eduKateSG · Why Science?
Identify a reference sequence by what it contains—not only by what one database happens to call it
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. refget is a set of GA4GH standards for identifying and retrieving reference biological sequences. Refget Sequences v2.0 is listed as current and approved by GA4GH. Its content-derived digests help distinguish identical sequence content from database-specific names, and its API can return whole sequences, subsequences and metadata. A digest identifies normalised sequence content; it does not by itself identify a genome assembly, coordinate system, biological sample or scientific interpretation. 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: GA4GH VRS and computable variant representation; gnomAD and population allele-frequency context; GA4GH Beacon v2 and federated genomic discovery; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 12 October 2026: Official GA4GH refget Sequences product page, approved v2.0; Official Refget API Specification v2.0.0; Official GA4GH refget specifications hub; Refget standard paper; 2026 Singapore–Cambridge O-Level Biology syllabus; MOE G2/G3 Lower Secondary Science teaching and learning syllabus; 2026 MOE G2 Computing syllabus. Standards, implementations and recommendations can change, so preserve the version, access date and documentation used.
| Stage | Question | Evidence | Responsible output |
|---|---|---|---|
| Normalise | Which symbols define the sequence content? | Whitespace removal, uppercase conversion and allowed alphabet | Canonical bytes for hashing |
| Identify | Which digest names those bytes? | GA4GH checksum or another supported algorithm | A content-derived sequence identifier |
| Retrieve | Can the exact sequence or range be returned? | Refget API, media type and Range request | Verifiable full or partial sequence |
| Contextualise | How was this sequence used? | Assembly, aliases, tool versions and provenance | A reproducible reference context |
Inside this guide
1–12 · Foundations and design
- 1. Start with naming ambiguity
- 2. Understand the approved product
- 3. Identify by sequence content
- 4. Normalise before hashing
- 5. Understand the GA4GH digest
- 6. Keep algorithm names attached
- 7. Treat TRUNC512 as deprecated
- 8. Use namespaced identifiers cautiously
- 9. Retrieve the full sequence
- 10. Retrieve a subsequence
- 11. Read metadata separately
- 12. Use service information
13–24 · Models and estimation
- 13. Respect media types
- 14. Follow redirects safely
- 15. Handle range errors
- 16. Do not use refget for private genomes
- 17. Connect refget to CRAM
- 18. Connect refget to VRS
- 19. Distinguish a sequence from an assembly
- 20. Use Refget Sequence Collections when needed
- 21. Recognise allowed biological sequences
- 22. Preserve aliases as context
- 23. Detect changed references
- 24. Cache by content safely
25–36 · Diagnostics and interpretation
- 25. Test compliance
- 26. Record provenance beyond the digest
- 27. Handle unknown identifiers honestly
- 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 reference report
- 35. Build a validation ladder
- 36. Finish with a reproducible bundle
Section 1 of 36
1. Start with naming ambiguity
The same chromosome sequence can appear under several database names, while similar names can refer to different releases.
Working checkpoint — refget: Record content identity separately from local display names. Keep the refget version, normalisation rule, algorithm, digest and retrieved bytes together.
For a classroom model, use short invented letter strings.
Section 2 of 36
2. Understand the approved product
GA4GH lists Refget Sequences v2.0 as current; the official technical page identifies Refget API Specification v2.0.0.
Working checkpoint — reference sequence: Pin that version when describing behaviour. Read every reference claim beside its content identity, range and assembly context.
For future study, connect hashing to genetics, mathematics and computing while checking official pathways separately.
Section 3 of 36
3. Identify by sequence content
refget derives a checksum from the sequence itself, reducing dependence on one central accessioning authority.
Working checkpoint — GA4GH digest: Recompute the digest from retrieved content. Preserve aliases and failed lookups; one service catalogue is not the whole reference universe.
For scientific writing, name the digest algorithm, refget version and assembly context.
Section 4 of 36
4. Normalise before hashing
The v2 specification removes non-base symbols such as line breaks and spaces, then uppercases the remaining A–Z sequence symbols.
Working checkpoint — checksum: Use the exact normalisation rule, not a home-made FASTA parser. Separate sequence identity from assembly identity and biological interpretation.
For students, ask whether two differently named strings contain exactly the same sequence.
Section 5 of 36
5. Understand the GA4GH digest
The ga4gh algorithm applies SHA-512 to the sanitised sequence, keeps the first 24 digest bytes, encodes them URL-safe and prefixes SQ.
Working checkpoint — subsequence: Test implementations against the published ACGT example. Treat sequence bytes, digest namespace and range boundaries as scientific data.
For parents, ask whether reference content was verified rather than trusted by filename.
Section 6 of 36
6. Keep algorithm names attached
The same sequence may have MD5, ga4gh or legacy TRUNC512 representations; a bare digest is ambiguous without its algorithm rules.
Working checkpoint — provenance: Store the namespace and checksum together. Keep the refget version, normalisation rule, algorithm, digest and retrieved bytes together.
For a classroom model, use short invented letter strings.
Section 7 of 36
7. Treat TRUNC512 as deprecated
The specification explains conversion compatibility but discourages continued use of the older TRUNC512 representation in favour of the GA4GH sequence checksum.
Working checkpoint — refget: Migrate while preserving legacy aliases. Read every reference claim beside its content identity, range and assembly context.
For future study, connect hashing to genetics, mathematics and computing while checking official pathways separately.
Section 8 of 36
8. Use namespaced identifiers cautiously
A service may resolve CURIE-like accessions such as an INSDC identifier, but conflicts must return multiple-choice or conflict responses rather than arbitrary selection.
Working checkpoint — reference sequence: Prefer content digests for unambiguous identity. Preserve aliases and failed lookups; one service catalogue is not the whole reference universe.
For scientific writing, name the digest algorithm, refget version and assembly context.
Section 9 of 36
9. Retrieve the full sequence
A standard request can return the sequence bytes corresponding to an identifier as plain US-ASCII without line breaks.
Working checkpoint — GA4GH digest: Verify length and checksum after retrieval. Separate sequence identity from assembly identity and biological interpretation.
For students, ask whether two differently named strings contain exactly the same sequence.
Section 10 of 36
10. Retrieve a subsequence
Range requests are the preferred method for asking a server for part of a sequence.
Working checkpoint — checksum: Record range convention and response headers. Treat sequence bytes, digest namespace and range boundaries as scientific data.
For parents, ask whether reference content was verified rather than trusted by filename.
Section 11 of 36
11. Read metadata separately
The metadata endpoint can describe known identifiers, length and related information without making the sequence body carry every context field.
Working checkpoint — subsequence: Archive metadata and sequence response together. Keep the refget version, normalisation rule, algorithm, digest and retrieved bytes together.
For a classroom model, use short invented letter strings.
Section 12 of 36
12. Use service information
A service endpoint can report implementation and capability details needed for compatibility checks.
Working checkpoint — provenance: Capture service identity and software version. Read every reference claim beside its content identity, range and assembly context.
For future study, connect hashing to genetics, mathematics and computing while checking official pathways separately.
Section 13 of 36
13. Respect media types
Protocol-specific text and JSON media types communicate version and response shape; a generic successful response can still be wrong for a client.
Working checkpoint — refget: Validate Content-Type before parsing. Preserve aliases and failed lookups; one service catalogue is not the whole reference universe.
For scientific writing, name the digest algorithm, refget version and assembly context.
Section 14 of 36
14. Follow redirects safely
A server may redirect retrieval to another location, and clients should follow valid redirects while maintaining security checks.
Working checkpoint — reference sequence: Reject insecure scheme downgrades. Separate sequence identity from assembly identity and biological interpretation.
For students, ask whether two differently named strings contain exactly the same sequence.
Section 15 of 36
15. Handle range errors
A 416 response tells the client that the requested range cannot be satisfied, which is different from a missing sequence.
Working checkpoint — GA4GH digest: Check length and interval boundaries before retrying. Treat sequence bytes, digest namespace and range boundaries as scientific data.
For parents, ask whether reference content was verified rather than trusted by filename.
Section 16 of 36
16. Do not use refget for private genomes
The specification says reference sequence is public and warns that adapting refget to retrieve private or sensitive genomic data is not a good use.
Working checkpoint — checksum: Use governed data-access services for personal sequence data. Keep the refget version, normalisation rule, algorithm, digest and retrieved bytes together.
For a classroom model, use short invented letter strings.
Section 17 of 36
17. Connect refget to CRAM
CRAM can rely on an external reference; a content-derived identifier helps a decoder obtain exactly the sequence used for compression.
Working checkpoint — subsequence: Verify the retrieved reference before decoding. Read every reference claim beside its content identity, range and assembly context.
For future study, connect hashing to genetics, mathematics and computing while checking official pathways separately.
Section 18 of 36
18. Connect refget to VRS
GA4GH VRS uses precise reference context for computable variation, and the refget ga4gh digest derives from work shared with VRS.
Working checkpoint — provenance: Keep the sequence identifier with every location-based allele. Preserve aliases and failed lookups; one service catalogue is not the whole reference universe.
For scientific writing, name the digest algorithm, refget version and assembly context.
Section 19 of 36
19. Distinguish a sequence from an assembly
One digest identifies one normalised sequence, while an assembly is a collection with ordering, names and coordinate-system context.
Working checkpoint — refget: Use Sequence Collections for whole-reference identity questions. Separate sequence identity from assembly identity and biological interpretation.
For students, ask whether two differently named strings contain exactly the same sequence.
Section 20 of 36
20. Use Refget Sequence Collections when needed
The companion standard identifies and compares collections such as genomes, transcriptomes and proteomes.
Working checkpoint — reference sequence: Do not call one chromosome digest a genome build. Treat sequence bytes, digest namespace and range boundaries as scientific data.
For parents, ask whether reference content was verified rather than trusted by filename.
Section 21 of 36
21. Recognise allowed biological sequences
The design can support DNA, RNA, cDNA, coding sequences or proteins represented as character strings.
Working checkpoint — GA4GH digest: Declare alphabet and biological type in surrounding metadata. Keep the refget version, normalisation rule, algorithm, digest and retrieved bytes together.
For a classroom model, use short invented letter strings.
Section 22 of 36
22. Preserve aliases as context
Database accessions remain useful for people and source provenance even when content digests establish identity.
Working checkpoint — checksum: Store aliases without letting them override a digest mismatch. Read every reference claim beside its content identity, range and assembly context.
For future study, connect hashing to genetics, mathematics and computing while checking official pathways separately.
Section 23 of 36
23. Detect changed references
A one-character content change produces a different digest, making silent reference drift visible.
Working checkpoint — subsequence: Fail the workflow when expected and observed digests differ. Preserve aliases and failed lookups; one service catalogue is not the whole reference universe.
For scientific writing, name the digest algorithm, refget version and assembly context.
Section 24 of 36
24. Cache by content safely
Content-derived IDs make immutable sequence caching practical, provided the implementation verifies the returned bytes.
Working checkpoint — provenance: Key the cache by algorithm and digest. Separate sequence identity from assembly identity and biological interpretation.
For students, ask whether two differently named strings contain exactly the same sequence.
Section 25 of 36
25. Test compliance
The official specification points implementers to a compliance suite and public implementation reports.
Working checkpoint — refget: Run conformance tests before trusting cross-service behaviour. Treat sequence bytes, digest namespace and range boundaries as scientific data.
For parents, ask whether reference content was verified rather than trusted by filename.
Section 26 of 36
26. Record provenance beyond the digest
The digest tells what content was used, not who curated it, why it was selected or which assembly named it.
Working checkpoint — reference sequence: Attach source accession, retrieval date and analysis purpose. Keep the refget version, normalisation rule, algorithm, digest and retrieved bytes together.
For a classroom model, use short invented letter strings.
Section 27 of 36
27. Handle unknown identifiers honestly
A 404 means the service cannot supply that identifier; it does not prove the sequence does not exist elsewhere.
Working checkpoint — GA4GH digest: Try another trusted refget service without changing the digest. Read every reference claim beside its content identity, range and assembly context.
For future study, connect hashing to genetics, mathematics and computing while checking official pathways separately.
Section 28 of 36
28. Connect to Primary Science
Two instruction cards with different titles but identical letter strings show how content can establish sameness.
Working checkpoint — checksum: Use short invented sequences and compare carefully. Preserve aliases and failed lookups; one service catalogue is not the whole reference universe.
For scientific writing, name the digest algorithm, refget version and assembly context.
Section 29 of 36
29. Connect to Secondary Science
Reference sequences, variation and mutations become clearer when students can separate a DNA string from the database name attached to it.
Working checkpoint — subsequence: Hash a fictional string and explain what the digest does not say. Separate sequence identity from assembly identity and biological interpretation.
For students, ask whether two differently named strings contain exactly the same sequence.
Section 30 of 36
30. Connect to mathematics
Hash functions map long inputs to fixed-length identifiers; the practical lesson is verification, not treating hashes as magic.
Working checkpoint — provenance: Change one character and compare outputs. Treat sequence bytes, digest namespace and range boundaries as scientific data.
For parents, ask whether reference content was verified rather than trusted by filename.
Section 31 of 36
31. Connect to computing
Normalisation, hashing, HTTP ranges, redirects and caching make refget a compact lesson in content-addressed systems.
Working checkpoint — refget: Test whitespace, case and invalid symbols. Keep the refget version, normalisation rule, algorithm, digest and retrieved bytes together.
For a classroom model, use short invented letter strings.
Section 32 of 36
32. Connect to pathways
Reference infrastructure links genetics, bioinformatics, software and data stewardship and can illuminate STEM pathways.
Working checkpoint — reference sequence: Check current official course and professional requirements separately. Read every reference claim beside its content identity, range and assembly context.
For future study, connect hashing to genetics, mathematics and computing while checking official pathways separately.
Section 33 of 36
33. Did you know?
The official v2 specification gives ACGT the GA4GH digest SQ.aKF498dAxcJAqme6QYQ7EZ07-fiw8Kw2, providing a precise interoperability test.
Working checkpoint — GA4GH digest: Reproduce the example before processing large references. Preserve aliases and failed lookups; one service catalogue is not the whole reference universe.
For scientific writing, name the digest algorithm, refget version and assembly context.
Section 34 of 36
34. Write a reader-ready reference report
Name the refget version, digest algorithm, identifier, length, range, aliases, service, retrieval date and assembly context.
Working checkpoint — checksum: Do not let an alias stand in for verified content. Separate sequence identity from assembly identity and biological interpretation.
For students, ask whether two differently named strings contain exactly the same sequence.
Section 35 of 36
35. Build a validation ladder
Normalise the sequence, compute its digest, retrieve by ID, verify bytes and range, then connect it to the correct assembly and analysis.
Working checkpoint — subsequence: Fix content identity before interpreting variation. Treat sequence bytes, digest namespace and range boundaries as scientific data.
For parents, ask whether reference content was verified rather than trusted by filename.
Section 36 of 36
36. Finish with a reproducible bundle
Deliver the permitted reference identifiers, checksums, metadata, exact sequences or access instructions, tool versions, tests and provenance notes. Include a small verification fixture so another implementation can repeat the normalisation, digest, retrieval and range checks without guessing which line endings, letter case or aliases were used.
Working checkpoint — provenance: Make every coordinate claim traceable to verified reference content, and keep the fixture separate from any personal or controlled sequence data. Keep the refget version, normalisation rule, algorithm, digest and retrieved bytes together.
For a classroom model, use short invented letter strings.
Contents · Previous section · Continue to the Science Learning Hub