eduKateSG · Why Science?
Read a long-sequence mapping in one line—while keeping query, target, strand, approximate matching, optional alignment detail and uncertainty visible
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 minimap2 manual defines Pairwise mApping Format, or PAF, as a tab-delimited text format with at least twelve fields per line. Those fields name a query and target, give their lengths and zero-based half-open intervals, state relative strand, count matching bases and alignment-block length, and report mapping quality from zero to 255 with 255 meaning missing. Optional fields use SAM-like typed tags. Minimap2 outputs PAF by default, and base-level alignment details such as CIGAR are optional. A PAF row is an algorithmic mapping record, not proof of uniqueness, homology, assembly correctness or biological function. 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: SAM human-readable sequence alignments; AXT pairwise genome alignment records; Chain files and genome coordinate mapping; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 12 October 2026: Official minimap2 manual, current project documentation, checked 12 October 2026; Official minimap2 project repository, checked 12 October 2026; Official minimap2 cookbook, checked 12 October 2026; 2026 Singapore–Cambridge O-Level Biology syllabus; MOE G2/G3 Lower Secondary Science teaching and learning syllabus; 2026 MOE G2 Computing syllabus. Specifications, tools and recommendations can change, so preserve dated documentation, exact source data, coordinate rules and software versions.
A deeper reading route
PAF is compact because one line records the outer geometry of a query-to-target mapping. Query and target names, full lengths and mapped intervals establish two coordinate spaces. The relative strand links their orientations. Match count, block length and mapping quality summarise evidence. The record remains useful even when base-by-base alignment detail is absent, but its limits must be stated.
Both query and target starts are zero-based and their ends are open. The span is end minus start, with no plus one. This convention makes interval arithmetic convenient but can conflict with one-based inclusive formats. A conversion test should include one-base mappings, chromosome starts and reverse-strand examples so an off-by-one error cannot hide inside a long alignment.
The strand field expresses whether query and target have the same or opposite relative orientation. It does not by itself say which sequence is biologically forward or ancestral. Query coordinates remain reported on the query’s original orientation under the format’s rules, so sequence extraction and display must use an explicit strand transform.
Field ten counts matching bases, while field eleven gives alignment block length including mismatches and gaps under the producer’s definition. Their ratio can summarise identity-like behaviour, but only when the denominator and generation mode are reported. Approximate mapping without base-level alignment should not be narrated as if every mismatch and indel were reconstructed.
Mapping quality occupies the twelfth field. The manual uses the zero-to-255 range and reserves 255 for missing. Minimap2 also warns that mapping quality is incorrect with a multi-part index under specified conditions. A report should therefore preserve index construction, reference size, options and competing mappings rather than treating one integer as universal certainty.
Optional tags follow a typed TAG:TYPE:VALUE pattern. The cg tag can carry CIGAR when base alignment is requested, while cs can encode sequence differences and other tags describe scores, minimizers or mapping type. Tags depend on tool version and options. An absent tag means unavailable detail, not that the corresponding event did not occur.
Primary, secondary and inversion-related classifications can appear in optional tags or be influenced by command options. Repetitive sequence can create several plausible mappings. Filters that retain only one row may simplify downstream work while discarding the alternatives needed to evaluate uniqueness, structural variation or assembly ambiguity.
Validation should first confirm twelve required fields and ranges, then resolve query and target names against exact FASTA manifests. Intervals must fit sequence lengths, starts must not exceed ends, strand must be legal and counts must be plausible. Optional tags need type-aware parsing. Small fixtures should cover unmapped output, missing quality and reverse orientation.
PAF often connects tools: assembly-to-reference comparison, read overlap, liftover-like tasks and dot plots can all consume related mappings. Each workflow imposes additional assumptions. A record suitable for visualising an approximate contig placement may be inadequate for calling variants, where base-level alignment, reference context and calibrated validation matter.
Sorting and indexing are not intrinsic guarantees of PAF meaning. Analysts may sort by target coordinate for regional work or group by query for assembly review. State the order, use stable secondary keys and confirm that regrouping preserves the same rows and tags. Never infer biological priority from file order unless the producer explicitly defines it.
For students, PAF offers a clean lesson in model layers. A mapping can say where two long sequences probably correspond before it says exactly how every base aligns. That is not a weakness when the question only needs placement; it is a reminder to match the resolution of the evidence to the resolution of the claim.
A reproducible PAF bundle includes exact query and target FASTA hashes, minimap2 version, presets and options, index settings, command, complete output, filtering and sorting rules, tag definitions and validation summaries. With those pieces, another analyst can explain why a row exists, why alternatives disappeared and which details were never computed.
| Stage | Question | Evidence | Responsible output |
|---|---|---|---|
| Identify | Which sequences participate? | Query and target names plus full lengths | Two versioned sequence identities |
| Locate | Which intervals and orientation map? | Zero-based half-open bounds and strand | A pairwise placement claim |
| Qualify | How much agreement and uncertainty are recorded? | Matches, block length, MAPQ and typed tags | A bounded mapping summary |
| Judge | What can responsibly be concluded? | Alternatives, parameters, references and validation | A reproducible interpretation |
Inside this guide
1–12 · Foundations and design
- 1. Start with the mapping question
- 2. Use the producing manual
- 3. Count twelve required fields
- 4. Read query name
- 5. Read query length
- 6. Read query bounds
- 7. Read relative strand
- 8. Read target name
- 9. Read target length
- 10. Read target bounds
- 11. Read matching bases
- 12. Read alignment block length
13–24 · Models and estimation
- 13. Read mapping quality
- 14. Handle absent base alignment
- 15. Read typed optional tags
- 16. Use cg only when generated
- 17. Use cs as a difference encoding
- 18. Keep mapping type visible
- 19. Respect repetitive sequence
- 20. Beware multi-part index quality
- 21. Separate mapping from alignment
- 22. Convert coordinates deliberately
- 23. Sort for a declared purpose
- 24. Validate sequence identities
25–36 · Diagnostics and interpretation
- 25. Audit retained alternatives
- 26. Protect sensitive sequence data
- 27. Separate placement from function
- 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 mapping note
- 35. Build a validation ladder
- 36. Finish with a reproducible bundle
Section 1 of 36
1. Start with the mapping question
PAF can represent read, contig or assembly mappings, so interpretation begins by naming the query, target and task.
Working checkpoint — PAF: Record both sequence sets and intended use. 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. Use the producing manual
PAF is defined in the minimap2 documentation and evolves with its output tags and options.
Working checkpoint — pairwise mapping: Pin tool version and retrieval 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 3 of 36
3. Count twelve required fields
Every ordinary mapped PAF line has at least twelve tab-separated fields before optional tags.
Working checkpoint — query interval: Reject shifted or truncated rows. 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. Read query name
Field one identifies the query sequence under its source manifest.
Working checkpoint — target interval: Resolve aliases and duplicate names. 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. Read query length
Field two gives total query length, not mapped span.
Working checkpoint — mapping quality: Validate it against the exact FASTA. 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 query bounds
Fields three and four give zero-based start and open end on the query.
Working checkpoint — typed tag: Require zero at minimum and end within length. 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. Read relative strand
Field five is plus for same orientation and minus for opposite orientation.
Working checkpoint — PAF: Carry orientation into extraction and display. 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. Read target name
Field six identifies the target sequence.
Working checkpoint — pairwise mapping: Resolve it against the target manifest. 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. Read target length
Field seven gives the full target length.
Working checkpoint — query interval: Compare it with the indexed reference. 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 target bounds
Fields eight and nine give zero-based start and open end on the target.
Working checkpoint — target interval: Check span and chromosome edges. 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. Read matching bases
Field ten counts matching bases under the generated alignment or mapping.
Working checkpoint — mapping quality: State the producer and mode. 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. Read alignment block length
Field eleven counts the block length including differences according to the format output.
Working checkpoint — typed tag: Publish the denominator for any identity ratio. 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. Read mapping quality
Field twelve ranges from zero to 255, with 255 meaning unavailable.
Working checkpoint — PAF: Keep 255 distinct from maximum confidence. 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. Handle absent base alignment
Default approximate mappings can lack CIGAR and base-level differences.
Working checkpoint — pairwise mapping: Do not invent mismatches or indels. 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. Read typed optional tags
Additional fields use a tag, type and value grammar familiar from SAM-like formats.
Working checkpoint — query interval: Validate tag syntax and allowed types. 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. Use cg only when generated
The cg tag carries CIGAR when minimap2 is asked for base alignment.
Working checkpoint — target interval: Record the option that produced it. 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. Use cs as a difference encoding
The cs tag can describe sequence differences at a chosen detail level.
Working checkpoint — mapping quality: Parse it with versioned documentation. 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 mapping type visible
Optional classification tags can distinguish primary, secondary or other mapping classes.
Working checkpoint — typed tag: Inspect alternatives before uniqueness claims. 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. Respect repetitive sequence
Repeated query or target regions can create several high-scoring placements.
Working checkpoint — PAF: Report multi-mapping and filtering. 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. Beware multi-part index quality
The manual warns that mapping quality is incorrect with a multi-part index in the documented setting.
Working checkpoint — pairwise mapping: Archive index size settings and warnings. 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. Separate mapping from alignment
Outer placement can be estimated without a full base alignment.
Working checkpoint — query interval: Match claim resolution to computed detail. 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. Convert coordinates deliberately
PAF spans are zero-based half-open and need explicit conversion to one-based formats.
Working checkpoint — target interval: Test one-base and boundary fixtures. 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. Sort for a declared purpose
Target-coordinate and query-grouped orders support different workflows.
Working checkpoint — mapping quality: State order and stable tie-breakers. 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. Validate sequence identities
Names and lengths can still collide across releases.
Working checkpoint — typed tag: Use hashes or persistent sequence identifiers. 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. Audit retained alternatives
Filtering secondary rows changes the scientific question about uniqueness.
Working checkpoint — PAF: Count mappings before and after each filter. 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. Protect sensitive sequence data
Custom queries can be individual genomic data even when the file only stores coordinates and tags.
Working checkpoint — pairwise mapping: Apply access control and minimisation. 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. Separate placement from function
A strong mapping does not establish gene function, phenotype or clinical meaning.
Working checkpoint — query interval: Add independent biological evidence for stronger claims. 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 labelled ribbons can model where one pattern fits another and where uncertainty remains.
Working checkpoint — target interval: Use invented symbols rather than real sequence data. 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
Pairwise mapping supports lessons on comparison, variation and model resolution.
Working checkpoint — mapping quality: Separate placement, alignment and mechanism. 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
Half-open intervals, ratios and logarithmic quality scales connect PAF to quantitative reasoning.
Working checkpoint — typed tag: Calculate spans and a match ratio by hand. 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
Streaming text, typed tags, indexes and stable sorting show how simple schemas support large workflows.
Working checkpoint — PAF: Test malformed columns and unknown tags. 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
Sequence mapping links biology, algorithms, statistics and data engineering and can illuminate STEM pathways.
Working checkpoint — pairwise mapping: Check current qualifications separately. 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?
PAF can report an approximate mapping without a CIGAR string; base-level alignment detail is optional.
Working checkpoint — query interval: Ask whether the intended claim needs that missing resolution. 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 mapping note
Name references, minimap2 version, preset, options, index mode, tags, filters and validation.
Working checkpoint — target interval: Keep it beside every derived plot or table. 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
Check fields, names, lengths, bounds, strand, counts, MAPQ, tag types, alternatives and fixtures.
Working checkpoint — mapping quality: Stop at the first unexplained record. 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 FASTA hashes, index settings, command, PAF hash, filters, summaries and bounded conclusions.
Working checkpoint — typed tag: Make every mapping traceable to its query and target. 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