eduKateSG · Why Science?
Connect sequences to specimens, laboratories and case investigations—without turning similarity into a transmission story
Start with the outbreak question, then follow specimen origin, laboratory process, genomic result, epidemiological context and uncertainty before linking cases.
Science becomes more useful when a genome sequence remains connected to the specimen, laboratory process, food or environment, time and case investigation that gave it meaning. The OBO Foundry says the Genomic Epidemiology Ontology, or GenEpiO, covers vocabulary needed to identify, document and research foodborne pathogens and associated outbreaks. Its official page names genomic laboratory testing, specimen and isolate metadata, and epidemiological case investigations as important subdomains. The official design describes GenEpiO as an application ontology intended for databases and software interfaces, with alignment to Basic Formal Ontology and the Ontology for Biomedical Investigations. GenEpiO can support data sharing, but sequence similarity alone does not prove transmission, outbreak membership or source. This guide supports Secondary Science, O-Level Science, O-Level Biology, computing, mathematics and STEM education; it is educational, not medical, food-safety, outbreak-management, admission or career advice.
Related eduKate reading: NCBI Taxonomy; FoodOn; Environment Ontology; Statistics Ontology; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 13 October 2026: OBO Foundry Genomic Epidemiology Ontology overview, checked 13 October 2026; Official GenEpiO source repository; Official GenEpiO design documentation; Canonical GenEpiO OWL release; 2026 Singapore–Cambridge O-Level Biology syllabus; MOE G2/G3 Lower Secondary Science teaching and learning syllabus; 2026 MOE G2 Computing syllabus. Ontologies, definitions and release products can change, so preserve dated documentation, exact source records and file versions.
How genomic evidence becomes an outbreak investigation record
A genome does not arrive without context. It is generated from a particular material sample through a chain of collection, culture or direct extraction, library preparation, sequencing and analysis.
The official OBO page focuses GenEpiO on foodborne pathogens and outbreak investigations. That scope makes specimen, food, laboratory and epidemiological metadata as important as the sequence itself.
The project repository describes genomics, laboratory, clinical and epidemiological context required for data sharing and integration. Each domain contributes a different piece of the evidence chain.
GenEpiO calls itself an application ontology because it is intended to provide terms for databases and user interfaces, not merely a taxonomic list. Usability and consistent data-entry fields are part of its design problem.
The official design aligns its hierarchy with BFO and OBI and discusses measurements, observables and information content. Imports help place terms in a broader semantic ecosystem.
Stable taxon identifiers support organism naming, while isolate and specimen identities preserve the material path. A taxon match does not make two isolates the same.
FoodOn can structure food sources and ingredients, and ENVO can structure environmental context. Those connections help distinguish a product, place or material from the pathogen record.
Genomic similarity depends on organism, method, quality, pipeline and threshold. A distance considered informative for one pathogen or time scale may be inappropriate for another.
Epidemiological linkage needs exposures, time, place and alternative explanations. Genomic closeness can strengthen or weaken a hypothesis without independently proving direction or source.
The optimistic value is connected evidence: investigators can exchange interoperable records while every displayed cluster remains traceable to samples, methods and uncertainty.
Begin with the investigation purpose. Routine surveillance, source tracing, contamination monitoring and research studies may collect similar fields for different decisions.
Assign persistent identifiers to specimen, isolate, extraction, sequencing run and assembly. One specimen can yield several laboratory derivatives that must not be counted as independent cases.
Record collection time as an interval when necessary. Manufacturing, sampling, symptom onset and sequencing dates are different events and should not share one vague date field.
Preserve geographic precision and restrictions. Exact farm, facility or household locations may be sensitive and should be governed rather than silently rounded.
Represent laboratory methods and quality thresholds. Coverage, contamination, mixed samples and missing regions can alter distance estimates and lineage calls.
Version the analysis pipeline, reference genome, database and parameters. Identical reads can produce different outputs under different software choices.
Store epidemiological evidence independently from the genomic cluster. Cases may share exposure without close sequences, or share sequences without a discovered common exposure.
Keep inclusion and exclusion decisions. Removing low-quality samples or redefining a cluster changes the apparent outbreak and must be reproducible.
For classroom learning, invented sequences and fictional food samples let students build an evidence table without implying a real outbreak.
Reader-ready outbreak evidence states what is linked, how strongly, by which methods and which alternatives remain plausible.
Whole-genome sequencing can discriminate closely related isolates, but resolution is not unlimited. Mutation rates, recombination, within-host variation and sequencing error influence how genomic distance should be interpreted.
A threshold is an operational decision, not a universal law. Fixed single-nucleotide-polymorphism cutoffs may fail across pathogens, pipelines or outbreak durations. Preserve the rule and scientific justification used at the time.
Sampling density changes cluster appearance. A sparse dataset can make distant cases look uniquely related, while intensive sampling may reveal intermediates or background diversity. Report who and what was eligible for sequencing.
Food supply chains create complex paths. A retail product can contain ingredients from several sources, be processed across facilities and be distributed widely. Product identity, lot and movement data need their own provenance.
Environmental persistence can complicate direction. A pathogen may remain in equipment or an environment across production cycles, so calendar order alone may not identify the source event.
Epidemiological questionnaires are subject to recall, selection and interviewer effects. Structured fields help comparison, but they do not turn remembered exposure into direct observation.
Data governance must consider people, farms, businesses and communities. Premature public linkage can cause harm. Use authorised workflows, minimum necessary disclosure and a clear distinction between hypothesis and confirmed finding.
A reproducible investigation bundle includes specimen manifests, chain-of-custody records, ontology releases, raw reads or governed access, pipeline containers, quality reports, epidemiological dictionaries, cluster rules and decision logs.
Statistical models can combine genomic and epidemiological information, yet their priors, missing-data handling and assumptions need review. A probability is not a substitute for explaining what evidence entered the model.
Science tuition can turn GenEpiO into a cross-subject exercise: Biology supplies pathogens, mathematics supplies distance and uncertainty, computing supplies pipelines, and English supplies calibrated evidence writing.
Students should learn that good surveillance is iterative. New samples can split a cluster, expand it or reveal a different explanation. Scientific knowledge remains revisable because evidence grows.
GenEpiO is most useful when it prevents the sequence from floating free of the real world. Every result should still point back to material, process, place, time and investigation context.
A GenEpiO investigation becomes most instructive when the sequence is not treated as the whole story. Start with three fictional isolates whose genomes are similar, then provide different specimen dates, collection sites, food histories and laboratory-quality notes. Ask learners to encode the specimen, isolate, assay, workflow, sequence-quality measures and epidemiological observations before drawing a network. They should discover that a small genetic distance can support relatedness while still leaving direction, source and transmission route unresolved. Conversely, incomplete sequencing or sparse sampling can make related cases look distant or invisible. A defensible report therefore names the analysis pipeline and reference, threshold rule, database and software versions, quality filters, sampling frame, missing records and non-genomic evidence. It distinguishes a laboratory cluster from an epidemiological outbreak and uses language such as consistent with, supports or does not exclude at the strength warranted by the evidence. For operational validation, rebuild a small benchmark containing known clusters, unrelated backgrounds, contamination scenarios and metadata conflicts. Run it whenever the ontology, reference database or pipeline changes, and have laboratory and epidemiology reviewers examine disagreements rather than letting a numeric threshold decide automatically. Preserve corrections and isolate merges as versioned events so published counts remain explainable. Governance matters because pathogen sequences, dates and fine-grained locations can expose patients, workers, farms or businesses even after names are removed. Release only the detail necessary for the public purpose and keep controlled linkage keys separately. This workflow shows why genomic epidemiology is a team science: ontologies help computers join the evidence, but microbiology, field investigation, statistics, data stewardship and public-health judgement decide what the connected evidence can responsibly mean. For students, that is a hopeful lesson—powerful tools become more useful, not less, when their limitations are explicit and every conclusion can be traced back to observations.
Before sharing a cluster visual, reviewers should also test the story without the most influential sample. If the conclusion changes sharply, say so and investigate sampling sensitivity. Show unlinked background isolates, not only the exciting cluster, and make missing collection dates visible. Students can then compare a sequence-only diagram with the full evidence table and identify which apparent links weaken. This simple counterfactual exercise teaches that a network is an analytical view built from choices, not a photograph of transmission. It also encourages precise questions about who was sampled, when, why and with which method before anyone treats graphical proximity as epidemiological proof.
| Stage | Question | Evidence | Responsible output |
|---|---|---|---|
| Collect | What specimen or isolate was obtained? | Source, time, location, method and chain of custody | A traceable material record |
| Generate | How was genomic data produced? | Culture, extraction, sequencing, quality and pipeline | A reproducible genomic result |
| Contextualise | Which food, environment or case information applies? | Exposure, taxonomy, geography and investigation data | A bounded epidemiological record |
| Link | What relationship is supported? | Genomic distance, sampling coverage and alternatives | A calibrated outbreak hypothesis |
Inside this guide
1–12 · Specimens, isolates and laboratory context
- 1. Begin with the investigation question
- 2. Use the GenEpiO identifier
- 3. Read the scope
- 4. Understand application ontology
- 5. Identify the specimen
- 6. Identify the isolate
- 7. Identify the taxon
- 8. Record collection
- 9. Record food context
- 10. Record environment
- 11. Record culture and extraction
- 12. Record sequencing
13–24 · Epidemiological linkage and validation
- 13. Record quality control
- 14. Record assembly and typing
- 15. Record genomic distance
- 16. Record case context
- 17. Record exposure history
- 18. Record outbreak definitions
- 19. Connect taxonomy
- 20. Connect FoodOn
- 21. Connect ENVO
- 22. Connect statistics
- 23. Challenge sequence-only clusters
- 24. Challenge universal cutoffs
25–36 · Learning, communication and pathways
- 25. Challenge duplicate sampling
- 26. Challenge ascertainment bias
- 27. Review ontology and pipeline migrations
- 28. Connect to Secondary Science
- 29. Connect to O-Level Biology
- 30. Connect to mathematics
- 31. Connect to computing
- 32. Connect to pathways
- 33. Did you know?
- 34. Write a reader-ready outbreak note
- 35. Build a validation ladder
- 36. Finish with a reproducible GenEpiO bundle
Section 1 of 36
1. Begin with the investigation question
Surveillance, source attribution and research comparisons require different evidence and metadata.
Working checkpoint — GenEpiO identifier: State the purpose and decision boundary. Keep specimen, isolate, sequence, pipeline, exposure context and uncertainty linked.
For mathematics, make distance thresholds and missing samples visible.
Section 2 of 36
2. Use the GenEpiO identifier
Stable terms connect database fields and software interfaces across projects.
Working checkpoint — specimen metadata: Store IRI, label, definition and release. Separate genomic similarity from epidemiological linkage and transmission direction.
For computing, preserve identifiers, pipeline versions and validation fixtures.
Section 3 of 36
3. Read the scope
The official scope centres foodborne pathogens, associated outbreaks and their laboratory and epidemiological context.
Working checkpoint — isolate record: Keep claims inside the represented domain. Pin taxonomies, ontologies, reference genomes and analysis versions.
For writing, call a cluster a hypothesis unless the wider evidence supports more.
Section 4 of 36
4. Understand application ontology
GenEpiO is designed for practical data systems as well as formal semantic alignment.
Working checkpoint — genomic result: Test both logical meaning and user-interface clarity. Preserve sampling frames, exclusions and cluster definitions.
For students, ask where the sequence came from and what else supports the link.
Section 5 of 36
5. Identify the specimen
Food, environmental, clinical and laboratory materials support different inferences.
Working checkpoint — case investigation: Assign a persistent specimen ID. Use qualified authorities for operational outbreak decisions.
For parents, the calm checkpoint is whether the learner can explain observation, comparison and inference separately.
Section 6 of 36
6. Identify the isolate
An isolate is a laboratory-derived biological record linked to its parent specimen.
Working checkpoint — outbreak provenance: Preserve parent–child provenance. Keep specimen, isolate, sequence, pipeline, exposure context and uncertainty linked.
For mathematics, make distance thresholds and missing samples visible.
Section 7 of 36
7. Identify the taxon
Species, serotype and lineage resolution should match the evidence.
Working checkpoint — GenEpiO identifier: Keep authoritative taxon and method. Separate genomic similarity from epidemiological linkage and transmission direction.
For computing, preserve identifiers, pipeline versions and validation fixtures.
Section 8 of 36
8. Record collection
Time, place, collector, method and container affect interpretation.
Working checkpoint — specimen metadata: Publish chain-of-custody fields. Pin taxonomies, ontologies, reference genomes and analysis versions.
For writing, call a cluster a hypothesis unless the wider evidence supports more.
Section 9 of 36
9. Record food context
Product, ingredient, lot, processing and supply-chain roles must remain distinct.
Working checkpoint — isolate record: Use FoodOn with local details. Preserve sampling frames, exclusions and cluster definitions.
For students, ask where the sequence came from and what else supports the link.
Section 10 of 36
10. Record environment
Facility, surface, water, soil and other environmental contexts shape exposure hypotheses.
Working checkpoint — genomic result: Use ENVO and measured conditions. Use qualified authorities for operational outbreak decisions.
For parents, the calm checkpoint is whether the learner can explain observation, comparison and inference separately.
Section 11 of 36
11. Record culture and extraction
Laboratory enrichment or culture can select organisms and alter representation.
Working checkpoint — case investigation: Name protocol and derivative identifiers. Keep specimen, isolate, sequence, pipeline, exposure context and uncertainty linked.
For mathematics, make distance thresholds and missing samples visible.
Section 12 of 36
12. Record sequencing
Platform, library preparation, read structure and run metadata influence quality.
Working checkpoint — outbreak provenance: Pin method and run identity. Separate genomic similarity from epidemiological linkage and transmission direction.
For computing, preserve identifiers, pipeline versions and validation fixtures.
Section 13 of 36
13. Record quality control
Coverage, contamination, mixed signals and missingness determine what analysis is defensible.
Working checkpoint — GenEpiO identifier: Retain thresholds and failures. Pin taxonomies, ontologies, reference genomes and analysis versions.
For writing, call a cluster a hypothesis unless the wider evidence supports more.
Section 14 of 36
14. Record assembly and typing
Assemblies, alleles, lineages and resistance predictions are generated outputs, not raw observations.
Working checkpoint — specimen metadata: Store pipeline and database versions. Preserve sampling frames, exclusions and cluster definitions.
For students, ask where the sequence came from and what else supports the link.
Section 15 of 36
15. Record genomic distance
Distance needs the compared representation, masking rules and calculation method.
Working checkpoint — isolate record: Do not publish a bare number. Use qualified authorities for operational outbreak decisions.
For parents, the calm checkpoint is whether the learner can explain observation, comparison and inference separately.
Section 16 of 36
16. Record case context
Symptoms, onset and demographics may be relevant while requiring privacy protection.
Working checkpoint — genomic result: Use governed minimum data. Keep specimen, isolate, sequence, pipeline, exposure context and uncertainty linked.
For mathematics, make distance thresholds and missing samples visible.
Section 17 of 36
17. Record exposure history
Food and place histories are reported evidence with recall limits.
Working checkpoint — case investigation: Preserve questionnaire and uncertainty. Separate genomic similarity from epidemiological linkage and transmission direction.
For computing, preserve identifiers, pipeline versions and validation fixtures.
Section 18 of 36
18. Record outbreak definitions
Cluster or case definitions specify inclusion rules for a purpose and time.
Working checkpoint — outbreak provenance: Version the definition. Pin taxonomies, ontologies, reference genomes and analysis versions.
For writing, call a cluster a hypothesis unless the wider evidence supports more.
Section 19 of 36
19. Connect taxonomy
NCBI Taxonomy stabilises organism identity across laboratory and sequence records.
Working checkpoint — GenEpiO identifier: Keep taxon-release context. Preserve sampling frames, exclusions and cluster definitions.
For students, ask where the sequence came from and what else supports the link.
Section 20 of 36
20. Connect FoodOn
FoodOn can distinguish foods, ingredients and processes involved in source hypotheses.
Working checkpoint — specimen metadata: Preserve brand and lot where authorised. Use qualified authorities for operational outbreak decisions.
For parents, the calm checkpoint is whether the learner can explain observation, comparison and inference separately.
Section 21 of 36
21. Connect ENVO
Environmental ontology terms provide comparable material and setting context.
Working checkpoint — isolate record: Keep coordinates and measurements separate. Keep specimen, isolate, sequence, pipeline, exposure context and uncertainty linked.
For mathematics, make distance thresholds and missing samples visible.
Section 22 of 36
22. Connect statistics
STATO-style provenance can describe models, tests and estimates used to assess clusters.
Working checkpoint — genomic result: Link every output to inputs and assumptions. Separate genomic similarity from epidemiological linkage and transmission direction.
For computing, preserve identifiers, pipeline versions and validation fixtures.
Section 23 of 36
23. Challenge sequence-only clusters
Close genomes can occur without a known direct epidemiological connection.
Working checkpoint — case investigation: Seek time, place and exposure evidence. Pin taxonomies, ontologies, reference genomes and analysis versions.
For writing, call a cluster a hypothesis unless the wider evidence supports more.
Section 24 of 36
24. Challenge universal cutoffs
A distance threshold may be pathogen-, method- and timescale-specific.
Working checkpoint — outbreak provenance: Document validation and exceptions. Preserve sampling frames, exclusions and cluster definitions.
For students, ask where the sequence came from and what else supports the link.
Section 25 of 36
25. Challenge duplicate sampling
Several derivatives from one specimen or facility can inflate apparent evidence.
Working checkpoint — GenEpiO identifier: Model sampling hierarchy. Use qualified authorities for operational outbreak decisions.
For parents, the calm checkpoint is whether the learner can explain observation, comparison and inference separately.
Section 26 of 36
26. Challenge ascertainment bias
Only a subset of cases, foods or facilities may be sampled and sequenced.
Working checkpoint — specimen metadata: Report the sampling frame. Keep specimen, isolate, sequence, pipeline, exposure context and uncertainty linked.
For mathematics, make distance thresholds and missing samples visible.
Section 27 of 36
27. Review ontology and pipeline migrations
New classifications or software versions can move records between clusters.
Working checkpoint — isolate record: Run regression tests before comparison. Separate genomic similarity from epidemiological linkage and transmission direction.
For computing, preserve identifiers, pipeline versions and validation fixtures.
Section 28 of 36
28. Connect to Secondary Science
Microorganisms, variation and experimental design become one connected evidence chain.
Working checkpoint — genomic result: Follow a fictional sample from collection to conclusion. Pin taxonomies, ontologies, reference genomes and analysis versions.
For writing, call a cluster a hypothesis unless the wider evidence supports more.
Section 29 of 36
29. Connect to O-Level Biology
Infectious disease and practices of science require analysis, evaluation and responsible action.
Working checkpoint — case investigation: Defend one inference with multiple evidence types. Preserve sampling frames, exclusions and cluster definitions.
For students, ask where the sequence came from and what else supports the link.
Section 30 of 36
30. Connect to mathematics
Distance, probability, sampling and uncertainty show why a threshold needs context.
Working checkpoint — outbreak provenance: Compare sensitivity under alternate rules. Use qualified authorities for operational outbreak decisions.
For parents, the calm checkpoint is whether the learner can explain observation, comparison and inference separately.
Section 31 of 36
31. Connect to computing
Databases, ontologies, pipelines and validation turn outbreak evidence into reproducible infrastructure.
Working checkpoint — GenEpiO identifier: Test identifiers and version changes. Keep specimen, isolate, sequence, pipeline, exposure context and uncertainty linked.
For mathematics, make distance thresholds and missing samples visible.
Section 32 of 36
32. Connect to pathways
GenEpiO skills matter in genomics, microbiology, epidemiology, food safety, bioinformatics and data stewardship.
Working checkpoint — specimen metadata: Check current official routes separately. Separate genomic similarity from epidemiological linkage and transmission direction.
For computing, preserve identifiers, pipeline versions and validation fixtures.
Section 33 of 36
33. Did you know?
The official design discusses interface-friendly labels because ontology terms must also work for people entering and reviewing data.
Working checkpoint — isolate record: Compare a formal label with a plain-language UI label. Pin taxonomies, ontologies, reference genomes and analysis versions.
For writing, call a cluster a hypothesis unless the wider evidence supports more.
Section 34 of 36
34. Write a reader-ready outbreak note
Name specimen, isolate, taxon, method, pipeline, distance, exposure context, definition and uncertainty.
Working checkpoint — genomic result: Keep linkage language calibrated. Preserve sampling frames, exclusions and cluster definitions.
For students, ask where the sequence came from and what else supports the link.
Section 35 of 36
35. Build a validation ladder
Check identity, custody, taxon, method, quality, pipeline, epidemiology, sampling and interpretation.
Working checkpoint — case investigation: Stop at the earliest unresolved rung. Use qualified authorities for operational outbreak decisions.
For parents, the calm checkpoint is whether the learner can explain observation, comparison and inference separately.
Section 36 of 36
36. Finish with a reproducible GenEpiO bundle
Deliver manifests, ontology files, sequences, pipelines, dictionaries, cluster rules, tests and review notes.
Working checkpoint — outbreak provenance: Make every link rebuildable. Keep specimen, isolate, sequence, pipeline, exposure context and uncertainty linked.
For mathematics, make distance thresholds and missing samples visible.
Contents · Previous section · Continue to the Science Learning Hub