eduKateSG · Why Science?
Profile familiar and complex splicing events across samples—while keeping library version, coverage and comparability 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. vast-tools is a command-line toolset for profiling and comparing alternative-splicing events in RNA-seq data. It uses curated species libraries and event definitions, processes samples separately, combines their event-level measurements and supports comparison of exon skipping, alternative splice-site choices and intron retention. Percent-spliced-in values and coverage grades are valuable only when the matching library, event denominator and sample design are recorded. 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: Whippet and lightweight splicing quantification; SUPPA2 and transcript-event uncertainty; rMATS and replicate-aware splicing differences; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 11 October 2026: Official vast-tools repository and documentation; Official vast-tools release notes; VAST-TOOLS and VastDB framework protocol in PubMed; VastDB atlas primary study full text; 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 |
|---|---|---|---|
| Library | Which events and references define the search space? | A versioned VASTDB species library | A reproducible event catalogue |
| Align | Which reads support expression and splice alternatives? | Raw RNA-seq reads and tool settings | Per-sample mapping and QC evidence |
| Combine | Which events are quantifiable across samples? | Per-sample outputs and coverage rules | A shared PSI and coverage matrix |
| Compare | Which event profiles differ by group? | Independent replicates and declared contrasts | Differential-splicing candidates for validation |
Inside this guide
1–12 · Foundations and design
- 1. Start with the species library
- 2. Understand the event-centred view
- 3. Prepare raw read inputs
- 4. Group technical read files correctly
- 5. Use the align command reproducibly
- 6. Read mapping quality summaries
- 7. Quantify exon-skipping events
- 8. Quantify alternative donor sites
- 9. Quantify alternative acceptor sites
- 10. Quantify intron retention
- 11. Interpret PSI as a conditional fraction
- 12. Keep unquantified values distinct
13–24 · Models and estimation
- 13. Use coverage grades
- 14. Combine samples after QC
- 15. Check event identity across releases
- 16. Build the biological design separately
- 17. Use compare with clear groups
- 18. Pair delta PSI with support
- 19. Respect replicate heterogeneity
- 20. Correct for the searched family
- 21. Inspect the generated command log
- 22. Use VastDB for context, not proof
- 23. Treat functional annotations as predictions
- 24. Test coverage thresholds
25–36 · Diagnostics and interpretation
- 25. Test sample influence
- 26. Test library-version sensitivity
- 27. Compare with Whippet fairly
- 28. Compare with SUPPA2 fairly
- 29. Connect to library catalogues
- 30. Connect to percentages
- 31. Connect to computing practice
- 32. Connect to study pathways
- 33. Use bounded language
- 34. Ask the parent question
- 35. Did you know?
- 36. Finish with a reproducible bundle
Section 1 of 36
1. Start with the species library
vast-tools relies on species-specific VASTDB libraries containing reference sequences, event annotations and supporting metadata. The library version is part of the measurement because it defines the events that can be quantified.
Working checkpoint — event library: Record species code, genome assembly and library release. 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. Understand the event-centred view
The workflow profiles defined alternative-splicing events rather than reconstructing every full-length transcript. Event-centric evidence is often easier to compare, but distant exon combinations remain unresolved.
Working checkpoint — percent spliced in: State the event unit before discussing isoforms. 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. Prepare raw read inputs
The standard workflow processes RNA-seq reads sample by sample. Read length, pairing, strandedness and technical grouping affect how evidence reaches transcript and junction references.
Working checkpoint — coverage grade: Keep original accessions and a manifest of read files. 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. Group technical read files correctly
Files derived from the same original sequenced molecule should not be counted as independent evidence. The official documentation describes read groups so subreads are handled without multiplying one molecule.
Working checkpoint — exon skipping: Document technical grouping before alignment. 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 the align command reproducibly
The align stage maps reads against the species library and produces expression, splicing and quality outputs. Thread count and performance options are practical settings; reference and mapping choices remain scientific settings.
Working checkpoint — alternative splice site: Archive the exact command and generated log. 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 mapping quality summaries
A sample with low usable mapping or strong bias has fewer opportunities to quantify events. Quality differences that align with biological groups can distort later comparisons.
Working checkpoint — intron retention: Review QC before combining samples. 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. Quantify exon-skipping events
Exon-skipping events use junction evidence for inclusion and exclusion paths. Complex cassette-exon arrangements may receive classes that reflect increasing structural complexity in the event library.
Working checkpoint — event library: Show all component junctions for a headline event. 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. Quantify alternative donor sites
Alternative 5-prime splice-site events compare donor choices. Strand orientation matters, and closely spaced sites can be difficult to distinguish with short or error-prone reads.
Working checkpoint — percent spliced in: Verify coordinates and overhang support. 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. Quantify alternative acceptor sites
Alternative 3-prime splice-site events compare acceptor choices under the same event framework. The familiar label does not remove alignment or mappability limitations.
Working checkpoint — coverage grade: Inspect both splice boundaries in context. 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. Quantify intron retention
Intron-retention estimates combine exon–intron and spliced evidence under library-specific rules. Nascent RNA, DNA contamination and uneven coverage can imitate retention.
Working checkpoint — exon skipping: Pair the number with library preparation and coverage review. 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. Interpret PSI as a conditional fraction
Percent Spliced In summarises inclusion relative to the alternatives defined for an event. PSI can change because inclusion rises, exclusion falls or both happen.
Working checkpoint — alternative splice site: Report component evidence beside PSI. 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. Keep unquantified values distinct
An event may be absent from a sample output because coverage is insufficient rather than because its inclusion is zero. Converting missingness to zero invents biological exclusion.
Working checkpoint — intron retention: Preserve missing values and coverage reason. 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. Use coverage grades
vast-tools reports qualitative coverage categories designed to communicate measurement support. A coverage grade is not a biological effect and should travel with every PSI value.
Working checkpoint — event library: Filter or stratify using a predeclared grade rule. 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. Combine samples after QC
The combine stage gathers per-sample event measurements into shared tables. A convenient matrix should not conceal that each cell may carry different support or missingness.
Working checkpoint — percent spliced in: Retain coverage columns with inclusion levels. 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. Check event identity across releases
Event identifiers and annotations can change when a species library is updated. Combining outputs from different releases risks aligning labels that do not represent identical structures.
Working checkpoint — coverage grade: Use one library release or create a documented crosswalk. 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. Build the biological design separately
vast-tools can organise measurements, but group inference still depends on independent samples, balanced batches and an explicit contrast. Sample count should refer to biological units, not FASTQ files.
Working checkpoint — exon skipping: Publish the sample-to-group table. 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 compare with clear groups
The compare workflow evaluates event differences across specified samples or groups. File lists and ordering are part of the analysis and must be checked before the command runs.
Working checkpoint — alternative splice site: Print resolved sample names and group labels. 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. Pair delta PSI with support
A large PSI difference can be unstable when one group has weak coverage; a small difference can be precise with many consistent replicates. Magnitude and evidence answer different questions.
Working checkpoint — intron retention: Show delta PSI, coverage and sample-level values together. 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 replicate heterogeneity
Group averages can hide bimodal or outlier samples. A credible pattern should be examined across each independent replicate rather than accepted from one summary row.
Working checkpoint — event library: Plot every sample for priority events. 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. Correct for the searched family
Many event comparisons create many opportunities for false positives. The tested family depends on coverage filters, event types and contrasts.
Working checkpoint — percent spliced in: Preserve full outputs and adjustment settings. 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. Inspect the generated command log
Current vast-tools releases create a log recording version, date and main options. That provenance should be archived with results rather than discarded after a successful run.
Working checkpoint — coverage grade: Include the log in the reproducibility bundle. 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. Use VastDB for context, not proof
VastDB can show event profiles, sequence features and functional annotations across many tissues or developmental stages. Context can motivate a hypothesis but does not replace evidence in the user’s own samples.
Working checkpoint — exon skipping: Separate database context from experiment-specific results. 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. Treat functional annotations as predictions
Protein domains, disorder regions, conservation and suggested primers can help prioritise events. These features do not demonstrate that an isoform is translated or functional in the current system.
Working checkpoint — alternative splice site: Use predicted language until validated. 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. Test coverage thresholds
Different defensible coverage rules can change the shared event set and apparent group pattern. Sensitivity analysis should expose, not optimise, that dependence.
Working checkpoint — intron retention: Report event retention across planned rules. 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. Test sample influence
Removing one independent sample can reveal whether a comparison is supported broadly or driven by a single library. Technical subreads should never be removed as though they were independent subjects.
Working checkpoint — event library: Repeat key comparisons one biological unit at a time. 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. Test library-version sensitivity
A later VASTDB library may revise events or annotations. For a major conclusion, compare coordinates and evidence across a justified update before treating the label as permanent.
Working checkpoint — percent spliced in: Track structural identity, not only the event ID. 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. Compare with Whippet fairly
Whippet uses a compact splicing-graph index and entropy-aware quantification, while vast-tools uses curated species libraries and event profiles. Similar PSI labels can arise from different reference structures.
Working checkpoint — coverage grade: Align event paths and denominators first. 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. Compare with SUPPA2 fairly
SUPPA2 derives events from transcript annotations and propagates expression uncertainty, whereas vast-tools directly profiles its curated event library from reads. Their missingness and uncertainty summaries differ.
Working checkpoint — exon skipping: Compare raw support and event definitions. 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 library catalogues
A species library resembles a carefully indexed reference shelf: it makes searching efficient, but what can be found depends on what the shelf contains and how it is labelled.
Working checkpoint — alternative splice site: Ask which biology lies outside the catalogue. 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 percentages
PSI is a percentage with a defined denominator. The same inclusion count can produce different PSI values when the competing evidence changes.
Working checkpoint — intron retention: Use coloured counters to make the denominator visible. 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 practice
Command logs, species libraries, file manifests and checksums turn a command-line workflow into an auditable experiment. Reproducibility is part of the scientific result.
Working checkpoint — event library: Make version capture automatic. 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 study pathways
The workflow brings together molecular biology, statistics and software engineering and can illuminate routes into bioinformatics, genomics and biotechnology.
Working checkpoint — percent spliced in: Avoid presenting tool familiarity as a guaranteed career outcome. 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. Use bounded language
A vast-tools comparison supports an event-level inclusion difference under a specific library, coverage rule and design. It does not alone prove full-length isoforms, protein change or clinical importance.
Working checkpoint — coverage grade: Match the verb to the evidence. 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. Ask the parent question
A useful question is: were the same events measured well in each independent sample, and did the analysis keep low coverage visible?
Working checkpoint — exon skipping: Good measurement discipline matters more than a dramatic heat map. 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. Did you know?
VAST expands to Vertebrate Alternative Splicing and Transcription, yet current libraries and workflows extend the framework beyond a single organismal story. The memorable name never replaces the exact species library.
Working checkpoint — alternative splice site: Put the library release in every methods 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 raw-read accessions, sample and read-group manifests, species library, vast-tools version, commands, logs, QC, combined tables, coverage grades, contrasts, full results and validation plans.
Working checkpoint — intron retention: A profile is trustworthy when another analyst can rebuild every cell in 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.
Contents · Previous section · Continue to the Science Learning Hub