eduKateSG · Why Science?
Search widely, verify at the source
Learn how OmicsDI helps students discover omics datasets across repositories while keeping source accessions, metadata differences and reuse limits visible.
Modern biology stores genomics, transcriptomics, proteomics and metabolomics data in specialist repositories. OmicsDI provides a cross-repository discovery layer, helping a learner search across those collections without pretending that every archive uses identical metadata or access rules.
The key reading habit is simple: use the index to find, then use the provider record to verify. An OmicsDI dataset page can surface a title, description, repository, primary and secondary accessions, files, molecules, publications or similar datasets. The source archive remains the place to inspect authoritative deposited detail.
Did you know? One biological project can appear through several accessions because different repositories own different evidence layers. That is useful when the relationships are explicit and confusing when identifiers are copied without labels. A good science note always says which identifier belongs to which object.
This guide treats search rankings and similarity suggestions as discovery aids. They can reveal promising neighbours, but they do not by themselves establish that two datasets are methodologically comparable or support the same biological conclusion.
Chapter index
Chapters 1–9: discovery objects and metadata
- 1. An index gathers routes, not ownership
- 2. Omics type frames the measurement
- 3. Primary and secondary accessions differ
- 4. Titles are clues, not full methods
- 5. Descriptions need source attribution
- 6. Repository labels explain access
- 7. Files require context before download
- 8. Publications connect data to narrative
- 9. Shared molecules support exploration
Chapters 10–17: source verification, similarity and reuse
A discovery index is most useful when you can see where its records came from. OmicsDI aggregates metadata from multiple omics resources so users can discover datasets through one interface. The dataset provider and repository labels show where the indexed information originated. For science learning, that separation prevents a convenient index from being mistaken for the final evidence source.
Try this
Evidence question: Which source repository supplied this record, and which accession should be used there? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Open one OmicsDI result and write a two-line route containing the OmicsDI page and the provider accession. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for an index gathers routes, not ownership, its identifier or version context, and one limit that must remain visible.
The interesting detail here is the hand-off between the index and the source repository. Records can represent genomics, transcriptomics, proteomics, metabolomics and other omics domains. These labels describe broad measurement classes, not identical protocols or evidence strength. It also shows why search, metadata harmonisation and biological interpretation are different tasks.
Try this
Evidence question: Which omics type is declared, and what was actually measured within that category? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Choose two records with the same omics label and list the method details that still need checking. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. If a field is missing, write “not shown” and follow the official documentation rather than inventing a value.
Evidence checkpoint
Record the evidence object for omics type frames the measurement, its identifier or version context, and one limit that must remain visible.
A search result looks simple, but its meaning depends on fields gathered from several archives. An indexed record may show a provider accession plus related or secondary identifiers. Each accession can point to a project, study or deposit in another system. The record becomes stronger when its source accession, omics type and linked publication remain visible.
Try this
Evidence question: Which accession is primary for this page, and what object does each secondary identifier name? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Build an accession table with columns for identifier, database, object type and verified URL. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Before comparing two records, check that their object types, methods, versions and units really align.
Evidence checkpoint
Record the evidence object for primary and secondary accessions differ, its identifier or version context, and one limit that must remain visible.
This is where data literacy becomes practical: one label can support discovery while another supports verification. Dataset titles support quick scanning but often omit sample design, platform, processing and restrictions. Abstracts and source metadata provide more context. A learner can therefore use the index as a map while treating the source repository as the owner of deposited detail.
Try this
Evidence question: Which important design detail is absent from the title? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Predict three details you need, then verify them at the source and mark any that remain unknown. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Finish by naming one supported statement and one stronger statement that still needs another evidence layer.
Evidence checkpoint
Record the evidence object for titles are clues, not full methods, its identifier or version context, and one limit that must remain visible.
A discovery index is most useful when you can see where its records came from. OmicsDI displays descriptions collected from provider records. Wording may summarise project aims or findings, yet the index display is not a substitute for the full protocol or article. For science learning, that separation prevents a convenient index from being mistaken for the final evidence source.
Try this
Evidence question: Is this sentence an aim, a method, a result or a provider-supplied interpretation? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Colour-code a description by sentence function and attach each claim to the next evidence source needed. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for descriptions need source attribution, its identifier or version context, and one limit that must remain visible.
The interesting detail here is the hand-off between the index and the source repository. The repository field identifies the archive responsible for the underlying record. Different repositories may expose files openly, require controlled access or specialise in particular data types. It also shows why search, metadata harmonisation and biological interpretation are different tasks.
Try this
Evidence question: What access route and licence apply at the provider repository? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Follow the provider link and record whether files, metadata and reuse terms are openly visible. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. If a field is missing, write “not shown” and follow the official documentation rather than inventing a value.
Evidence checkpoint
Record the evidence object for repository labels explain access, its identifier or version context, and one limit that must remain visible.
A search result looks simple, but its meaning depends on fields gathered from several archives. An indexed file list can help locate evidence, but filenames and formats need methods, checksums, sample mapping and version context. The record becomes stronger when its source accession, omics type and linked publication remain visible.
Try this
Evidence question: Which file is raw, processed, metadata or documentation? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Classify five filenames and verify your classification against the source record rather than relying on extensions alone. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Before comparing two records, check that their object types, methods, versions and units really align.
Evidence checkpoint
Record the evidence object for files require context before download, its identifier or version context, and one limit that must remain visible.
This is where data literacy becomes practical: one label can support discovery while another supports verification. Linked papers can explain rationale, methods and interpretation. A publication identifier helps connect narrative claims to deposited data, while the archive record preserves the reusable objects. A learner can therefore use the index as a map while treating the source repository as the owner of deposited detail.
Try this
Evidence question: Does the paper cite the same accession, and which claims depend on the article rather than the dataset card? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Create paired notes for one dataset field and one paper conclusion. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Finish by naming one supported statement and one stronger statement that still needs another evidence layer.
Evidence checkpoint
Record the evidence object for publications connect data to narrative, its identifier or version context, and one limit that must remain visible.
A discovery index is most useful when you can see where its records came from. OmicsDI can surface molecules associated with indexed datasets. Such links aid discovery across studies but may arise from annotation, indexing or reported analysis. For science learning, that separation prevents a convenient index from being mistaken for the final evidence source.
Try this
Evidence question: What evidence connects the molecule to this dataset, and is the relationship measured or annotated? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Select one molecule link and trace it back to a file, field or provider explanation. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for shared molecules support exploration, its identifier or version context, and one limit that must remain visible.
The interesting detail here is the hand-off between the index and the source repository. Similarity features can suggest neighbouring records based on indexed metadata. Similarity does not guarantee compatible organisms, samples, platforms or statistical designs. It also shows why search, metadata harmonisation and biological interpretation are different tasks.
Try this
Evidence question: Which shared terms likely drove the suggestion, and which critical variables differ? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Compare one suggested pair using a five-field compatibility matrix. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. If a field is missing, write “not shown” and follow the official documentation rather than inventing a value.
Evidence checkpoint
Record the evidence object for similar datasets are suggestions, its identifier or version context, and one limit that must remain visible.
A search result looks simple, but its meaning depends on fields gathered from several archives. A cross-repository search depends on indexed text, identifiers, ontology mappings and filters. A narrow query may miss records that use different names. The record becomes stronger when its source accession, omics type and linked publication remain visible.
Try this
Evidence question: Which synonyms, identifiers or ontology terms could expand the search? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Run a plain-language query and an identifier-based query, then explain their different result sets. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Before comparing two records, check that their object types, methods, versions and units really align.
Evidence checkpoint
Record the evidence object for search terms shape recall, its identifier or version context, and one limit that must remain visible.
This is where data literacy becomes practical: one label can support discovery while another supports verification. Filters for repository, omics type, organism, date or publication status can make results manageable. Every facet also excludes records outside the selected value. A learner can therefore use the index as a map while treating the source repository as the owner of deposited detail.
Try this
Evidence question: Which filter most changes the scientific meaning of the set? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Save the unfiltered and filtered query states and describe what each can support. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Finish by naming one supported statement and one stronger statement that still needs another evidence layer.
Evidence checkpoint
Record the evidence object for facets change the visible universe, its identifier or version context, and one limit that must remain visible.
A discovery index is most useful when you can see where its records came from. Cross-repository systems map fields into a shared view, but source schemas and curation practices differ. Empty or broad fields may reflect that translation. For science learning, that separation prevents a convenient index from being mistaken for the final evidence source.
Try this
Evidence question: Is a missing value absent at source, not indexed, or not mapped into the common view? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Check the same field in OmicsDI and the provider record before labelling it missing. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for metadata harmonisation has limits, its identifier or version context, and one limit that must remain visible.
The interesting detail here is the hand-off between the index and the source repository. The OmicsDI REST API provides programmatic access to much of the discovery index and returns JSON for integration and analysis. It also shows why search, metadata harmonisation and biological interpretation are different tasks.
Try this
Evidence question: Which endpoint, query and pagination settings produced the retrieved set? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Record an API URL and describe the object type and result window before using the response. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. If a field is missing, write “not shown” and follow the official documentation rather than inventing a value.
Evidence checkpoint
Record the evidence object for the api makes searches repeatable, its identifier or version context, and one limit that must remain visible.
A search result looks simple, but its meaning depends on fields gathered from several archives. Result totals describe indexed records matching a query at a particular time. They are not participant counts, experiment counts or measures of scientific importance. The record becomes stronger when its source accession, omics type and linked publication remain visible.
Try this
Evidence question: What exactly is being counted, and could duplicate project layers appear? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Rewrite a misleading “there are N studies” statement as a precise index-result statement. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Before comparing two records, check that their object types, methods, versions and units really align.
Evidence checkpoint
Record the evidence object for counts are discovery counts, its identifier or version context, and one limit that must remain visible.
This is where data literacy becomes practical: one label can support discovery while another supports verification. Before reanalysis, verify licence, access status, file version, sample mapping, methods and documentation at the source repository. A learner can therefore use the index as a map while treating the source repository as the owner of deposited detail.
Try this
Evidence question: Which conditions permit reuse, and what transformations are needed? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Create a reuse checklist for one record and stop before analysis if a required item is unresolved. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Finish by naming one supported statement and one stronger statement that still needs another evidence layer.
Evidence checkpoint
Record the evidence object for reuse starts at the provider, its identifier or version context, and one limit that must remain visible.
A discovery index is most useful when you can see where its records came from. Bioinformaticians, data curators, librarians and research analysts all build routes across repositories. Their shared skill is keeping identifiers and provenance attached to evidence. For science learning, that separation prevents a convenient index from being mistaken for the final evidence source.
Try this
Evidence question: Which hand-off could fail if a provider accession were lost? Answer in three parts: identify the visible field or file, explain how it bears on the question, and state the boundary of that evidence. Preserve the database’s own object names and accession rather than flattening everything into the word ‘data’.
Try this: Write a short analyst handover that another person can reproduce without the title search. Turn the result into a compact claim-evidence-reasoning note, then ask a partner to retrace it. Keep the identifier, retrieval date and source page in the note so the route can be repeated.
Evidence checkpoint
Record the evidence object for discovery skills connect science careers, its identifier or version context, and one limit that must remain visible.
Contents · Previous chapter · Continue at the Science Learning Hub
A broad search becomes useful when the source route stays intact
OmicsDI can shorten the journey from a science question to candidate datasets, especially when evidence is spread across specialist archives. Its value comes from connection and discovery rather than replacing those archives.
Keep the provider repository, accession, query, filters and access date in every note. Then inspect methods, samples, files and reuse conditions at the source before making a comparison.
| Reading object | What it can support | Question to keep visible |
|---|---|---|
| Index record | Cross-repository discovery view | Which provider owns the deposited record? |
| Provider accession | Stable route into the source archive | What object type does it identify? |
| Similarity link | A candidate neighbouring dataset | Which design variables still differ? |
| API query | Repeatable discovery request | Which filters and page window were used? |
Sources and next reading
Substantive resource claims in this article were checked against the primary pages below. A database record is a structured route into evidence; it does not by itself prove a new biological conclusion. Recheck the live record, evidence labels and release context before using it in assessed work or research.
Primary sources
- OmicsDI home and repository list — accessed 24 October 2026.
- OmicsDI API overview — accessed 24 October 2026.
- OmicsDI ENA-project example record — accessed 24 October 2026.
- OmicsDI ArrayExpress example record — accessed 24 October 2026.
- OmicsDI open-source project — accessed 24 October 2026.