eduKateSG · Why Science?
Turn a large genomic text table into a region-searchable resource—while keeping its columns, coordinate convention, sort order and compression contract 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. Tabix is an HTSlib tool and index model for fast regional retrieval from suitable genomic text files that have been position sorted and compressed with bgzip. The TBI specification stores the preset or generic-format configuration, sequence, beginning and optional end columns, comment character, skipped header lines, reference names, bins, chunks and a linear index. The current tabix manual also supports CSI output for coordinates beyond the TBI limit. A successful query depends on the declared columns and coordinate convention matching the file; tabix does not understand an arbitrary spreadsheet or judge the biological quality of its rows. 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: VCF and structured variant call records; BED and genomic feature coordinates; BCF and binary variant call storage; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.
Primary and official sources checked on 12 October 2026: Official Tabix index file-format specification; Official HTSlib tabix manual, current documentation viewed 12 October 2026; Official HTSlib bgzip manual, current documentation viewed 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 the dated documentation, exact source bytes, parser rules and software versions used.
A deeper reading route
Tabix sits between readable text and scalable access. Researchers often prefer VCF, BED, GFF or custom tab-separated records because rows can be inspected directly. Sequentially scanning gigabytes for every region is inefficient. bgzip supplies independently addressable compressed blocks, and tabix supplies a coordinate map from genomic intervals to candidate blocks.
The index must know how to interpret each line. Presets configure familiar formats, while generic mode records which column carries the sequence name, which carries the beginning coordinate and which optionally carries the end. It also records whether positions are zero based, which character starts metadata lines and how many initial lines to skip. These are scientific semantics, not cosmetic parser switches.
TBI combines hierarchical bins with a linear index, echoing BAI. Reference names are stored in index order, and chunks use BGZF virtual offsets. A query identifies its reference, enumerates possible bins, seeks to candidate chunks, decodes lines and tests exact overlap. The index accelerates the route; the text row remains the evidence.
The TBI coordinate range follows the BAI-style hierarchy and therefore cannot address every possible long reference. The tabix manual documents CSI creation as the alternative and notes a higher default limit with configurable minimum interval size. Workflows should decide from contig lengths and consumer support before indexing, not after a failed query.
Text creates special hazards. Tabs can be replaced by spaces, comment lines can appear unexpectedly, coordinate values can be malformed, and locale-aware sorting can order chromosome labels or numbers differently from genomic expectations. Recompressing a corrected file changes virtual offsets. A robust workflow validates text, performs the intended sort, compresses, indexes and then compares queries with a full scan.
Remote querying adds HTTP range semantics, authentication and object stability. A client may cache an index while the text object changes. Matching basenames or URLs do not guarantee matching versions. Immutable object identifiers, hashes, sizes and clear cache invalidation protect both performance and scientific correctness.
A useful production test includes deliberately awkward rows: a one-base interval at the first coordinate, an interval ending exactly at a bin boundary, a long feature crossing several bins, an empty reference region, repeated positions and a final record at the end of a contig. These fixtures reveal off-by-one, sort and stale-index failures that ordinary mid-chromosome examples can miss.
For students, tabix feels like adding a map to a very long notebook. The notebook stays readable, but the map lets us open near one labelled location. The important scientific habit is to write the map’s rules: which column names the place, where an interval starts and ends, and which coordinate system the notebook uses.
| Stage | Question | Evidence | Responsible output |
|---|---|---|---|
| Define | How does each line express a genomic interval? | Preset or explicit sequence, begin and end columns | A parser contract |
| Prepare | Is the text ordered and block-compressed correctly? | Validated sort plus bgzip stream | An indexable source |
| Query | Which chunks may contain the requested region? | TBI or CSI bins, linear bounds and virtual offsets | Candidate lines |
| Verify | Do indexed results equal a full format-aware scan? | Hashes, boundary fixtures and record comparisons | Trustworthy regional text access |
Inside this guide
1–12 · Foundations and design
- 1. Start with the text schema
- 2. Use current official documentation
- 3. Distinguish tool from index
- 4. Require bgzip compression
- 5. Require position sorting
- 6. Use a preset when it truly matches
- 7. Configure the sequence column
- 8. Configure the begin column
- 9. Configure the end column
- 10. Declare zero-based input explicitly
- 11. Declare the meta character
- 12. Declare skipped lines
13–24 · Models and estimation
- 13. Preserve reference-name order
- 14. Understand TBI format codes
- 15. Read bins and chunks
- 16. Use the linear index
- 17. Know the TBI coordinate limit
- 18. Choose CSI deliberately
- 19. Query with matching reference names
- 20. Use half-open query reasoning
- 21. Keep headers with query output
- 22. Do not equate no result with no biology
- 23. Protect index freshness
- 24. Validate malformed lines
25–36 · Diagnostics and interpretation
- 25. Use stable sorting rules
- 26. Compare with a full scan
- 27. Plan remote URLs carefully
- 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 tabix note
- 35. Build a validation ladder
- 36. Finish with a reproducible bundle
Section 1 of 36
1. Start with the text schema
Tabix needs a line-oriented genomic format with identifiable sequence and coordinate fields. Begin by naming the format and one example row.
Working checkpoint — tabix: Document delimiter, columns, coordinate convention and metadata lines. Keep raw text, bgzip stream and index hashes, schema, preset or columns, coordinate convention, sort and tool versions together.
For a classroom model, index a fictional habitat table with no personal data.
Section 2 of 36
2. Use current official documentation
The TBI quick reference defines the index layout, while the HTSlib tabix manual consulted here is dated 9 July 2026.
Working checkpoint — TBI index: Pin the tool release and index format. Read every empty or positive result beside header definitions, reference-name identity and a full-scan comparison.
For future study, connect tabular evidence to biology, mathematics and computing while checking pathways separately.
Section 3 of 36
3. Distinguish tool from index
tabix is the command-line program; .tbi and .csi are index formats it can create. The source remains bgzip-compressed text.
Working checkpoint — genomic text: Name all three layers in methods. Preserve headers, malformed-line reports and records outside the query; returned rows are only a regional slice.
For scientific writing, report schema, compression, index format, object identity and boundary tests.
Section 4 of 36
4. Require bgzip compression
Regional seeking relies on BGZF block boundaries and virtual offsets. Ordinary gzip may decompress sequentially yet lacks the required blocked access structure.
Working checkpoint — preset: Validate the source with BGZF-aware tools. Separate successful text retrieval from measurement quality, variant interpretation and biological causation.
For students, ask which columns define place and how the map finds candidate lines.
Section 5 of 36
5. Require position sorting
The text must be sorted by sequence and coordinate according to the chosen parser contract. An index cannot make unsorted rows safe.
Working checkpoint — coordinate column: Scan actual order and reject backward coordinates. Treat column numbers, delimiters, sort order and coordinate base as scientific inputs: a one-column error can look plausible.
For parents, ask whether readable text and fast search were both checked for correctness.
Section 6 of 36
6. Use a preset when it truly matches
Presets for formats such as VCF, BED, SAM or GFF encode established column and coordinate rules. A similar-looking custom file may not qualify.
Working checkpoint — meta line: Compare every relevant column with the preset definition. Keep raw text, bgzip stream and index hashes, schema, preset or columns, coordinate convention, sort and tool versions together.
For a classroom model, index a fictional habitat table with no personal data.
Section 7 of 36
7. Configure the sequence column
Generic indexing records which one-based text column contains the reference sequence name. Wrong selection partitions rows under meaningless keys.
Working checkpoint — tabix: Check the chosen column against header and sample rows. Read every empty or positive result beside header definitions, reference-name identity and a full-scan comparison.
For future study, connect tabular evidence to biology, mathematics and computing while checking pathways separately.
Section 8 of 36
8. Configure the begin column
The beginning coordinate column is mandatory for generic indexing. Its text position and genomic base convention are separate facts.
Working checkpoint — TBI index: Record both column number and coordinate system. Preserve headers, malformed-line reports and records outside the query; returned rows are only a regional slice.
For scientific writing, report schema, compression, index format, object identity and boundary tests.
Section 9 of 36
9. Configure the end column
An optional end column represents intervals. Without one, format rules determine an implied end or record length.
Working checkpoint — genomic text: Test multi-base and single-position records. Separate successful text retrieval from measurement quality, variant interpretation and biological causation.
For students, ask which columns define place and how the map finds candidate lines.
Section 10 of 36
10. Declare zero-based input explicitly
The generic tabix interface can be told that input positions are zero based. BED-like and VCF-like files therefore need different settings.
Working checkpoint — preset: Add first-base fixtures that would fail under the wrong convention. Treat column numbers, delimiters, sort order and coordinate base as scientific inputs: a one-column error can look plausible.
For parents, ask whether readable text and fast search were both checked for correctness.
Section 11 of 36
11. Declare the meta character
Comment or header lines can be identified by a leading character and excluded from coordinate parsing. That metadata may still define the file.
Working checkpoint — coordinate column: Preserve all header lines with the source and query output. Keep raw text, bgzip stream and index hashes, schema, preset or columns, coordinate convention, sort and tool versions together.
For a classroom model, index a fictional habitat table with no personal data.
Section 12 of 36
12. Declare skipped lines
The index can record an initial number of lines to skip before ordinary parsing. Silent extra lines later in the file remain a hazard.
Working checkpoint — meta line: Validate the whole stream, not only the beginning. Read every empty or positive result beside header definitions, reference-name identity and a full-scan comparison.
For future study, connect tabular evidence to biology, mathematics and computing while checking pathways separately.
Section 13 of 36
13. Preserve reference-name order
TBI stores a concatenated list of sequence names and one index structure per name. Name order is therefore part of the binary interpretation.
Working checkpoint — tabix: Compare the list with the source’s actual grouping. Preserve headers, malformed-line reports and records outside the query; returned rows are only a regional slice.
For scientific writing, report schema, compression, index format, object identity and boundary tests.
Section 14 of 36
14. Understand TBI format codes
The TBI configuration records generic, SAM or VCF format identifiers together with coordinate flags. The code changes interval interpretation.
Working checkpoint — TBI index: Decode configuration before diagnosing query errors. Separate successful text retrieval from measurement quality, variant interpretation and biological causation.
For students, ask which columns define place and how the map finds candidate lines.
Section 15 of 36
15. Read bins and chunks
Each reference has hierarchical bins containing BGZF virtual-offset chunk ranges. Chunks identify candidate bytes, not exact record matches.
Working checkpoint — genomic text: Apply the format-specific overlap test after retrieval. Treat column numbers, delimiters, sort order and coordinate base as scientific inputs: a one-column error can look plausible.
For parents, ask whether readable text and fast search were both checked for correctness.
Section 16 of 36
16. Use the linear index
TBI also stores offsets for successive 16-kilobase intervals to avoid chunks ending before a query start.
Working checkpoint — preset: Test empty and sparse intervals. Keep raw text, bgzip stream and index hashes, schema, preset or columns, coordinate convention, sort and tool versions together.
For a classroom model, index a fictional habitat table with no personal data.
Section 17 of 36
17. Know the TBI coordinate limit
The BAI-style hierarchy underlying TBI has a fixed maximum coordinate range. Long references or coordinates require CSI.
Working checkpoint — coordinate column: Inspect every reference length before choosing .tbi. Read every empty or positive result beside header definitions, reference-name identity and a full-scan comparison.
For future study, connect tabular evidence to biology, mathematics and computing while checking pathways separately.
Section 18 of 36
18. Choose CSI deliberately
The tabix manual offers CSI creation and a configurable minimum interval size. CSI improves range flexibility but consumers must support it.
Working checkpoint — meta line: Test all readers before switching production indexes. Preserve headers, malformed-line reports and records outside the query; returned rows are only a regional slice.
For scientific writing, report schema, compression, index format, object identity and boundary tests.
Section 19 of 36
19. Query with matching reference names
chr1, 1 and an accession can name different or equivalent sequences depending on the source. tabix performs key matching, not reference reconciliation.
Working checkpoint — tabix: Use a declared reference dictionary or mapping. Separate successful text retrieval from measurement quality, variant interpretation and biological causation.
For students, ask which columns define place and how the map finds candidate lines.
Section 20 of 36
20. Use half-open query reasoning
Command interfaces and file formats can expose different coordinate conventions. Convert the user’s intended region exactly once at a documented boundary.
Working checkpoint — TBI index: Test a one-base record at coordinate one and zero-based position zero. Treat column numbers, delimiters, sort order and coordinate base as scientific inputs: a one-column error can look plausible.
For parents, ask whether readable text and fast search were both checked for correctness.
Section 21 of 36
21. Keep headers with query output
Many tabix workflows can return header lines with selected records. Headers carry field definitions, reference metadata and provenance.
Working checkpoint — genomic text: Include the header whenever rows would otherwise be ambiguous. Keep raw text, bgzip stream and index hashes, schema, preset or columns, coordinate convention, sort and tool versions together.
For a classroom model, index a fictional habitat table with no personal data.
Section 22 of 36
22. Do not equate no result with no biology
An empty query can result from reference-name mismatch, coordinate conversion, stale index, filters or genuine absence.
Working checkpoint — preset: Check controls and parser configuration before biological interpretation. Read every empty or positive result beside header definitions, reference-name identity and a full-scan comparison.
For future study, connect tabular evidence to biology, mathematics and computing while checking pathways separately.
Section 23 of 36
23. Protect index freshness
Editing, resorting or recompressing text changes content or virtual offsets. A stale .tbi may still look ordinary on disk.
Working checkpoint — coordinate column: Regenerate and hash the index after every source rewrite. Preserve headers, malformed-line reports and records outside the query; returned rows are only a regional slice.
For scientific writing, report schema, compression, index format, object identity and boundary tests.
Section 24 of 36
24. Validate malformed lines
Missing tabs, non-numeric coordinates, negative intervals and late metadata can confuse indexing or silently drop rows.
Working checkpoint — meta line: Run a full schema validation before compression. Separate successful text retrieval from measurement quality, variant interpretation and biological causation.
For students, ask which columns define place and how the map finds candidate lines.
Section 25 of 36
25. Use stable sorting rules
Lexical sorting under different locales can order sequence names or numeric coordinates unexpectedly.
Working checkpoint — tabix: Set and record a deterministic genomic sort procedure. Treat column numbers, delimiters, sort order and coordinate base as scientific inputs: a one-column error can look plausible.
For parents, ask whether readable text and fast search were both checked for correctness.
Section 26 of 36
26. Compare with a full scan
Representative regional results should match a sequential parser applying the same schema and overlap rule.
Working checkpoint — TBI index: Compare complete lines or stable row identifiers. Keep raw text, bgzip stream and index hashes, schema, preset or columns, coordinate convention, sort and tool versions together.
For a classroom model, index a fictional habitat table with no personal data.
Section 27 of 36
27. Plan remote URLs carefully
tabix can work with remote files when servers support byte ranges. Index and source must refer to the same immutable object version.
Working checkpoint — genomic text: Record URLs, hashes, sizes and cache validators. Read every empty or positive result beside header definitions, reference-name identity and a full-scan comparison.
For future study, connect tabular evidence to biology, mathematics and computing while checking pathways separately.
Section 28 of 36
28. Connect to Primary Science
A long table of fictional habitat observations can be indexed by region while remaining readable line by line.
Working checkpoint — preset: Ask pupils to define the place and start columns. Preserve headers, malformed-line reports and records outside the query; returned rows are only a regional slice.
For scientific writing, report schema, compression, index format, object identity and boundary tests.
Section 29 of 36
29. Connect to Secondary Science
Tabix shows that faster retrieval depends on a correct data model: wrong coordinates can return confidently wrong evidence.
Working checkpoint — coordinate column: Separate parser validation from scientific inference. Separate successful text retrieval from measurement quality, variant interpretation and biological causation.
For students, ask which columns define place and how the map finds candidate lines.
Section 30 of 36
30. Connect to mathematics
Sorting, half-open intervals, nested bins and unions of byte ranges connect text indexing to discrete mathematics.
Working checkpoint — meta line: Trace one toy query from interval to candidate chunks. Treat column numbers, delimiters, sort order and coordinate base as scientific inputs: a one-column error can look plausible.
For parents, ask whether readable text and fast search were both checked for correctness.
Section 31 of 36
31. Connect to computing
Text parsing, compression, binary indexes, remote ranges and cache consistency form a practical data-engineering lesson.
Working checkpoint — tabix: Test corrupted headers, wrong presets and stale indexes. Keep raw text, bgzip stream and index hashes, schema, preset or columns, coordinate convention, sort and tool versions together.
For a classroom model, index a fictional habitat table with no personal data.
Section 32 of 36
32. Connect to pathways
Indexed genomic text connects biology, bioinformatics, cloud systems and data stewardship and can illuminate STEM pathways.
Working checkpoint — TBI index: Check current official qualifications separately. Read every empty or positive result beside header definitions, reference-name identity and a full-scan comparison.
For future study, connect tabular evidence to biology, mathematics and computing while checking pathways separately.
Section 33 of 36
33. Did you know?
A tabix-indexed file stays human-readable after decompression, yet regional queries can avoid scanning most of its lines because BGZF blocks and the index work together.
Working checkpoint — genomic text: Explain both the readable layer and the binary companion. Preserve headers, malformed-line reports and records outside the query; returned rows are only a regional slice.
For scientific writing, report schema, compression, index format, object identity and boundary tests.
Section 34 of 36
34. Write a reader-ready tabix note
Name text format, coordinate convention, columns or preset, sort command, bgzip and tabix versions, index type, hashes and tests.
Working checkpoint — preset: Include one verified boundary query. Separate successful text retrieval from measurement quality, variant interpretation and biological causation.
For students, ask which columns define place and how the map finds candidate lines.
Section 35 of 36
35. Build a validation ladder
Check text schema, coordinates, metadata, sort order, BGZF, index configuration, reference names, limits and full-scan equivalence.
Working checkpoint — coordinate column: Stop at the first mismatch. Treat column numbers, delimiters, sort order and coordinate base as scientific inputs: a one-column error can look plausible.
For parents, ask whether readable text and fast search were both checked for correctness.
Section 36 of 36
36. Finish with a reproducible bundle
Deliver source and compressed hashes, schema, commands, .tbi or .csi hash, headers, boundary fixtures and comparison logs.
Working checkpoint — meta line: Make every returned line traceable to exact source text. Keep raw text, bgzip stream and index hashes, schema, preset or columns, coordinate convention, sort and tool versions together.
For a classroom model, index a fictional habitat table with no personal data.
Contents · Previous section · Continue to the Science Learning Hub