eduKateSG · Why Science?
Describe the biological variation precisely enough that different systems can recognise the same object without guessing from a string
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 Global Alliance for Genomics and Health Variation Representation Specification, GA4GH VRS, is an open, community-developed framework for precise computable representation and exchange of biological sequence variation. Its model separates locations, sequence expressions and variation objects, supplies machine-readable schemas and supports identifiers derived from normalised content. This helps different databases recognise equivalent variation without relying on ambiguous prose or local identifiers. VRS answers “what variation object is this?” It does not decide whether the variation is pathogenic, associated with a trait or relevant to one person. 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: Ensembl VEP and transcript-aware variant consequences; BCFtools/csq and haplotype-aware consequences; ClinVar and evidence-aware variant classifications; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 11 October 2026: Official GA4GH VRS product page; Official GA4GH VRS stable introduction; Official GA4GH VRS repository; GA4GH VRS primary 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 |
|---|---|---|---|
| Reference | Which sequence and coordinate system define the location? | Sequence identifier, interval and specification version | A precise reference context |
| Represent | What state or change exists at that location? | VRS variation object and sequence expression | A computable biological object |
| Identify | Can equivalent normalised objects share an identifier? | Canonical serialisation and computed digest | A federated identifier |
| Interpret | What evidence is asserted about the object? | Separate annotation, classification and provenance models | Meaning linked to, not hidden inside, identity |
Inside this guide
1–12 · Foundations and design
- 1. Start with the communication problem
- 2. Distinguish human-readable from computable
- 3. Separate location from state
- 4. Identify the reference sequence
- 5. Use inter-residue coordinates carefully
- 6. Represent an allele as an object
- 7. Normalise before identifying
- 8. Understand computed identifiers
- 9. Do not hash arbitrary JSON
- 10. Read the schema
- 11. Pin the specification version
- 12. Preserve source expressions
13–24 · Models and estimation
- 13. Map VCF deliberately
- 14. Map HGVS deliberately
- 15. Map SPDI deliberately
- 16. Handle repetitive sequence
- 17. Treat reference context as evidence
- 18. Represent larger changes cautiously
- 19. Validate identifiers across implementations
- 20. Use a reference implementation transparently
- 21. Keep provenance outside the digest
- 22. Keep classification separate
- 23. Keep association separate
- 24. Design round-trip tests
25–36 · Diagnostics and interpretation
- 25. Design collision and mismatch tests
- 26. Use services with contracts
- 27. Distinguish VRS from Cat-VRS and VA-Spec
- 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 the communication problem
The same biological variant can be written as genomic, transcript or protein strings and can receive local database identifiers. Computers need a precise shared object beneath those labels.
Working checkpoint — variation object: Define the biological object before merging annotations. 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. Distinguish human-readable from computable
HGVS and VCF are invaluable communication and file standards, but equivalent forms and context-dependent parsing can complicate federated matching. VRS complements rather than replaces them.
Working checkpoint — sequence location: Store source expressions alongside the computed representation. 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. Separate location from state
A sequence variation object needs both where it occurs and what sequence state is present. Combining them implicitly invites coordinate and allele mistakes.
Working checkpoint — sequence expression: Validate location and state as separate fields. 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. Identify the reference sequence
A coordinate is meaningful only on a named reference sequence. Assembly labels alone may be too broad when transcripts and contigs are involved.
Working checkpoint — computed identifier: Use stable sequence identifiers and retain the source accession. 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. Use inter-residue coordinates carefully
VRS models intervals between residues, which avoids some start/end ambiguity but differs from common one-based inclusive conventions.
Working checkpoint — normalisation: Test boundary conversions with single-base substitutions and empty intervals. 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. Represent an allele as an object
An allele links a sequence location with a sequence expression. The object says what sequence state exists at the defined place.
Working checkpoint — interoperability: Keep representation separate from predicted consequence. 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. Normalise before identifying
Equivalent repeat-region changes can be written at different positions. VRS normalisation rules aim to produce consistent objects for identification.
Working checkpoint — variation object: Compare equivalent indel inputs in regression tests. 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. Understand computed identifiers
A GA4GH computed identifier is derived from normalised object content rather than assigned by one central database. Matching content can therefore match across systems.
Working checkpoint — sequence location: Record the algorithm and specification version used. 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. Do not hash arbitrary JSON
A digest is stable only when the object has been normalised and serialised under the defined rules. Whitespace-free JSON alone is not a VRS identifier.
Working checkpoint — sequence expression: Use the reference implementation or a validated conforming library. 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 schema
Machine-readable schemas define allowed object shapes and types. Schema validation catches missing fields but cannot prove biological correctness.
Working checkpoint — computed identifier: Run schema and biological validation separately. 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. Pin the specification version
VRS evolves through community releases. An object or identifier produced under one version may need migration or compatibility testing.
Working checkpoint — normalisation: Store the VRS and implementation versions with every identifier. 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. Preserve source expressions
The original VCF, HGVS, SPDI or database identifier carries provenance and can help diagnose conversion errors. A canonical object should not erase its inputs.
Working checkpoint — interoperability: Maintain a bidirectional mapping table. 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. Map VCF deliberately
VCF coordinates are one-based and alleles include anchoring conventions for indels. Conversion must reconstruct the intended sequence change before creating a VRS object.
Working checkpoint — variation object: Test substitutions, insertions, deletions and symbolic alleles. 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. Map HGVS deliberately
HGVS descriptions depend on reference sequences, transcript versions and variant class rules. Parsing a string without sequence validation can produce the wrong object.
Working checkpoint — sequence location: Retain the exact expression and resolution warnings. 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. Map SPDI deliberately
SPDI expresses sequence, position, deletion and insertion. It can be a useful bridge, but reference and normalisation details still matter.
Working checkpoint — sequence expression: Verify the resolved reference sequence before comparison. 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. Handle repetitive sequence
Repeat regions create multiple equivalent placements for an indel. Canonical normalisation supports matching but does not reveal biological mechanism.
Working checkpoint — computed identifier: Keep the local sequence window for audit. 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. Treat reference context as evidence
Two coordinates on different assemblies or transcripts can describe related but not automatically identical sequence states. Liftover is a hypothesis that requires allele checking.
Working checkpoint — normalisation: Verify both source and destination reference alleles. 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. Represent larger changes cautiously
Variation beyond simple alleles may involve copy number, haplotypes or systemic observations. Use only the VRS object types supported by the pinned version and implementation.
Working checkpoint — interoperability: Reject unsupported objects instead of approximating silently. 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. Validate identifiers across implementations
Federated identification works only when independent implementations normalise and serialize the same object consistently.
Working checkpoint — variation object: Exchange a shared conformance set and compare computed IDs. 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. Use a reference implementation transparently
The official repository supplies code and schemas that can accelerate adoption. Production pipelines still need pinned dependencies and tests.
Working checkpoint — sequence location: Archive package versions and conformance results. 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. Keep provenance outside the digest
A variation identifier should remain stable across institutions, while assertions, submitters and dates change. Provenance belongs in linked records.
Working checkpoint — sequence expression: Model who said what and when separately. 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. Keep classification separate
A VRS object identifies variation; it does not say benign, pathogenic or uncertain. Several organisations can attach different classifications to the same object.
Working checkpoint — computed identifier: Link ClinVar-style assertions without embedding them in the identity. 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. Keep association separate
A GWAS signal or functional annotation is evidence about a variation object, not part of its sequence identity.
Working checkpoint — normalisation: Attach evidence with source and study 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 24 of 36
24. Design round-trip tests
Convert a source expression to VRS and back to an equivalent source form. Losses and warnings reveal unsupported assumptions.
Working checkpoint — interoperability: Include reference sequence, coordinate and allele checks in the round trip. 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. Design collision and mismatch tests
Different objects must not share an identifier, while equivalent normalised objects should. Both false merges and missed merges are serious.
Working checkpoint — variation object: Test nearby indels, transcript mappings and multiallelic records. 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. Use services with contracts
A service that resolves or translates variants should publish supported references, versions, errors and rate limits. A correct model can still be wrapped by an unreliable service.
Working checkpoint — sequence location: Log request, response and service version. 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. Distinguish VRS from Cat-VRS and VA-Spec
GA4GH now maintains related products for categorical variation and variant annotations. They address different layers from precise variation identity.
Working checkpoint — sequence expression: Name the specification responsible for each field. 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
Two cards can use different words for the same shape. A shared construction rule lets the class decide whether the objects are identical.
Working checkpoint — computed identifier: Use geometric objects rather than 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
DNA changes can be described on genome, transcript and protein sequences. Reference context explains why strings differ even when related.
Working checkpoint — normalisation: Trace one fictional substitution through each sequence level. 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
Coordinates, intervals, equivalence classes and deterministic functions underpin normalisation and computed identifiers.
Working checkpoint — interoperability: Prove boundary cases with small sequence examples. 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
Schemas, canonical serialisation, digests and conformance tests turn biological identity into interoperable infrastructure.
Working checkpoint — variation object: Separate syntactic validity from semantic correctness. 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
Genomic standards join molecular biology, ontology design, cryptographic identifiers and software engineering and can illuminate STEM education and career pathways.
Working checkpoint — sequence location: Use current official programme information for real pathway 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?
VRS identifiers can be computed from normalised content, so two systems can recognise the same represented variation without asking one registry to assign the ID first.
Working checkpoint — sequence expression: The result works only when both systems follow the same pinned 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 34 of 36
34. Write a reader-ready result
A strong report names the source expression, reference sequence, VRS object type, normalised object, computed identifier, specification version and conversion warnings.
Working checkpoint — computed identifier: Keep interpretation and evidence in linked fields. 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
Validate source syntax, resolve sequences, convert coordinates, normalise the object, check the schema, compare identifiers and then link external evidence.
Working checkpoint — normalisation: Do not interpret an object that has not passed identity checks. 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 source records, reference-sequence manifest, VRS objects, computed identifiers, spec and library versions, conversion code, conformance tests, warnings and provenance links.
Working checkpoint — interoperability: Make every merge and mismatch explainable. 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