time deliberately
- 8. Represent diagnoses without overclaiming
- 9. Record measurements
- 10. Describe medical actions carefully
- 11. Connect biosamples
- 12. Represent interpretations
13–24 · Models and estimation
- 13. Reference external files
- 14. Include resource metadata
- 15. Use HPO for phenotype meaning
- 16. Use disease ontologies responsibly
- 17. Connect variants through GA4GH standards
- 18. Do not confuse a link with the data
- 19. Choose JSON or Protocol Buffers knowingly
- 20. Validate against the schema
- 21. Validate terminology
- 22. Preserve provenance
- 23. Connect carefully with FHIR
- 24. Minimise sensitive data
25–36 · Diagnostics and interpretation
- 25. Respect consent and purpose
- 26. Plan for reanalysis
- 27. Use packages in benchmarking
- 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 packet report
- 35. Build a validation ladder
- 36. Finish with a reproducible bundle
Section 1 of 36
1. Start with the case, not the file
A Phenopacket is useful when it preserves the connected evidence of a case rather than becoming a bag of disconnected codes.
Working checkpoint — Phenopacket: Write the intended question and permitted use before choosing fields. Keep the source record, Phenopacket, schema, terminology releases and transformation log together.
For a classroom model, use fictional creatures and invented measurements rather than health records.
Section 2 of 36
2. Understand the Phenopacket standard
GA4GH describes Phenopackets as a human- and machine-readable structure for clinical and phenotypic data.
Working checkpoint — phenotypic feature: Treat the standard as a representation layer, not a diagnosis engine. Read every structured field beside its original observation, time context and review status.
For future study, connect data standards to biology, medicine, computing and ethics while checking official pathways separately.
Section 3 of 36
3. Know the approved version
The official product page identifies version 2.0 as the last approved version. Schema versions matter because field definitions and validation rules can change.
Working checkpoint — time element: Record the exact schema version in every exchange. Preserve unknown and excluded states; a clean packet is not proof of completeness.
For scientific writing, name the Phenopacket schema and ontology versions.
Section 4 of 36
4. Represent the subject
The individual record can carry an identifier, sex, vital status and time-at-last-encounter information appropriate to the use case.
Working checkpoint — biosample: Use a governed identifier and omit fields that are not needed. Separate technical schema validity from scientific correctness and sharing permission.
For students, ask which fields are direct observations, mappings or later interpretations.
Section 5 of 36
5. Capture phenotypic features
PhenotypicFeature objects link observations to ontology terms and can include evidence, exclusions, severity and temporal context.
Working checkpoint — ontology identifier: Preserve the source wording beside every mapping. Treat subject, sample, ontology and variant identifiers as scientific data: a silent mismatch can join the wrong evidence.
For parents, ask whether a portable record is complete, reviewed and authorised for its stated use.
Section 6 of 36
6. Separate present, excluded and unknown
A specifically excluded finding is evidence; an unrecorded or unassessed finding is not the same thing.
Working checkpoint — computable case representation: Use negation only when the observation was actually assessed. Keep the source record, Phenopacket, schema, terminology releases and transformation log together.
For a classroom model, use fictional creatures and invented measurements rather than health records.
Section 7 of 36
7. Add time deliberately
Age, age range, gestational age, interval and timestamp options can place observations in a clinical timeline.
Working checkpoint — Phenopacket: Choose the least identifying time representation that still answers the question. Read every structured field beside its original observation, time context and review status.
For future study, connect data standards to biology, medicine, computing and ethics while checking official pathways separately.
Section 8 of 36
8. Represent diagnoses without overclaiming
Disease entries can record a diagnosis and timing, but the code does not itself establish that the diagnosis is correct or current.
Working checkpoint — phenotypic feature: Attach provenance and review status to the asserted concept. Preserve unknown and excluded states; a clean packet is not proof of completeness.
For scientific writing, name the Phenopacket schema and ontology versions.
Section 9 of 36
9. Record measurements
Measurements can connect an assay, value, unit, reference range and time. Units and methods are part of the meaning.
Working checkpoint — time element: Validate units and keep the laboratory method visible. Separate technical schema validity from scientific correctness and sharing permission.
For students, ask which fields are direct observations, mappings or later interpretations.
Section 10 of 36
10. Describe medical actions carefully
Version 2 supports medical actions such as procedures, treatments and related records when relevant to the case.
Working checkpoint — biosample: Do not infer response or causality merely from sequence in time. Treat subject, sample, ontology and variant identifiers as scientific data: a silent mismatch can join the wrong evidence.
For parents, ask whether a portable record is complete, reviewed and authorised for its stated use.
Section 11 of 36
11. Connect biosamples
Biosample records link specimens to tissue, collection context and derived samples. A sample is not interchangeable with the person.
Working checkpoint — ontology identifier: Track sample lineage and identifiers explicitly. Keep the source record, Phenopacket, schema, terminology releases and transformation log together.
For a classroom model, use fictional creatures and invented measurements rather than health records.
Section 12 of 36
12. Represent interpretations
Interpretations can connect a genomic interpretation, candidate diagnosis and supporting variant evidence.
Working checkpoint — computable case representation: Label status and evidence instead of turning an interpretation into fact. Read every structured field beside its original observation, time context and review status.
For future study, connect data standards to biology, medicine, computing and ethics while checking official pathways separately.
Section 13 of 36
13. Reference external files
Large sequence or imaging files can be referenced rather than embedded, preserving location and metadata.
Working checkpoint — Phenopacket: Use governed URIs, checksums and access rules. Preserve unknown and excluded states; a clean packet is not proof of completeness.
For scientific writing, name the Phenopacket schema and ontology versions.
Section 14 of 36
14. Include resource metadata
Ontology and reference resources change. Resource metadata tells a recipient which version supplied an identifier or definition.
Working checkpoint — phenotypic feature: Pin prefixes, versions and namespaces. Separate technical schema validity from scientific correctness and sharing permission.
For students, ask which fields are direct observations, mappings or later interpretations.
Section 15 of 36
15. Use HPO for phenotype meaning
HPO identifiers allow computational comparison of phenotypic abnormalities while labels keep the record readable.
Working checkpoint — time element: Review specificity, negation and age-dependent observations. Treat subject, sample, ontology and variant identifiers as scientific data: a silent mismatch can join the wrong evidence.
For parents, ask whether a portable record is complete, reviewed and authorised for its stated use.
Section 16 of 36
16. Use disease ontologies responsibly
Mondo, ORDO and other disease resources can support interoperable disease identifiers, but their mappings and scopes differ.
Working checkpoint — biosample: Retain source-system identifiers and mapping provenance. Keep the source record, Phenopacket, schema, terminology releases and transformation log together.
For a classroom model, use fictional creatures and invented measurements rather than health records.
Section 17 of 36
17. Connect variants through GA4GH standards
GA4GH states that Phenopackets can interoperate with VRS so genomic variation can be represented consistently.
Working checkpoint — ontology identifier: Validate the variant representation, assembly and reference sequence. Read every structured field beside its original observation, time context and review status.
For future study, connect data standards to biology, medicine, computing and ethics while checking official pathways separately.
Section 18 of 36
18. Do not confuse a link with the data
A file reference or external identifier may fail, change or require permission even when the packet validates.
Working checkpoint — computable case representation: Test access under the recipient’s actual authorisation. Preserve unknown and excluded states; a clean packet is not proof of completeness.
For scientific writing, name the Phenopacket schema and ontology versions.
Section 19 of 36
19. Choose JSON or Protocol Buffers knowingly
The schema supports serialisations suited to APIs, files and generated software classes. Equivalent structure does not remove implementation differences.
Working checkpoint — Phenopacket: Round-trip a test packet through the intended toolchain. Separate technical schema validity from scientific correctness and sharing permission.
For students, ask which fields are direct observations, mappings or later interpretations.
Section 20 of 36
20. Validate against the schema
Schema validation catches missing required fields and wrong data shapes; it cannot detect every implausible phenotype or swapped sample.
Working checkpoint — phenotypic feature: Run technical and domain review as separate gates. Treat subject, sample, ontology and variant identifiers as scientific data: a silent mismatch can join the wrong evidence.
For parents, ask whether a portable record is complete, reviewed and authorised for its stated use.
Section 21 of 36
21. Validate terminology
An accepted ontology identifier can still be too broad, obsolete or mismatched to the clinical observation.
Working checkpoint — time element: Open the current term and inspect its definition and status. Keep the source record, Phenopacket, schema, terminology releases and transformation log together.
For a classroom model, use fictional creatures and invented measurements rather than health records.
Section 22 of 36
22. Preserve provenance
A reproducible packet explains who or what created each mapping, when it was reviewed and which source record supports it.
Working checkpoint — biosample: Keep transformation logs and reviewer decisions. Read every structured field beside its original observation, time context and review status.
For future study, connect data standards to biology, medicine, computing and ethics while checking official pathways separately.
Section 23 of 36
23. Connect carefully with FHIR
The GA4GH product page describes interoperability with HL7 FHIR; interoperability is implemented through mappings and profiles, not assumed identity between every field.
Working checkpoint — ontology identifier: Test the exact implementation guide and round-trip semantics. Preserve unknown and excluded states; a clean packet is not proof of completeness.
For scientific writing, name the Phenopacket schema and ontology versions.
Section 24 of 36
24. Minimise sensitive data
Rare combinations of phenotype, age, family structure and genotype can identify someone even after names are removed.
Working checkpoint — computable case representation: Share only fields required by the approved purpose. Separate technical schema validity from scientific correctness and sharing permission.
For students, ask which fields are direct observations, mappings or later interpretations.
Section 25 of 36
25. Respect consent and purpose
A technically portable record is not automatically authorised for research, care, publication or cross-border transfer.
Working checkpoint — Phenopacket: Check consent, governance and recipient obligations. Treat subject, sample, ontology and variant identifiers as scientific data: a silent mismatch can join the wrong evidence.
For parents, ask whether a portable record is complete, reviewed and authorised for its stated use.
Section 26 of 36
26. Plan for reanalysis
New observations, ontology releases and variant interpretations can change a case. Overwriting loses the history of scientific reasoning.
Working checkpoint — phenotypic feature: Version packets and describe what changed. Keep the source record, Phenopacket, schema, terminology releases and transformation log together.
For a classroom model, use fictional creatures and invented measurements rather than health records.
Section 27 of 36
27. Use packages in benchmarking
PhEval and Phenopacket Store show how standardised case packages can support reproducible phenotype-driven evaluation.
Working checkpoint — time element: Keep corpus governance and truth-label provenance visible. Read every structured field beside its original observation, time context and review status.
For future study, connect data standards to biology, medicine, computing and ethics while checking official pathways separately.
Section 28 of 36
28. Connect to Primary Science
A fictional observation card can separate what was seen, when it was seen and how confidently it was classified.
Working checkpoint — biosample: Use imaginary organisms and no personal health data. Preserve unknown and excluded states; a clean packet is not proof of completeness.
For scientific writing, name the Phenopacket schema and ontology versions.
Section 29 of 36
29. Connect to Secondary Science
Genetics, variation, inheritance and phenotype become a connected evidence record rather than isolated vocabulary.
Working checkpoint — ontology identifier: Ask students which fields are observations and which are interpretations. Separate technical schema validity from scientific correctness and sharing permission.
For students, ask which fields are direct observations, mappings or later interpretations.
Section 30 of 36
30. Connect to mathematics
Units, ranges, ages and timelines show that numbers need definitions and reference frames.
Working checkpoint — computable case representation: Convert a toy measurement only after checking its unit. Treat subject, sample, ontology and variant identifiers as scientific data: a silent mismatch can join the wrong evidence.
For parents, ask whether a portable record is complete, reviewed and authorised for its stated use.
Section 31 of 36
31. Connect to computing
Schemas, APIs, identifiers and validation rules make data portable while exposing why strict types matter.
Working checkpoint — Phenopacket: Write tests for missing identifiers and invalid time intervals. Keep the source record, Phenopacket, schema, terminology releases and transformation log together.
For a classroom model, use fictional creatures and invented measurements rather than health records.
Section 32 of 36
32. Connect to pathways
Computable clinical data links biology, medicine, software, statistics and governance and can illuminate STEM pathways.
Working checkpoint — phenotypic feature: Check current official education and professional requirements separately. Read every structured field beside its original observation, time context and review status.
For future study, connect data standards to biology, medicine, computing and ethics while checking official pathways separately.
Section 33 of 36
33. Did you know?
GA4GH explains that Phenopackets version 2 expanded the model beyond rare disease to support cancer and common-disease use cases.
Working checkpoint — time element: A broader schema still needs a specific purpose and governed profile. Preserve unknown and excluded states; a clean packet is not proof of completeness.
For scientific writing, name the Phenopacket schema and ontology versions.
Section 34 of 36
34. Write a reader-ready packet report
Name the schema version, resources, included entities, transformations, validation results, privacy limits and intended use.
Working checkpoint — biosample: Keep content quality separate from technical conformance. Separate technical schema validity from scientific correctness and sharing permission.
For students, ask which fields are direct observations, mappings or later interpretations.
Section 35 of 36
35. Build a validation ladder
Validate syntax, identifiers, time and units, then review scientific meaning, provenance, consent and recipient access.
Working checkpoint — ontology identifier: Fix the earliest weak rung before exchange. Treat subject, sample, ontology and variant identifiers as scientific data: a silent mismatch can join the wrong evidence.
For parents, ask whether a portable record is complete, reviewed and authorised for its stated use.
Section 36 of 36
36. Finish with a reproducible bundle
Deliver the permitted packet, schema and ontology versions, transformation code, validation report, checksums, review notes and access conditions.
Working checkpoint — computable case representation: Make the case reconstructable without exposing unnecessary identity. Keep the source record, Phenopacket, schema, terminology releases and transformation log together.
For a classroom model, use fictional creatures and invented measurements rather than health records.
Contents · Previous section · Continue to the Science Learning Hub